Chapter 1 — Definition, Landscape and Stakes

A business that has grown by making some things and buying others owns the boundary of its firm whether or not anyone chose it, so this chapter defines vertical integration and its related terms in this book's own words, draws the value chain as a map of what a business could own, reads the research on firm boundaries at the level it was read, draws the lines between this book and its neighbors, names control, capability and capital as the three constraints, calls the rules that touch integration to the books that own them, and opens the case of Hydrosentinelle inc. on day 1, which is day 366 of Book A9.

What You'll Be Able to Do

• Define vertical integration, backward integration, forward integration, make-or-buy and the hybrid form in the working words this book uses, and say what each of them is not.
• Draw the value chain of a product business as a row of stages, and mark for each stage whether the business makes it, buys it or contracts it.
• Decide one whole activity at a time, on purpose, with the reason written down and a trigger for reading the decision again.
• Read the research on firm boundaries at the level it was read, with its setting and its sample, and never as a promise of a saving.
• Place a question about the boundary of the firm in the book that owns it, and keep for this book what concerns the ownership of a whole activity.
• Name the three constraints on ownership — control, capability and capital — and the signs that a stage is owned for the feeling of control rather than for a reason.

Why It Matters

Every business has a boundary whether or not anyone drew it. A business that started by making one thing and buying the rest, or that kept a step inside because a person knew how to do it, owns a set of stages that history assembled, and the boundary is real even when it was never written down. It decides what the business depends on, what it can change on its own decision, and what it pays for whether or not it sells anything in a given quarter. A business that cannot say which stages it owns cannot say what it depends on, and a dependence that is not named is a dependence that will be discovered on the day it hurts.

The decision is not whether to own as much as possible. It is which stages to own and why. A business that owns too little can be held up by a supplier, and a business that owns too much carries fixed costs it cannot move and skills it does not have. The research read for this book does not promise that either choice is better in general: one study found no significant direct effect of outsourcing peripheral and near-core tasks on firm performance, and another found that neither outsourcing nor internalization per se results in superior performance. What the research supports is a decision made on the facts of one business, one activity and one year.

This chapter sets the words, the map, the decision, the evidence, the lines, the constraints, the landscape and the rules, so that a reader finishes with a vocabulary, a map and a way of placing a question, and can begin the work that the rest of the book describes.

In This Chapter

Section 1.1 defines vertical integration, backward and forward integration, make-or-buy and the hybrid form. Section 1.2 draws the value chain as a map of what a business could own. Section 1.3 treats the choice to make, buy or contract as a decision rather than a default. Section 1.4 reads the research on firm boundaries at the level of its abstracts and its openings. Section 1.5 draws the lines between this book and its neighbors. Section 1.6 names the three constraints on ownership. Section 1.7 dates the landscape and states what is at stake. Section 1.8 calls the rules to the books that own them. Section 1.9 opens the case of Hydrosentinelle inc. on day 1 and walks the plan of this book.

1.1 What Vertical Integration Is, and What It Is Not

A business is vertically integrated in the measure that it owns more than one stage of the chain that takes a product or a service from its inputs to its customer. That is this book's working definition, and it is deliberately a matter of measure rather than a yes or a no. A business that designs its product, buys its boards, assembles its kits in its own workshop, installs them through its own staff and runs the service that watches them owns several stages and buys others. Each stage is either owned or not, and the set of owned stages is the boundary of the firm.

The definition has a negative side that is worth stating as plainly as the positive one. Buying a good or a service from a supplier is not integration. A long-term agreement that leaves ownership where it was is not integration either, however close the two businesses become. A business that signs a three-year supply agreement, holds dedicated tooling at a supplier's plant and shares its production plan with that supplier has moved closer to the supplier, and it has not integrated the stage. Ownership carries the fixed cost, the management, the hiring and the risk, and an agreement carries a promise and a price. Chapter 5 takes the forms that stand between the market and ownership.

The lecture read for this book puts the decision in a form that a small business can use. Taking the technological core as given, the lecture says, attention focuses on a series of separable make-or-buy decisions, backward, forward and lateral, to decide which transactions stay within the ownership boundary of the firm (Williamson, 2009). The lecture is the Nobel Prize lecture of December 8, 2009, read on October 15, 2026 in its published PDF, and it is a theoretical account with no number and no decision rule for a small firm. What it gives is a way of seeing the question: not one large decision about how integrated to be, but a series of separable decisions, each about one transaction. The lecture also says that the hybrid mode was introduced later, in the paper of 1991 that section 1.3 reads.

Integration is a matter of degree, and it runs in two directions from the core of the business. A stage that supplies the business sits upstream, and owning it is backward integration. A stage between the business and its customer sits downstream, and owning it is forward integration. The words are directional and not moral. A business that owns its tooling has integrated backward into the assets that make its product, and a business that installs its own product has integrated forward into the last step before the customer; both decisions commit capital and management, and both can be reversed only at a cost.

Every degree of integration is a decision that someone made or that history made by default. A founder who bought boards because a supplier existed made a decision, even if it was never written down. The point of this book is not to move a business toward more ownership or toward less; it is to make the boundary visible, to decide each stage on purpose, to write the reason down, and to read the decision again when a trigger says to.

The table below sets out the five words this book uses for the boundary of the firm, what each means here, and how the company of the case reads each one. The readings in the third column are posed by this book and are told in full in section 1.9. Read the table as a vocabulary.

Table 1.1 — Five words of this book and the company's reading of each

DEFINITION The words this book uses for vertical integration, backward and forward integration, make-or-buy and a hybrid form

These are this book's working words, and no source read defines them in this sense. Vertical integration is the ownership by one business of more than one stage of the chain from inputs to customer. Backward integration is the ownership of a stage that supplies the business, such as components, boards or tooling. Forward integration is the ownership of a stage between the business and its customer, such as distribution, installation, service or the customer relationship. Make-or-buy is the decision, for one whole activity, to make it inside the business, to buy it from the market or to contract it in a form between. A hybrid form is an arrangement between market and ownership, such as a long-term supply agreement, dedicated tooling held at a supplier or a joint investment. The value chain and the activity system are Book A3's (Strategy and Competitive Positioning, chapter 2), and the buying of work and services is Book I8's (Freelancers, Agencies and Outsourcing, chapter 2); neither is redefined here.

1.2 The Value Chain as a Map of What You Could Own

The value chain is the map on which the boundary of the firm is drawn. The page read for this book describes it as a tool that breaks a company into its strategically relevant activities to find where advantage comes from, and it says that the activities that matter command higher prices or lower costs (Institute for Strategy and Competitiveness, n.d.). The page was read on October 15, 2026 and shows no date. It also says that a company's chain sits within a larger value system that includes upstream suppliers and downstream channels. The page read gives no list of primary and support activities, and this chapter prints none. Strategy, the value chain and the activity system are Book A3's (Strategy and Competitive Positioning, chapter 2), cited here and never redefined.

This book cuts the chain of a product business into six stages, and the cut is this book's own rather than the page's list. The stages run from the left to the right: inputs and components; tooling and equipment; making and assembly; distribution and delivery; installation and service; and the customer relationship and data. The cut is a practical one. Each stage is a whole activity that a business can own or not own, and each one has a supplier market on one side and a customer on the other. A service business can use the same six stages with different names.

For each stage there are three ways to organize it, and the three ways are the subject of this book. The first is to make it inside the business, which means owning the assets, the people and the management that the stage needs. The second is to buy it from the market, which means paying a price and leaving the ownership where it is. The third is to contract it in a form between the two, such as a long-term supply agreement, dedicated tooling held at a supplier or a joint investment. The three ways are not a ladder from bad to good; they are three answers to the same question, which is which one fits this stage, this business and this year.

Figure 1.1 lays out the value chain of a product business as a row of stages, each of which can be made, bought or contracted, with the neighboring books under it.

Figure 1.1 — The value chain as a map of what you could own. Each stage can be made, bought or contracted; owning upstream is backward integration, owning downstream is forward integration.

The figure is read from left to right. Inputs and components are the boards, the sensor elements and the parts that go into the product. Tooling and equipment are the molds, the jigs and the test benches that make the product possible. Making and assembly turns parts into a finished product. Distribution and delivery moves the finished product to the customer. Installation and service puts the product into use and keeps it working. The customer relationship and data is the record of who the customer is, what the customer bought and what the product reports. Under each stage the figure shows the same three ways to organize it, and around the row it names the neighboring books.

KEY CONCEPT Every stage of the chain can be made, bought or contracted, and the question is which stages to own and why

The chain is a row of decisions and not a row of obligations. A business owns a stage on purpose when the activity matters enough to justify the volume, the skill and the capital that ownership commits, and it buys or contracts the others. The figure shows six stages and three ways to organize each one, and the three ways are answers to the same question rather than steps on a ladder. Owning more is not better in general, and owning less is not better in general: the research read in section 1.4 finds no direct effect of outsourcing on firm performance and no superior performance from either outsourcing or internalization per se. What a business can do is make the boundary visible, decide one stage at a time, write the reason down and read the decision again on a trigger.

1.3 Make, Buy or Something in Between: A Decision, Not a Default

The paper read for this book distinguishes three generic forms of economic organization, and the distinction is the backbone of this chapter. The paper argues, as a conceptual analysis, that institutional economics, contract law and organization theory together separate market, hybrid and hierarchy by their coordinating and control mechanisms and by their abilities to adapt to disturbances (Williamson, 1991). Each form is supported by a distinctive type of contract law, the paper argues, and the cost-effective form varies systematically with the attributes of transactions. Changes in property rights, contract law, reputation effects and uncertainty shift the comparative costs of governance. The paper is read at abstract only, and it offers no test for a small firm.

A market transaction coordinates by price: a business asks for a quote, compares it with another, and buys. A hierarchy coordinates by ownership and management: the business directs the work, hires the people and carries the fixed cost. A hybrid coordinates by an agreement that lasts, with dedicated assets or shared plans, so that the business and the supplier commit to each other for a period without either owning the other. The paper argues that the cost-effective form varies with the transaction rather than with the fashion.

The practice guidance of a federal Crown corporation says the same thing in the language of a small business. The page read for this book, updated October 4, 2022, says that a business can subcontract the production of a good, one manufacturing step, a service or storage (Business Development Bank of Canada, 2022). It says that a supplier may produce more cheaply, mostly as a matter of volume, so the decision should be revisited as the company grows. It names the risks as dependence, the loss of in-house expertise and the sharing of inside information. It recommends being transparent with the subcontractor by giving it a solid overview of the work ahead and sharing the right information. The page is the practice guidance of a federal Crown corporation and never an obligation, and it gives no figure.

A business that never decides ends with the boundary that its history gave it. That is this book's own reasoning and not a finding of any source. A stage bought because a supplier existed ten years ago may still be bought, at a price that has moved, for a reason that nobody remembers, and a stage kept inside because the founder liked doing it may still be inside, with a fixed cost that does not fall when volume does. The remedy is to decide each whole activity on purpose, to write the reason down, and to read the decision again on triggers. Book A1 (From Idea to Business Model, chapter 6) and Book I2 (Organizational Design, Roles, Delegation and Accountability, chapter 2) each told the reader that the choice between doing work inside the business and buying or sharing it outside belongs to this book, and Chapters 3 and 5 keep those promises.

The decision has questions, and the chapters that follow answer them one at a time: which stage and which direction in Chapter 2, what made and bought cost on the same basis in Chapter 3, how work moves out of the business and back in Chapter 4, the forms between market and ownership in Chapter 5, ownership of data in Chapter 6, the numbers over the life of the asset in Chapter 7, and the cadence, the roles, the tools and the plan of ninety days in Chapters 8 to 11.

The table below sets the three forms beside how each coordinates and what supports it, and beside where the activities of the company of the case sit. The third column is posed by this book and is scored in Chapter 3; the activities it names are printed by Books A5, H5 and G8 and read here by this book. Read the table as a map of the three forms: the same business can use all three at once.

Table 1.2 — Market, hybrid and hierarchy, and where the company's activities sit

WHO DOES THIS The owner decides what to own, and the leads assess the work and the numbers as the business grows

Who carries a make-or-buy decision depends on the size of the business. A solo owner decides alone and writes the reason down, because a decision that lives only in memory will be taken again on the same facts. In a small team the owner decides and an operations lead assesses the work and the numbers, so that the person who runs the stage has a voice in whether it should be owned. In an established business an operations lead, a purchasing manager and a finance lead prepare the decision and the owner or the executive team approves it, and Chapter 9 reads the roles in full. In the case of this book, posed by this book, Laurence Bergeron and Karim Haddad sign the verdicts together on day 28, with no hire and no dollar. The pattern is this book's own suggestion and not a rule of any source.

1.4 What the Research on Firm Boundaries Supports

The research on the boundary of the firm is a body of theory and a set of studies, and this book reads it at the level it was read. A theory holds what follows from its assumptions, a study reports what it found in one industry, one period and one sample, and a review read at its abstract is not printed beyond that abstract. A finding read as a promise becomes a decision on a fact that does not exist.

The oldest reading is the one that started the question. The text read for this book, read on October 15, 2026 where it is legible, which is its first half, says that the main reason why it is profitable to establish a firm would seem to be that there is a cost of using the price mechanism (Coase, 1937). The most obvious cost is the cost of discovering what the relevant prices are, and then the costs of negotiating and concluding a separate contract for each exchange. The later argument of the paper is not printed here, because the deposited text is illegible there. A business that makes a stage inside avoids that cost, and it pays for the avoidance in management and fixed cost.

Two theories of ownership follow from that starting point. The first says that contractual rights are specific or residual, and that when it is costly to list all specific rights over assets in a contract, it may be optimal to let one party purchase all residual rights; ownership is the purchase of these residual rights (Grossman & Hart, 1986). The second theory identifies a firm with the assets its owners control and says that the crucial difference between owning a firm and contracting for a service from another party who owns it is that under integration the owner can selectively stop working with the people of the firm, while without integration the owner can only stop dealing with the whole firm and its assets (Hart & Moore, 1990). Both are theories read at abstract only, and neither reports a measured effect.

A third reading ties capabilities to transaction costs. The abstract reports that transaction costs and capabilities are intertwined in determining vertical scope, that capability differences are a necessary condition for vertical specialization, and that lower transaction costs lead to specialization only when capabilities along the chain are heterogeneous, illustrated by mortgage banking in the United States and Swiss watch manufacturing (Jacobides & Winter, 2005).

The review read for this book reports the state of the empirical work. The abstract reports a review of the empirical studies, accumulated over twenty-five years, of two questions: which transactions are best brought within the firm, and what vertical integration does to prices, quantities, investment and profits (Lafontaine & Slade, 2007). The review is read at abstract only, and nothing beyond its abstract is printed.

Three studies report what they found in their own settings. The first reports that standard performance models can wrongly suggest a positive link between outsourcing and technological performance, and that models which account for firm and transaction features indicate that neither outsourcing nor internalization per se results in superior performance, which is contingent on the alignment between the governance decision and the degree of contractual hazards (Leiblein et al., 2002). The second reports that in the luxury automobile segment, initial performance is associated with a decline as the level of vertical integration rises, while performance improvement over the product lifecycle is associated with an increase in it (Novak & Stern, 2008). The third reports that no significant direct effect of outsourcing peripheral and near-core tasks on firm performance was found, while firm strategy and environmental dynamism moderated the relationship (Gilley & Rasheed, 2000). None of the three promises a saving or a loss to a reader.

The article by two consultants is read at its opening only, the opening that the publisher shows, published April 15, 1993, with the rest behind a subscription, and this chapter says so once and prints nothing from the parts not visible. The opening states, as the position of two consultants and never as evidence or a rule, that vertical integration can be important but is hard to implement and costly to reverse, and that management's record on it is described as poor (Stuckey & White, 1993). It says that integration coordinates adjacent stages when trading between them is not beneficial, and it names four reasons for integrating: a risky or unreliable market, adjacent stages holding more market power, integration that would create or exploit market power, and a young market to develop or a declining one that independents are leaving.

The table below sets eight readings on firm boundaries beside the level at which each was read and its setting, what it reports, and what it does not support. The rows are in the order the section reads them, and the last column is the part that matters most: a reading that is a theory supports no measured effect, and a study of one industry supports no finding outside it. Read the table as a set of limits.

Table 1.3 — Eight readings on firm boundaries, at the level read

THE RULE A finding is printed with its setting, and no gain is promised (Gilley & Rasheed, 2000)

The abstract of the study read for this book reports that no significant direct effect of outsourcing peripheral and near-core tasks on firm performance was found, while firm strategy and environmental dynamism moderated the relationship (Gilley & Rasheed, 2000). The finding is one study's, in its own setting and period, and it promises no saving and no loss to a reader's business. The rule this book applies is that a finding is printed with its setting, that a theory is written as a theory, and that no gain is promised from outsourcing, insourcing, integration, a contract manufacturer or the purchase of a supplier. A business therefore decides on its own grid and its own numbers. No source read for this book measures a typical saving, a typical premium, a typical hidden cost or a typical failure rate of outsourcing, and none is printed here.

1.5 Where the Lines Fall: Outsourcing, Purchasing, Partnerships and Platforms

This book holds one thing: the boundary of the firm. It holds the decision to own, to have made or to contract a whole activity or a stage of the chain, its grid, its financial analysis, its hybrid forms, its transitions in both directions and its review on triggers. It does not hold the buying of a piece of work, the buying of goods, a negotiated alliance, a platform, a distributor, the financing of an asset or the ownership of a business. Those belong to the books that own them, and this section draws the lines so that a reader knows where to look.

The buying of work and services by a small business is Book I8's (Freelancers, Agencies and Outsourcing), including the core and the non-core, the full cost of bought work and the grid of seven families of work of the case (chapter 2 of Book I8). This book cites Book I8 and never redoes it: Book I8 takes one piece of work and asks how to buy it well, and this book takes a whole activity and asks whether to own it at all.

Book J5 (Supply Chain, Procurement and Inventory) lies outside this book, and Book E9 (Tenders, RFPs and Public Procurement) treats public purchasing and private requests for proposals (its section 1.2) and lies outside this book. Negotiated alliances and joint ventures are Book L4's, platforms and the dependence on dominant offers are Book A9's, and distributors are Book F3's; Table 1.4 sets what each holds beside what this book holds.

The financing of the tooling and its loan is Book A5's, strategy and the value chain are Book A3's, and Book M6 (Buying a Business) is only called in Chapter 2.

The table below sets each neighbor of this book beside what the neighbor holds and what this book holds. The rows are in the order the section reads them, and the third column is the part that matters most: it says what a reader will find here rather than there. Read the table as a set of lines.

Table 1.4 — Where this book stops and its neighbors begin

METHOD Placing a question about the boundary of the firm in the right book in five steps

First, write the question in one sentence, because a question that cannot be written in one sentence is two questions. Second, ask whether it concerns the buying of a piece of work, the buying of goods, a negotiated alliance, a platform, a distributor, the financing of an asset or the ownership of a whole activity. Third, send it to the book that owns that part. Fourth, keep for this book what concerns the boundary of the firm. Fifth, read the neighbor first when a rule is written there, because this book cites the neighbor and never rewrites it. The sequence is this book's own and no source tests it. The example is a customer asking whether the company could sell its kits through a distributor: the distributor is Book F3's, and whether to own the downstream stage is this book's. The method is a way of reading rather than a rule of law, and it promises no result.

1.6 Control, Capability and Capital: The Three Constraints

A business can own a stage only within three limits, and the three limits are this book's own way of organizing the question. The first is control, which is what the owner can direct and measure and what the owner can only ask for: a stage that a supplier owns is a stage the business can influence through a contract and a relationship, and no further. The second is capability, which is what the business does well and what it could learn in time; a stage that needs a skill nobody in the business has is a stage that ownership will expose. The third is capital, which is the money, the floor space and the hours that ownership commits, now and when the asset must be replaced. A stage that needs a machine, a mold or a bench will need the money again when the asset wears out.

The three constraints are not a checklist to be scored one at a time but a set of conditions that must agree. Integration is a decision when the business can direct the stage, staff it and pay for it, and it is a trap when only one of the three holds. Chapter 3 scores each activity on one grid with these constraints among its criteria, and Chapter 7 compares made and bought on the same basis.

The tooling of the case shows what capital means in a small business. Book A5 printed that tooling and test equipment cost $95,000 and are capital spending, and that the enclosure's molds would need to be replaced or redesigned within about five years. Book A5 also printed that few buyers exist for a mold made for someone else's product, so an asset specific to one product cannot be sold when the product changes. Owning the tooling therefore commits the business to the money, the floor space and the replacement, and it buys control of a specific asset.

Control has a cost that is easy to miss. A stage that the business owns must be staffed in a busy week, carries a fixed cost that does not fall when volume falls, and may hold its skill in one person who can leave. None of these is a reason never to own a stage; each is a reason to own a stage on purpose rather than for the feeling of control.

The remedy for a stage owned for the wrong reason is this book's own reasoning: score the activity on one grid (Chapter 3), compare made and bought on the same basis (Chapter 7), write down what would bring the work back (Chapter 4), and read the decision again on a trigger (Chapter 8).

WATCH OUT Owning a stage for the feeling of control can leave a business with a fixed cost and a skill it does not have

The trap is a stage kept inside because it feels safer, and the signs are three. The first is a stage that nobody can staff in a busy week, so the stage that was meant to protect the chain slows it. The second is a fixed cost that does not fall when volume does, so a slow quarter carries a cost that a bought stage would not. The third is a skill that one person holds, so the stage is owned by the business and controlled by a person who may leave. The remedy is to score the activity on one grid, to compare made and bought on the same basis, and to write down what would bring the work back, which are the subjects of Chapters 3, 7 and 4. The reasoning is this book's own and not a finding of any source, and no figure is printed here.

1.7 The Landscape in 2026 and What Is at Stake

The landscape is told in words, and the words come from one release of a statistics agency. The release of May 27, 2026 on the Canadian Survey on Business Conditions, second quarter 2026, read on October 15, 2026, reports that 64.3 percent of businesses expected cost-related obstacles, up from 58.9 percent in the first quarter (Statistics Canada, 2026). The release reports that 48.8 percent expected inflation and 28.4 percent the cost of inputs to be obstacles. In manufacturing, 58.2 percent expected inflation and 48.1 percent the cost of inputs to be obstacles, and 54.0 percent expected a negative impact of United States tariffs, against 34.0 percent of all businesses. The figures are shares of businesses, and those of manufacturing are shares of the businesses of that sector. The release gives no figure on supply chain challenges, on outsourcing or on small businesses alone. The page was modified on August 27, 2026 after the release, so the release date is the date of the source.

What the release means for the boundary of the firm is a matter of reasoning rather than of measurement. A business that buys a component watches its price move, and a business that makes it watches its own costs move. The release does not say which way a business should go, and this book does not either; it dates the landscape. No source read for this book measures the typical saving of outsourcing, the typical premium of a contract manufacturer, the typical hidden cost or the typical failure rate, which is why the figures of the case are posed and said to be posed.

The stakes are three sentences, and they are this book's own reasoning. A business that does not know which stages it owns cannot say what it depends on, because dependence is a property of the boundary and not of a supplier list. A business that outsources its core hands away what makes it different, because the stage that carries the difference is the stage that a competitor can buy from the same supplier. A business that owns a stage without the volume, the skill or the capital carries a cost it cannot move, because a fixed cost is a cost that stays when the volume goes.

WHAT CHANGED Cost-related obstacles reached 64.3 percent of businesses in the second quarter of 2026, up from 58.9 percent (Statistics Canada, 2026)

The dated change comes from the release of May 27, 2026 on the Canadian Survey on Business Conditions, read on October 15, 2026. For the second quarter of 2026, the release reports that 64.3 percent of businesses expected cost-related obstacles, up from 58.9 percent in the first quarter (Statistics Canada, 2026). The shares are the agency's, for the businesses surveyed, and the collection ran from April 1 to May 6, 2026, with expectations for the next three months. The release gives no figure on outsourcing, on supply chain challenges or on small businesses alone, and the shares will change with the next quarter. The date is the source's and never a date of the case, and no other date is printed here.

TOOLS OF THE TRADE What a business needs to see its boundary: a map, a list of decisions, stored agreements, a calendar and one register

Five functions carry the boundary of the firm, and a business should look for them rather than for a product: a map of the chain that a person can correct, so that the stages and their handoffs are visible; a list of the activities with the verdict and the reason for each, so that a decision can be read again on its facts; a stored copy of each agreement with the day it was signed, so that the term of the commitment is known; a calendar of the dates when a review is due, so that the decision is read again on a trigger rather than on a memory; and one register of decisions where each activity is written with its triggers, so that the trigger is a fact rather than an intention. Chapter 10 gives the functions in full. No brand name belongs on this list and no price; the requirement is the function.

1.8 The Rules That Apply to Integration and Outsourcing

None of the texts read for this book obliges a business to integrate or forbids it to. The rules that touch integration and outsourcing are rules about competition, about contracts and about employment, and each of them is held by another book of this encyclopedia. This section tells each rule in one sentence with its jurisdiction and its date, calls it to the book that holds it, and judges no real agreement, clause, exclusivity, merger or transfer of work. Only the Tribunal or a court decides whether a given arrangement complies with a rule, and this book does not say that no other rule exists.

In Canada, competition law is held by Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements). The Competition Act, R.S.C. 1985, c. C-34, ss. 45, 77, 90.1, 91, 92, 109 and 110, read on the Justice Laws Website on October 16, 2026 at its consolidation of September 30, 2026, last amended March 26, 2026, makes it an offense under s. 45(1) for a person, with a competitor in respect of a product, to agree to fix the price, to allocate sales, territories, customers or markets, or to fix, control, prevent, lessen or eliminate production or supply, with the ancillary-restraint defense of s. 45(4) and with "competitor" including a person likely to compete absent the agreement under s. 45(8). Section 77(1) defines exclusive dealing, market restriction and tied selling; s. 77(2) lets the Tribunal act on exclusive dealing or tied selling, and s. 77(3) on market restriction, only where the practice of a major supplier or a widespread practice is likely to lessen competition substantially. Section 90.1(1) allows a civil order where an agreement between persons of whom two or more are competitors prevents or lessens competition substantially, and s. 90.1(1.01) even between non-competitors where a significant purpose is to prevent or lessen competition. Section 91 defines a merger as the acquisition of control over or a significant interest in the whole or part of a business of a competitor, supplier, customer or other person, so that the purchase of a supplier or a customer is a merger, and s. 92(1) lets the Tribunal act on a merger that prevents or lessens competition substantially. Sections 109(1) and 110 make pre-merger notification apply only above the party-size threshold of $400 million and the transaction-size threshold.

The release of the competition regulator of March 2, 2026, read on October 15, 2026, reports that as of 2026 the transaction-size threshold for pre-merger notification stays at $93 million, measured on the assets in Canada of the business acquired or the revenues they generate from sales in, from or into Canada, and that notice is also generally required when the parties and their affiliates exceed $400 million in combined assets or revenues (Competition Bureau, 2026). The thresholds decide only who must notify, and they are never an exemption: s. 92 lets the Tribunal act on any merger, including the purchase of a supplier or a customer. The guidelines of the competition regulator of 2021, dated May 6, 2021 and modified June 27, 2024, read on October 15, 2026, carry a notice that the Bureau is reviewing them after recent amendments to the Act, so they are the Bureau's approach, under review, and never the law (Competition Bureau, 2021). The guidelines say that purely vertical agreements between a supplier and its customer will be assessed under the civil provisions of Part VIII rather than under s. 45, while a vertical arrangement that contains an agreement between competitors to fix prices, allocate markets or limit supply will generally be assessed under s. 45. The page predates subsections 90.1(1.01) and (1.3) of the Act read today.

In Québec, contracts are held by Book H1 (Contracts and Commercial Agreements) and by a lawyer. The Civil Code of Québec, CQLR c CCQ-1991, arts. 1708, 1726, 1729, 2098, 2099, 2100, 2103, 2125, 2126 and 2129, read in LégisQuébec on October 15, 2026, up to date to August 12, 2026, says in art. 1708 that a sale transfers ownership of property for a price in money; in art. 1726 that the seller warrants the property free of latent defects; and in art. 1729 that a defect is presumed to have existed at the sale by a professional seller if the property malfunctions or deteriorates prematurely. Article 2098 defines a contract of enterprise or for services as one by which a contractor or provider undertakes to carry out physical or intellectual work or to supply a service for a price; and art. 2103 says the contractor supplies the property necessary for the work unless only the work is stipulated, and that the contract is a sale, not a contract of enterprise, where the work is merely accessory to the value of the property supplied. Article 2125 lets the client unilaterally resiliate the contract even while the work is in progress; art. 2126 lets the contractor resiliate only for a serious reason and never at an inopportune moment; and art. 2129 says that on resiliation the client pays, in proportion to the agreed price, the actual costs, the value of the work performed and of the property supplied that can be used.

In Québec, employment law is held by Book H7 (Employment Law Essentials (Québec and Canada)). The Labour Code, CQLR c C-27, s. 45, read in LégisQuébec on October 15, 2026, up to date to August 12, 2026, says that the alienation or operation by another, in whole or in part, of an undertaking does not invalidate a certification or a collective agreement, and that the new employer is bound by them; the second paragraph does not apply to the transfer of part of the operation where the transfer does not entail, besides functions or the right to operate, most of the elements that characterize the part of the undertaking involved. The case has no certified union and makes no acquisition, so no question of certification or of notification is raised for it.

The table below sets each provision of the rules read for this book beside its jurisdiction and date, what it says in this book's words, and the book that holds it. The rows are in the order the section reads them, and the last column is the part that matters most: it says where a reader will find the rule rather than here. Read the table as a set of referrals.

Table 1.5 — The rules called to the books that hold them, with their provision and date

QUÉBEC & CANADA NOTE Competition, contracts and employment are held by other books, and this book explains the law as published and is not legal advice

This book explains the law as published and is not legal advice. The Canadian point is competition: the Competition Act, current to September 30, 2026, with Book H4 called, the purchase of a supplier or a customer being a merger under s. 91, and the thresholds of $93 million and $400 million deciding only who must notify, never an exemption. The Québec points are contracts and employment: the Civil Code of Québec, up to date to August 12, 2026, with Book H1 and a lawyer called, and the Labour Code, s. 45, with Book H7 called. Whether a given agreement, clause, exclusivity, acquisition or transfer of work complies with a rule is a question for a lawyer or the Tribunal, never decided by this book.

1.9 The Case and the Plan of This Book

The case of this book opens on day 1, and day 1 is day 366 of Book A9 (Platform Strategy: Integrating with Dominant Offers and Partner Ecosystems), the day after Book A9's one year later. It is day 731 of Book G8's count, day 1,096 of Book G7's count and day 1,461 of Book I8's count. The case is counted in days and never in calendar years or months, and day 365 of this book is entirely posed. Hydrosentinelle inc. is a fictional business created for this encyclopedia; its founders, its employees, its customers, its partners, its suppliers, its figures and the events described are invented, and any resemblance to a real business is coincidental.

The facts of the case are printed by published books, and this book cites them and never recounts them. Book A5 printed that Hydrosentinelle inc. is a fictional company in Sherbrooke that designs and sells connected water-leak detectors for commercial and rental buildings, with two founders, Laurence Bergeron, the president (product and engineering), and Karim Haddad (sales and operations), that tooling and test equipment cost $95,000, and that the enclosure's molds would need to be replaced or redesigned within about five years. Book A7 printed that the kit sells for $4,800 and costs $2,100. Book F9 printed the illustrative rates of $45 an hour for the staff installer and $300 an hour for the outside lawyer, and Book G7 printed $60 an hour for each founder and $120 an hour for the accountant. Book H5 printed that the firmware of the sensors and of the gateway, and the mobile application, are assets of the company. Book I8 printed that the testing of the gateway firmware and of modules of the mobile application is rented under control, and that the strategic make-or-buy decision of a whole activity is this book's. Book G8 printed that the workshop, the stock of detectors and components and the test bench are among the company's exposures, and that a supplier's component shortage that stops kit assembly is rated at $40,000 and 30 days, reduced by a second source and a safety stock. The company does not use La Machine, as Books G8, I8 and L5 print.

The question of this book is posed by this book: which of six activities to make, have made or contract, written one activity at a time and reviewed by triggers. The company buys boards from the electronics supplier of Book A5, holds the tooling Book A5 financed, assembles and tests its kits in its own workshop, installs them through the staff installer and the certified installers of Book F9, and owns the monitoring service and the alert logic. The firmware and the alert logic stay in-house, and the grid of six activities and its verdicts are Chapter 3's subject.

The plan of this book covers ninety days in three slices, with gates on days 30, 60 and 90, and a gate that is not met is read again 7 days later, with nothing new started until it passes. The plan is told in full in Chapter 11.

The table below sets the case on day 1 beside what published books printed and what this book poses. The rows are in the order the section reads them, and the third column is the part that matters most: it says where each fact comes from, so that a reader can tell a printed fact from a posed one. Read the table as a list of sources.

Table 1.6 — The case on day 1: what published books printed and what this book poses

In La Machine, as of the application copy of September 7, 2026, the page that touches this chapter is Control center → Tools (inventory) (/outils), an inventory of the installation's software tools with their role, state, cost and a measured version and date. It holds no purchase order, no bill of materials, no supplier record or scorecard, no inventory, no production order and no cost or make-or-buy model; check it in your installation. The case does not use La Machine, this book is valid with any tool, and La Machine is described and never sold.

CASE IN POINT Hydrosentinelle inc. opens its book on day 1 with a chain built by habit and six activities to decide

Hydrosentinelle inc. is a fictional business created for this encyclopedia; its founders, its employees, its customers, its partners, its suppliers, its figures and the events described are invented, and any resemblance to a real business is coincidental. The company opens this book on day 1, which is day 366 of Book A9 (Platform Strategy: Integrating with Dominant Offers and Partner Ecosystems). Laurence Bergeron, the president (product and engineering), and Karim Haddad (sales and operations) hold the chain that published books printed: boards from the electronics supplier of Book A5, tooling financed by Book A5, kits assembled in the workshop, installation by the staff installer and by the certified installers of Book F9, and the firmware and the alert logic as assets of the company. The plan of ninety days is to come, with gates on days 30, 60 and 90, and no result is credited to the plan. The firmware and the alert logic stay in-house, and the six activities are scored in Chapter 3.

DO THE MATH From a plan of 121 hours to $89,540 committed: the case in numbers

Illustrative figures; the case is fictional. Step 1 — The plan takes 62 hours of Laurence Bergeron, 36 of Karim Haddad, 16 of the staff installer, 2 of the accountant and 5 of the outside lawyer, which are 121 hours, posed by this book. Step 2 — At $60 an hour for each founder, $45 for the staff installer, $120 for the accountant and $300 for the outside lawyer, those hours are $8,340. Step 3 — The duplicate test bench is $9,200, posed by this book. Step 4 — The second-generation molds are $72,000, posed by this book. Step 5 — $8,340 plus $9,200 plus $72,000 are $89,540 committed in ninety days, and every hour, price and investment is posed by this book and illustrative.

IN LA MACHINE La Machine lists tools with a measured version and holds no purchase order, supplier record or make-or-buy model

In the application copy of September 7, 2026, La Machine holds no supplier record, no supplier evaluation and no supplier scorecard; no purchase order and no receiving; no bill of materials, no production order and no production planning; no stock and no inventory of parts; no register of tooling, of molds or of assets held at a supplier; no traceability of lots; no total cost, break-even, present value or make-or-buy model; and no register of sourcing decisions. Control center → Tools (inventory) (/outils) lists the installation's tools with a name, a role, a category, a state, a cost, a count of tasks and an identity measured by a script, a version, a location and a date. It can serve as a model for an inventory of what a business depends on, for software only, and it measures no dependence on a supplier of goods, no total cost and no volume. Check it in your installation. The case does not use La Machine, and La Machine is described and never sold.

Costly Mistakes

• Treating the boundary of the firm as a fact of history rather than a decision. A business that never decides ends with the stages its past gave it.
• Owning a stage for the feeling of control. A stage kept inside because it feels safer can leave the business with a fixed cost that does not fall when volume does and a skill that one person holds.
• Reading a theory as evidence. A theory holds what follows from its assumptions, and a study of one industry reports what it found there; neither promises a saving to a reader's business.
• Printing a finding without its setting. A finding read without its setting will be applied where it does not hold.
• Sending a question about the boundary to the wrong book. The buying of a piece of work is Book I8's, a negotiated alliance is Book L4's, a distributor is Book F3's, and the ownership of a whole activity is this book's.
• Reading a pre-merger notification threshold as an exemption. The thresholds of $93 million and $400 million decide only who must notify, and s. 92 lets the Tribunal act on any merger.
• Promising a saving that cannot be proved. No source read for this book measures a typical saving, a premium, a hidden cost or a failure rate of outsourcing, so a promise made on such a figure cannot be kept.

Key Takeaways

• The boundary of the firm is the set of stages a business owns, and it is real whether or not anyone chose it.
• Vertical integration is a matter of degree, and it runs in two directions from the core: owning a stage that supplies the business is backward integration, and owning a stage between the business and its customer is forward integration.
• The value chain is a row of stages, each of which can be made, bought or contracted in a form between; the question is which stages to own and why.
• The research on firm boundaries is read at the level it was read: a theory holds what follows from its assumptions, and a study reports what it found in one industry and one period.
• The lines between this book and its neighbors are drawn by subject: the buying of work is Book I8's, alliances are Book L4's, platforms are Book A9's, distributors are Book F3's, and the ownership of a whole activity is this book's.
• The rules that touch integration and outsourcing are held by other books: competition law by Book H4, contracts by Book H1 and a lawyer, employment law by Book H7, and the thresholds of $93 million and $400 million decide only who must notify.

Your Action Plan

1. Write the working words down. Write the five words of Table 1.1 — vertical integration, backward integration, forward integration, make-or-buy and a hybrid form — in one sentence each, in the words the business will use. Deliverable: a one-page vocabulary of the boundary of the firm. Verification: each word is defined in one sentence, and each sentence says what the word is not.
2. Draw the chain as a row of stages. Draw the six stages of section 1.2 for the business, from inputs and components to the customer relationship and data, and mark for each stage whether the business makes it, buys it or contracts it. Deliverable: a one-page map of the chain with a mark on each stage. Verification: every stage carries one of the three marks, and the map names the neighboring book that holds what the business does not decide here.
3. List the activities and write the reason for each. List the whole activities of the business, and beside each one write the reason it is made, bought or contracted today. Deliverable: a list of activities with a reason for each. Verification: every activity has a reason, and no reason is "it has always been this way".
4. Name the three constraints for each activity. For each activity on the list, write what the business can direct, what it can staff and what it can pay for. Deliverable: a table of activities with the three constraints beside each. Verification: each activity has a sentence for control, a sentence for capability and a sentence for capital.
5. Place the open questions in the right book. Take the questions the list has raised, and place each one in the book that owns it, using the five steps of section 1.5. Deliverable: a list of questions with the book that holds each. Verification: every question is placed, and the questions that concern the ownership of a whole activity stay in this book.

In La Machine

In the application copy of September 7, 2026, the page that touches this chapter is Control center → Tools (inventory) (/outils), an inventory of the installation's software tools with their role, state, cost and a measured version and date (La Machine, 2026). The page can serve as a model for an inventory of what a business depends on and of what it chose to buy rather than make, for software only. It is one worked example of a business platform, and this book is valid with any tool.

What a business needs for the boundary of the firm is elsewhere. La Machine holds no supplier record, no purchase order and no make-or-buy model. The map of the chain, the list of activities with their verdicts, the stored agreements, the calendar of reviews and the register of decisions live in a spreadsheet, a document or a tool the business chooses by function, and the case does not use La Machine (Books G8, I8 and L5). Check it in your installation.

Self-Check

1. What is vertical integration, and what is it not?
2. What are the six stages of the value chain this book cuts, and what are the three ways to organize each one?
3. What do the three forms of economic organization coordinate, and what does the paper read for this book argue about them?
4. What does the research read for this book support, and what does it not support?
5. Where do the lines fall between this book and its neighbors, and what do the thresholds of $93 million and $400 million decide?

Answers. 1. Vertical integration is the ownership by one business of more than one stage of the chain from inputs to customer, and it is a matter of degree; buying a good or a service from a supplier is not integration, and neither is a long-term agreement that leaves ownership where it was. 2. The six stages are inputs and components, tooling and equipment, making and assembly, distribution and delivery, installation and service, and the customer relationship and data; each can be made, bought or contracted in a form between. 3. A market coordinates by price, a hierarchy by ownership and management, and a hybrid by an agreement that lasts; the paper argues that the forms are distinguished by their coordinating and control mechanisms and their abilities to adapt to disturbances, and that the cost-effective form varies systematically with the attributes of transactions (Williamson, 1991). 4. The research supports a set of theories, a review read at its abstract and a set of studies each in its own industry and period; one study found no significant direct effect of outsourcing peripheral and near-core tasks on firm performance, and another found that neither outsourcing nor internalization per se results in superior performance (Gilley & Rasheed, 2000; Leiblein et al., 2002). It does not support a typical saving, premium, hidden cost or failure rate. 5. The lines fall by subject: the buying of work and services is Book I8's, alliances are Book L4's, platforms are Book A9's, distributors are Book F3's, the financing of the tooling is Book A5's, the ownership of a whole activity is this book's, Book J5 (Supply Chain, Procurement and Inventory) treats sourcing, purchasing, inventory and logistics in its chapter 1, and Book E9 (Tenders, RFPs and Public Procurement) treats public tenders and bid responses in its chapter 1; the thresholds of $93 million and $400 million decide only who must notify, and they are never an exemption, because s. 92 lets the Tribunal act on any merger.

Chapter References

Business Development Bank of Canada. (2022). How to outsource: A decision-making guide for entrepreneurs (updated October 4, 2022). Retrieved October 15, 2026, from https://www.bdc.ca/en/articles-tools/operations/operational-efficiency/outsourcing-pros-cons

Civil Code of Québec, CQLR c CCQ-1991, arts. 1708, 1726, 1729, 2098, 2099, 2100, 2103, 2125, 2126 and 2129 (Québec). LégisQuébec, up to date to August 12, 2026 — read October 15, 2026. https://www.legisquebec.gouv.qc.ca/en/document/cs/CCQ-1991

Coase, R. H. (1937). The nature of the firm. Economica, 4(16), 386–405. https://doi.org/10.1111/j.1468-0335.1937.tb00002.x

Competition Act, R.S.C. 1985, c. C-34, ss. 45, 77, 90.1, 91, 92, 109 and 110 (Canada). Justice Laws Website, current to September 30, 2026 (last amended March 26, 2026) — read October 16, 2026. https://laws-lois.justice.gc.ca/eng/acts/C-34/

Competition Bureau. (2021). Competitor collaboration guidelines. Government of Canada. (Enforcement guidelines dated May 6, 2021; page modified June 27, 2024.) Retrieved October 15, 2026, from https://competition-bureau.canada.ca/en/how-we-foster-competition/education-and-outreach/competitor-collaboration-guidelines

Competition Bureau. (2026, March 2). Pre-merger Competition Bureau notification threshold to remain at $93M in 2026 [News release]. Government of Canada. Retrieved October 15, 2026, from https://www.canada.ca/en/competition-bureau/news/2026/03/pre-merger-competition-bureau-notification-threshold-to-remain-at-93m-in-2026.html

Gilley, K. M., & Rasheed, A. (2000). Making more by doing less: An analysis of outsourcing and its effects on firm performance. Journal of Management, 26(4), 763–790. https://doi.org/10.1177/014920630002600408

Grossman, S. J., & Hart, O. D. (1986). The costs and benefits of ownership: A theory of vertical and lateral integration. Journal of Political Economy, 94(4), 691–719. https://doi.org/10.1086/261404

Hart, O., & Moore, J. (1990). Property rights and the nature of the firm. Journal of Political Economy, 98(6), 1119–1158. https://doi.org/10.1086/261729

Institute for Strategy and Competitiveness. (n.d.). The value chain. Harvard Business School. Retrieved October 15, 2026, from https://www.isc.hbs.edu/strategy/business-strategy/Pages/the-value-chain.aspx

Jacobides, M. G., & Winter, S. G. (2005). The co-evolution of capabilities and transaction costs: Explaining the institutional structure of production. Strategic Management Journal, 26(5), 395–413. https://doi.org/10.1002/smj.460

La Machine. (2026). Documentation — route reference inventory. Internal product documentation, read October 16, 2026, together with the code of the pages this book describes: /outils.

Labour Code, CQLR c C-27, s. 45 (Québec). LégisQuébec, up to date to August 12, 2026 — read October 15, 2026. https://www.legisquebec.gouv.qc.ca/en/document/cs/C-27

Lafontaine, F., & Slade, M. (2007). Vertical integration and firm boundaries: The evidence. Journal of Economic Literature, 45(3), 629–685. https://doi.org/10.1257/jel.45.3.629

Leiblein, M. J., Reuer, J. J., & Dalsace, F. (2002). Do make or buy decisions matter? The influence of organizational governance on technological performance. Strategic Management Journal, 23(9), 817–833. https://doi.org/10.1002/smj.259

Novak, S., & Stern, S. (2008). How does outsourcing affect performance dynamics? Evidence from the automobile industry. Management Science, 54(12), 1963–1979. https://doi.org/10.1287/mnsc.1080.0922

Statistics Canada. (2026, May 27). Canadian Survey on Business Conditions, second quarter 2026. The Daily. Retrieved October 15, 2026, from https://www150.statcan.gc.ca/n1/daily-quotidien/260527/dq260527a-eng.htm

Stuckey, J., & White, D. (1993, Spring). When and when not to vertically integrate. Sloan Management Review. https://sloanreview.mit.edu/article/when-and-when-not-to-vertically-integrate/

Williamson, O. E. (1991). Comparative economic organization: The analysis of discrete structural alternatives. Administrative Science Quarterly, 36(2), 269–296. https://doi.org/10.2307/2393356

Williamson, O. E. (2009, December 8). Transaction cost economics: The natural progression [Prize lecture]. The Nobel Foundation. https://www.nobelprize.org/uploads/2018/06/williamson_lecture.pdf

Chapter 2 — Backward and Forward Integration

A stage next to the core of a business can be reached from two directions, so this chapter says what backward and forward integration are and what each secures, treats the service around a product as a stage of its own, distinguishes full, tapered and quasi-integration, compares building a stage, buying a company and contracting the work, names what ownership commits and the signals that an adjacent stage deserves ownership, and draws the chain of Hydrosentinelle inc. from components to installed sensors on day 1.

What You'll Be Able to Do

• Place the core of a business and say which stages of its chain lie upstream of that core, which lie downstream and which are lateral, and why each direction is a separate decision rather than one decision taken twice.
• Say what a business seeks and what it gives up when it owns a stage that supplies it, and read a study of backward integration at the level the study was read.
• Read a study of forward integration at the level it was read and in its own industry, and resist the pull of a visible margin whose fixed cost, skills and volume are invisible.
• Treat the service around a product — installation, support, monitoring, repair and renewal — as a stage that can be owned, bought or shared, and name who is accountable for it.
• Distinguish full, tapered and quasi-integration, and choose between building a stage, buying a company that already runs it and contracting the work.
• Name what ownership of a stage commits a business to in volume, skills and fixed cost, and write the signals that put an adjacent stage on the grid of Chapter 3.

Why It Matters

Every business sits at the center of a chain it did not design and can still change. Upstream are the stages that supply what the business needs, and downstream are the stages that carry what it makes to the person who pays for it. A stage becomes important when a supplier's quality decides the quality of the product, when a distributor holds the customer relationship the business thought it had, when the service after the sale becomes the reason a customer renews. At that point the question of who owns the stage stops being background and becomes a decision, and the decision is worth writing down.

The two directions are not symmetrical and they are not equally risky. Backward integration buys control of an input and sells the freedom to change supplier; forward integration buys the customer and the margin and takes on the fixed cost, the skills and the volume of a channel. Both are decided one stage at a time, and a small business can take either in a tapered form.

This chapter decides nothing that belongs to its neighbors. The value chain as a strategy tool is Book A3's (chapter 2 of Book A3), the buying of work and services is Book I8's (chapter 2 of Book I8), distributors are Book F3's (chapter 2 of Book F3), and the partners block of a business model is Book A1's. What stays here is the question an owner actually faces: for this stage, on this day, is the right answer to own it, to have it made, to share it or to buy it, and what does each answer commit the business to.

In This Chapter

Section 2.1 places the core of a business and defines upstream, downstream, a lateral stage and the three degrees of integration. Section 2.2 covers backward integration and section 2.3 forward integration. Section 2.4 treats the service around a product as a stage of its own. Section 2.5 distinguishes full, tapered and quasi-integration. Section 2.6 compares building a stage, buying a company and contracting the work. Section 2.7 sets out what ownership commits in volume, skills and fixed cost. Section 2.8 names the signals that put an adjacent stage on the grid of Chapter 3. Section 2.9 compares product firms and service firms. Section 2.10 draws the chain of Hydrosentinelle inc. from components to installed sensors, with one backward and one forward integration set aside.

2.1 Upstream and Downstream: Two Directions From One Core

The core of a business is the activity that the business would not give away. For Hydrosentinelle inc., a fictional company created for this encyclopedia, the core is the firmware and the alert logic: the sensors report a leak within minutes (Book A7), and the logic that decides what a reading means is an asset of the company (Book H5). Everything else in the chain sits on one side of that core or the other, and the work of this chapter begins by saying which side.

Upstream are the stages that supply the core: the components, the boards, the enclosure and its molds, the assembly and the test. Downstream are the stages that carry the output of the core to the customer: the installation, the monitoring service that a building pays for each month, the support and the renewal. A lateral stage is a related activity beside the core rather than before or after it — a second product built on the same firmware, a service offered to the same customer through a different route. The three positions are not ranks. A stage is upstream or downstream of a particular core, and a change in what the business considers its core moves every stage around it.

The reason the directions deserve separate names is that each one is a separate decision. The lecture delivered on December 8, 2009, read in its published text on October 15, 2026, takes the technological core as given and focuses attention on a series of separable make-or-buy decisions, backward, forward and lateral, each of which decides whether a transaction stays within the ownership boundary of the firm (Williamson, 2009). That is a theoretical account and it carries no number.

The chain itself is Book A3's. The page read on October 15, 2026, which shows no date, describes the value chain as a tool that breaks a company into its strategically relevant activities to find where advantage comes from, and places a company's chain inside a larger value system that includes upstream suppliers and downstream channels (Institute for Strategy and Competitiveness, n.d.).

The table below sets this chapter beside the books whose subjects touch the same ground, and says for each neighbor what that book holds and what this chapter holds. Read it as a frontier rather than as a division of importance.

Table 2.1 — Where this chapter stops and its neighbors begin

What the table leaves in this chapter is a decision about a whole stage, and the word whole is doing real work. Buying an afternoon of a freelancer's time is a purchase, and Book I8 (Freelancers, Agencies and Outsourcing) holds it; deciding that the assembly of every kit the business sells will be done by another company, or by the business itself, is a boundary decision and it is this chapter's.

DEFINITION The words this book uses for upstream and downstream, a stage, full integration, tapered integration and quasi-integration

These are this book's working words, and no source read defines them in this sense. Upstream and downstream are the stages that supply the core of a business and the stages that carry its output to the customer. A stage is one whole activity of the chain that can be made, bought or contracted. Full integration is the making of all that a stage needs. Tapered integration is the making of part and the buying of part of the same input or service, at the same time. Quasi-integration is the holding of some benefits of ownership through something less than ownership, such as dedicated tooling, a stake or a long-term agreement. The value chain that these stages sit in is Book A3's (Strategy and Competitive Positioning, chapter 2) and is not redefined here.

2.2 Backward Integration: Securing Inputs, Quality and Cost

A business integrates backward when it takes ownership of a stage that supplies it. The reasons it does so are few and they are practical: continuity of supply; control of quality and of specifications; and a cost that the business can see. Those three are what a business seeks. What it gives up is equally concrete: the freedom to change supplier when a better one appears, the benefit of a specialist's volume and the capital and the skills that the stage itself requires.

That trade is the heart of the matter, and it is this book's own reasoning rather than a finding. Backward integration is worth its price when the input is specific, scarce or central to the quality of the product, and it is a poor bargain when the input is standard, abundant and easy to check. A specific input is one made for this product and hard to sell to anyone else; the molds of Hydrosentinelle inc. are the example, because the enclosure is shaped for these detectors only. A scarce input is one that few suppliers can make at the volume the product needs, and an input central to quality is one whose failure shows up in the customer's hands.

The evidence available on backward integration is narrow and it is read at its narrowest. The abstract of a study of firms' backward integration choices reports that a logit analysis of a sample of thirty-four chemical products is consistent with transactions cost theories and with a demand variability model (Lieberman, 1991, abstract only). It supports no finding outside those products and it is never a forecast for a reader.

The specific assets in a supply relationship carry a further cost that a price list does not show. A supplier that has invested in tooling for this business, or a business that has invested in tooling held at a supplier, has created a relationship in which each side can hurt the other: the supplier can raise its price once the tooling exists, and the business can move the work once the supplier has built the capacity. This book calls that exposure the hold-up problem, and Chapter 3 treats it beside uncertainty and the specificity of assets. Ownership is one answer to it, and a long-term agreement with safeguards, which Chapter 5 develops, is another.

The hybrid forms between buying and owning are Chapter 5. What belongs here is the decision to own the supplying stage outright, and the question an owner asks before taking it is whether the three things sought are worth the three things given up. For Hydrosentinelle inc., the electronic boards come from the electronics supplier of Book A5 and the company has no assembly line for them; the question of making them in-house was put to the grid and the grid returned a buy, a verdict Chapter 3 records.

KEY CONCEPT Backward integration buys control of an input and sells the freedom to change supplier

The central idea of this section is this book's own reading and never a finding of a source. Owning a stage that supplies the business secures continuity, control of quality and a cost that can be seen, and it costs the freedom to change supplier, the volume advantage of a specialist and the capital and skills the stage needs. The bargain is worth taking when the input is specific to the product, scarce at the volume required, or central to the quality the customer experiences, and it is a poor bargain when the input is standard and easy to check. No figure of a source is used here, and no saving is promised from owning a stage.

2.3 Forward Integration: Reaching the Customer and the Margin

A business integrates forward when it takes ownership of a stage between itself and its customer. What it seeks is the customer relationship, the margin that an intermediary takes, and control of the last mile — the installation, the handover, the support call. What it must carry is the volume that a channel needs to run, the skills of a trade it may not know, and the capital that a channel ties up in stock, premises and people.

Two readings bear on the question, each at abstract only and each in a single industry. The first reports that in the electronic components industry, integration of personal selling is associated with higher asset specificity, with the difficulty of evaluating performance and with their combination, while contrary to the transaction cost model neither the frequency of transactions nor the interaction of specificity and uncertainty is significantly related to integration (Anderson & Schmittlein, 1984, abstract only). The second reports that for United States semiconductor companies entering foreign markets, choosing between captive agents and independent intermediaries is a question of downstream vertical integration, that integration is associated with the transaction specificity of assets in the distribution function and with a highly differentiated product, and that a product tends to be sold through the channel already in place (Anderson & Coughlan, 1987, abstract only).

Each of those is an association in one industry and one period, and none of them is a rule. What they have in common is narrow and useful: where the assets used to sell are specific to the product and where performance is hard to judge from outside, firms lean toward owning the selling stage. That does not say that a reader should.

The trap in this direction is the shape of the arithmetic itself. The margin an intermediary takes is visible on every invoice, and the fixed cost that the intermediary carries — the warehouse, the vehicles, the sales team, the slow months — is not. The remedy is unglamorous: count the cost of owning the stage on the same basis as the price of buying it, one activity at a time, which is what Chapter 3 does with its grid and Chapter 7 does with its numbers.

Distributors and wholesale channels are Book F3's (chapter 2 of Book F3); the arrangement in which another company sells under its own brand is Book F2's (chapter 2 of Book F2); and the direct channel that a digital business keeps beside its partners is Chapter 6. What belongs here is the decision to own the downstream stage rather than to sell through it, and the case gives a modest example: Hydrosentinelle inc. installs through its staff installer and through the certified installers of Book F9 (Building and Running a Partner Program), and the question of whether installation is a stage the company should own outright is one the grid scores as a hybrid, which Chapter 3 records.

The table below sets the three readings on integration side by side at the level each was read, with its setting and the limit of what its abstract supports. Each of these studies is one industry, and a finding carried into another setting is something the study does not offer.

Table 2.2 — Three readings on integration, each in its industry

WATCH OUT A forward stage looks like margin until its fixed cost, its skills and its volume are counted

The trap of this section is the forward stage that looks like free margin. The intermediary's margin appears on every invoice the business sends, and the intermediary's costs do not, so the arithmetic of owning the stage looks better than the arithmetic of buying it before a single hour has been counted. Two signs say that the comparison has been done too quickly: a gain computed on a sales volume the business does not have, and a channel that nobody on the team has ever run. The remedy is to count the cost of the stage on the same basis as the price of buying it, one activity at a time, which is what Chapters 3 and 7 do. The reasoning here is this book's own and carries no figure.

2.4 Integrating the Service Around a Product

A product is surrounded by a service, and the service is a stage of the chain rather than an afterthought. The installation of the thing sold, the support that follows it, the monitoring that watches it, the repair when it fails and the renewal that keeps the relationship alive can each be owned, bought or shared, and each is a stage with its own volume, skills and fixed cost.

The difference between owning a product and owning what happens after the sale is a difference of recurring duties. A product is sold once, and a service owned is a promise that repeats, with a queue of calls, a schedule of maintenance visits, a monthly report and an obligation to answer.

This section rests on this book's own reasoning and not on a source. The service stages of a chain have the same three constraints as the material ones — control, capability and capital, as Chapter 1 sets them out — and the same grid applies to them in Chapter 3. A service stage that a business owns is scored on the same six criteria as a factory stage, and the verdict for the service can be a make while the verdict for the hardware next to it is a buy.

The case gives a worked picture without a number. Hydrosentinelle inc. runs the monitoring service and owns the alert logic (Books H5 and A7), and it installs through its staff installer and through the certified installers of Book F9 (Building and Running a Partner Program). The firmware and the alert logic stay in-house, and Book I8 (Freelancers, Agencies and Outsourcing) records that only the testing of the gateway firmware is rented under control. The company therefore owns the part of the service that carries its knowledge, and it shares the installation in buildings it does not own with the certified installers of Book F9.

WHO DOES THIS One person is accountable for each stage, and in the smallest business that person is the owner

A service stage is owned by whoever answers for it when it fails. In a solo business the owner is accountable for every stage, and the practical discipline is to keep a short written list of the stages and the promise each one carries. In a small team the owner assigns each stage to one person and writes the assignment in a place the team can read, because a stage assigned to a team is a stage assigned to nobody. In an established business a service lead owns the stage, reports its readings and answers to the owner, which is the arrangement Chapter 9 sets out for expertise and roles. In the case, and this is posed by this book, the staff installer (Book F9's illustrative rate of $45 an hour) installs beside the certified installers of Book F9 (Building and Running a Partner Program), and no hire and no amount follow from that arrangement.

2.5 Full, Tapered and Quasi-Integration

Ownership of a stage comes in degrees, and the degrees are worth naming because they are what makes an integration decision practical rather than all-or-nothing. Full integration is the making of all that a stage needs: every kit the business sells is assembled in its own workshop, every board is built in its own line. Tapered integration is the making of part and the buying of part of the same input or service at the same time. Quasi-integration is this book's own name for holding some benefits of ownership through something less than ownership — dedicated tooling held at a supplier, a minority stake, a long-term agreement with reserved capacity — and Chapter 5 develops it as the hybrid middle between buying and owning.

The three degrees are not stages of a journey from buying to owning; they are three answers that can each be right for a different stage of the same chain. A business can own its firmware fully, taper its assembly and hold its molds quasi-integrated at a molder, and the three answers can all be correct at once.

Four readings bear on the degrees, each at abstract only. The framework of hypotheses developed in a review article proposes the dimensions of vertical integration strategies and the key factors that might augment their use, and its author states that these are hypotheses and conjectures about make-or-buy decisions that require empirical testing, and that strategists could formulate better hybrid strategies if the framework is valid (Harrigan, 1984, abstract only). A later study reports that the business units studied made fewer products and services in-house, and firms were engaged in fewer stages of processing, when demand was highly uncertain, that internal transfers to adjacent units were more numerous when synergies were substantial, and that firms with high market shares sought higher ownership stakes in adjacent stages (Harrigan, 1985, abstract only) — associations in the units studied and never a cause. A synthesis of theory and evidence on vertical financial ownership relative to vertical contracts reports that without agency and transaction costs the two are equivalent, and that with them the question becomes when market mechanisms suffice, when intermediate forms of vertical contracting become necessary and when ownership is preferred (Mahoney, 1992, abstract only); it proposes a framework and measures nothing. And on a panel of more than 3,500 product introductions in the global microcomputer industry, balancing vertical integration and strategic outsourcing — which that abstract calls taper integration — is associated with a richer product portfolio, product success and superior firm performance (Rothaermel et al., 2006, abstract only).

None of those four is a rule. The first is a framework awaiting tests, the second reports associations in a set of business units, the third is a synthesis, and the fourth reports an association in one industry.

The table below sets the four readings side by side, with the setting of each and the limit of what its abstract supports.

Table 2.3 — Four readings on degrees of integration, at the level read

Tapering has an administrative cost that full integration does not. A business that makes part of a stage and buys part of it runs two sets of records, two sets of quality checks and two relationships with the same output, and it should write down why it keeps the internal part, whether for repairs, pilot builds, urgent orders or knowledge that must not be lost, rather than keeping it out of habit.

THE RULE An association is printed as an association, with its sample (Rothaermel et al., 2006)

The abstract of Rothaermel et al. reports that on a panel of more than 3,500 product introductions in the global microcomputer industry, taper integration is associated with a richer product portfolio, product success and superior firm performance (Rothaermel et al., 2006, abstract only). That sentence is an association in one industry over one period and it is never a cause: the panel does not show that tapering produced the result, and it says nothing about any other industry. The case's split of 207 kits bought and 23 made, posed by this book in section 2.10, is posed; it credits nothing, it promises nothing, and it is never an illustration of what tapering will do for another business.

2.6 Building a Stage, Buying a Company or Contracting the Work

When a stage has been judged worth owning, there are three routes to it, and they commit the business to very different things. The first is to build the stage: buy the equipment, hire or train the people, learn the process, and take the time to reach the quality the product needs. The second is to buy a company that already runs the stage, which transfers a working operation at once and transfers its obligations with it. The third is to contract the work: an agreement, an oversight arrangement and a supplier who owns the stage.

The question of which route to take is the question the synthesis read for section 2.5 poses — when market mechanisms suffice, when intermediate forms of vertical contracting become necessary and when ownership is preferred (Mahoney, 1992, abstract only). That is a framework and not a measured result, and it answers nothing by itself; it names the three answers an owner is choosing between. Table 2.4 sets out what each route commits and what it keeps open.

Buying a company is a merger in the language of Canadian competition law, and the provision is worth stating once at the level it was read. Section 91 of the Competition Act, R.S.C. 1985, c. C-34, read on the Justice Laws Website on October 16, 2026, at its consolidation of September 30, 2026, last amended on March 26, 2026, defines a merger as the acquisition of control over or a significant interest in the whole or part of a business of a competitor, supplier, customer or other person, so that the purchase of a supplier or of a customer is a merger. Section 92(1) lets the Tribunal act on a merger that prevents or lessens competition substantially, including among the sources from which a trade obtains a product and among its outlets, and it applies to a merger of any size. Sections 109(1) and 110 make pre-merger notification apply only above the party-size threshold of $400 million and the transaction-size threshold. Those provisions are called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements), which holds competition law, and this book draws no conclusion about any real purchase.

The threshold that governs notification is a rule about who must notify and never an exemption from review, and its current value is a dated fact. The news release of March 2, 2026, read on October 15, 2026, reports that the transaction-size threshold for pre-merger notification stays at $93 million, measured on the assets in Canada of the business acquired or the revenues they generate from sales in, from or into Canada, and that notice is also generally required when the parties and their affiliates exceed $400 million in combined assets or revenues (Competition Bureau, 2026). A purchase below those thresholds can still be examined under section 92, and the thresholds decide only who must notify.

The table below sets the three routes side by side, with what each commits, what it keeps open and where the matter is treated. Read the second column as the price of the route and the third as its flexibility.

Table 2.4 — Three routes to a stage and what each commits

The second route, buying a company, is named here by the code of its book alone, and the competition-law dimension of it belongs to Book H4 (chapter 5 of Book H4). The case makes no acquisition: Hydrosentinelle inc. buys boards from the electronics supplier of Book A5, holds a mold at a molder, contracts assembly and test, and owns its firmware and alert logic, so that no question of notification arises for the company.

QUÉBEC & CANADA NOTE Buying a supplier or a customer is a merger in the Act's words, and Book H4 holds what follows

The Canadian point of this section is the reach of the merger provisions. Section 91 of the Competition Act, R.S.C. 1985, c. C-34, current to September 30, 2026, read on October 16, 2026, defines a merger as the acquisition of control over or a significant interest in the whole or part of a business of a competitor, supplier, customer or other person, and section 92 applies to a merger of any size. The pre-merger notification thresholds decide only who must notify and are never an exemption from review. That law is called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements), which holds it, and no conclusion about any real purchase is drawn here. The case makes no acquisition. Book M6 (Buying a Business) is only named. This book recommends no vendor, no supplier, no adviser and no tool.

WHAT CHANGED The transaction-size threshold for pre-merger notification stays at $93 million in 2026 (Competition Bureau, 2026)

The dated change comes from the news release of March 2, 2026, read on October 15, 2026: as of 2026 the transaction-size threshold for pre-merger notification stays at $93 million, measured on the assets in Canada of the business acquired or the revenues they generate from sales in, from or into Canada, and notice is also generally required when the parties and their affiliates exceed $400 million in combined assets or revenues (Competition Bureau, 2026). The thresholds decide only who must notify; they are never an exemption from review, because a merger of any size can be examined under section 92 of the Competition Act, and the law is called to Book H4. The threshold can change, and it is printed here with the date of the release that states it and never as a permanent figure. The date is a date of the source and never a date of the case.

2.7 What Ownership Commits You To: Volume, Skills and Fixed Cost

Owning a stage commits a business to three things that buying the same output does not, and each of the three has a way of becoming expensive quietly. The first is volume: a stage that the business owns must be kept busy, and a stage running at a fraction of its capacity costs more per unit than the price it was replacing. The second is skills: the know-how needed to run the stage has to stay in the business, which means hiring, training and keeping people whose work is not the business's core work. The third is fixed cost: the cost of the stage does not fall when the volume does, so a slow quarter is felt on the cost side as well as on the revenue side.

Skills leave and cannot be bought back at a moment's notice: a business that loses the two people who ran a stage has the equipment and no longer has the stage. Fixed cost is felt at the first slow month, when the equipment payment falls due beside a smaller order book. Both are the reason that full integration of a stage is a heavier decision than the arithmetic of a unit cost suggests.

The practice guidance of a federal Crown corporation, read on October 15, 2026, on a page that shows no date, treats equipment as a decision to be planned within a capital expenditure plan, and it names the components that a total cost of ownership includes: the principal, floor space, financing interest, installation, training and the downtime associated with maintenance (Business Development Bank of Canada, n.d.). It is that author's own practice guidance and never a rule or an obligation. The same page quotes a vice-president of financing of the corporation saying that a large piece of equipment can take six, eight or 12 months to be made and that, with progress payments to a foreign manufacturer, 90 percent of the money can be out the door before the equipment is on site. Those are the quoted words of one vice-president, they are never a norm, and they are never a figure of the corporation. The page also notes that outsourcing may be preferable when the equipment would serve only a short period or a small part of the work, and it prints no cost benchmark.

The kit of Hydrosentinelle inc. gives the three commitments a concrete shape without adding a figure. The tooling and test equipment of the company cost $95,000 and were capital spending, as Book A5 (Financing a Business) prints, and the enclosure's molds would need to be replaced or redesigned within about five years, with few buyers existing for a mold made for someone else's product (Book A5). Those two facts together say what ownership of the tooling commits: a capital sum that has been spent, a set of molds that must be replaced on a schedule the company does not control, and a resale market that is thin. The financing of the tooling and its loan are Book A5's, cited here and never retold.

Volume, skills and fixed cost are also what a tapered arrangement buys back. A business that keeps a small in-house cell and buys the rest of the volume carries a smaller fixed cost and keeps the option of bringing the rest of the work back, at the price of running two arrangements. That trade is written down in Chapter 3 and priced in Chapter 7.

TOOLS OF THE TRADE What a business needs to see what it owns: a list of assets, a replacement calendar, a usage record and one register

Four functions cover what this section asks an owner to keep, and they are functions rather than products. The first is a list of the assets the business owns, each with its location and its date of purchase, so that the equipment held at a molder or a supplier is not forgotten. The second is a calendar of the dates on which an asset must be replaced or reviewed, because a mold that must be replaced has a date, even when the date is approximate. The third is a place to record the hours and the volume that each owned stage actually uses, so that a stage running at a fraction of its capacity is visible before its cost is read. The fourth is one register of decisions, where each owned stage is written with the triggers that would send it back to the grid. Chapter 10 gives these functions in full, with their features and their pricing models, without amounts.

2.8 Signals That an Adjacent Stage Deserves Ownership

Five signals put an adjacent stage on the grid of Chapter 3. The first is a supplier who has already hurt the business — a quality problem that reached a customer, a shortage that stopped assembly. Book G8 (Business Insurance and Risk Financing) rates one such shortage for the case at $40,000 and 30 days if a supplier's component stops kit assembly, reduced by a second source and a safety stock, and that is a printed fact of the case and not a forecast. The second signal is a stage whose price the business cannot check, because the supplier is the only source and there is no comparable quote. The third is a stage whose know-how is the business's own and would be lost if the stage left. The fourth is a volume large enough to keep an owned unit busy. The fifth is a market so young, so small or so uncertain that no supplier yet exists for the stage at all.

The signals that argue against are equally concrete. A demand the business cannot forecast makes an owned stage a fixed cost against a moving revenue. A skill nobody in the business has makes the stage a hiring problem before it is a production problem. A capital sum the business needs elsewhere — for the core, or for the working capital that growth consumes — makes the stage a competitor for the money rather than a use of it. And a stage that a supplier already runs well at a price the business can check is a stage that ownership is unlikely to improve.

Two bodies of work bear on these signals, each at the level it was read. The position of two consultants, published on April 15, 1993, and read at its opening only on October 15, 2026, with nothing printed here from the parts not visible, is that vertical integration can be important but is hard to implement and costly to reverse, and that management's record on it is described as poor (Stuckey & White, 1993). That is the position of two consultants and never evidence or a rule. The study of business units reports that the units studied made fewer in-house products and stages when demand was highly uncertain (Harrigan, 1985, abstract only), and the study of thirty-four chemical products reports choices consistent with transactions cost theories and a demand variability model (Lieberman, 1991, abstract only). None of the three tells a reader what to do about a particular stage.

The signals are useful because each can be written on a line with a date, and a line with a date can be reviewed. Chapter 3 scores the activity on its grid, and Chapter 8 sets the cadence of the review.

METHOD Reading whether an adjacent stage deserves ownership in five steps

First, name the adjacent stage and the activity it actually performs, in one sentence, so that the decision is about a stage and not about a vague ambition. Second, write the volume the stage would carry and the skills it would need, in units and in roles, because a stage with no volume and no named skill is a stage that cannot be planned. Third, write what the business would lose if the stage failed, in the customer's terms, because a stage whose failure stops a promise matters more than one whose failure delays a report. Fourth, compare ownership with a hybrid form and with a purchase, at the level of the same activity, using the forms of Chapter 5 and the grid of Chapter 3. Fifth, write the trigger that sends the stage to the grid, with a date or an event, so that the review happens on its own rather than when someone remembers. This sequence is this book's own; no source tests it, and no figure follows from it.

2.9 Integration in Product Firms and in Service Firms

The same three limits — control, capability and capital — govern integration in every business, and they take a different shape depending on whether the stages are mostly things or mostly people. In a product firm the stages are components, tooling, assembly, stock and delivery, and ownership of a stage means assets and floor space: a molding machine, a bench, a rack of parts, a vehicle. In a service firm the stages are the front office, the delivery team and the specialist skill, and ownership of a stage means employees and know-how: a trained crew, a method that lives in their hands, a standard that has to be taught.

That difference changes nothing about the grid and everything about the exit. Equipment is hard to sell when it is made for one product, as Book A5 notes when it observes that few buyers exist for a mold made for someone else's product. A team, once assembled and trained, cannot be sold; bringing a service stage back inside the business or moving it out means moving people, and that is a decision with an employment dimension that Book H7 (Employment Law Essentials (Québec and Canada)) holds.

A business can be both kinds at once, and the case is. Hydrosentinelle inc. is a product firm with a monitoring service, and its chain contains stages of each kind: boards and molds and assembly on one side, installation and support and monitoring on the other. The two sides are scored on the same grid, with the same six criteria, and the verdict for each stage is read on its own.

For a service stage, the practical question is whether the business can keep the method without keeping the people. A method that has been written down, such as a checklist, a script or a standard for a good installation, is what allows a service stage to be shared with partners, contracted out or brought back in. The installers' program of the case is Book F9's (Building and Running a Partner Program, chapter 4 of Book F9), and the buying of work from freelancers and agencies is Book I8's (chapter 2 of Book I8); this chapter cites them and does not repeat them.

2.10 The Case's Chain From Components to Installed Sensors

The chain of Hydrosentinelle inc. runs from a sensor component that the company buys to a monitoring service that a building pays for each month, and the company sits at its center with the firmware and the alert logic that it owns. Part of the chain is printed by published books of this encyclopedia and part of it is posed by this book, and the two are separated here so that a reader can tell which is which. The chain is mapped by day 9 of the plan of Chapter 11, with the technical side of the map done by Laurence Bergeron, the president, product and engineering (Book A5); no hour is given to the map in this chapter.

What is printed: the electronic boards come from the electronics supplier of Book A5; the tooling of $95,000 was financed, and the enclosure's molds would need to be replaced or redesigned within about five years, with few buyers existing for a mold made for someone else's product (Book A5); the kits are assembled and tested in the company's own workshop, and the workshop, the stock of detectors and components and the test bench are among the company's exposures (Book G8); installation is done by the staff installer and by the certified installers of Book F9 (Building and Running a Partner Program); the sensors report a leak within minutes (Book A7); and the firmware of the sensors and of the gateway, and the mobile application, are assets of the company (Book H5). What is posed by this book: the seven stages as the company organizes them on day 1 and by the end of the plan, and the two integrations that were examined and set aside.

The table below sets the seven stages of the chain in the order they run, with who does each, whether the stage lies upstream or downstream of the core, and what this book poses about it. Read it as a map rather than as a set of verdicts: the verdicts are Chapter 3's, and the table says only who holds each stage and where a figure in it is posed by this book.

Table 2.5 — The case's chain on day 1: seven stages and who does each

Two integrations were examined and set aside, and both are posed by this book. The backward one is assembling the electronic boards in-house: it would need an assembly line that the company has neither the skill nor the volume for, and the grid of Chapter 3 scores it as a buy. The forward one is buying a firm of certified installers: it would need capital, it would be a merger under section 91 of the Competition Act called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements), and it would put the company in competition with the installers of Book F9 (Building and Running a Partner Program) whose program the company runs. The case makes no acquisition, and nothing in the case involves a question of notification.

Figure 2.1 follows the case's chain from components to installed sensors and shows the two integrations the company examined and set aside.

Figure 2.1 — The case's chain. Boards and hosting bought, firmware and alert logic made, molds owned at the molder, assembly bought at 90 percent, installation shared; one backward and one forward integration set aside (posed).

Read from the left, the figure shows the sensor component and the electronic boards bought, the enclosure and its molds owned by the company and held at the molder, the firmware and the alert logic made in-house at the center, kit assembly and test bought at 90 percent, installation shared, and the monitoring service made with its hosting rented. Above the chain, dashed, sits the backward option set aside, and below it the forward option.

In La Machine, as of the application copy of September 7, 2026, the page that touches this chapter is CRM → Companies (business/marketing/companies), where a supplier, a molder or an installer can be recorded as a company with its sector, size, site, status and notes. It holds no purchase order, no bill of materials, no supplier scorecard, no inventory and no production order; check it in your installation. Hydrosentinelle inc. does not use La Machine, as Books G8, I8 and L5 record.

CASE IN POINT Hydrosentinelle inc. draws its chain on day 1 and sets aside one backward and one forward integration

Hydrosentinelle inc. is a fictional business created for this encyclopedia; its founders, its employees, its customers, its partners, its suppliers, its figures and the events described are invented, and any resemblance to a real business is coincidental. Seven stages run from the sensor component that the company buys, through the electronic boards bought from the electronics supplier of Book A5, the enclosure and its molds owned by the company and held at the molder, and the firmware and the alert logic made in-house (Books H5 and A7). Kit assembly and test is bought for 207 kits and kept in-house for 23. Installation is done by the staff installer and by the certified installers of Book F9. The monitoring service is made by the company while its hosting is rented. Two integrations were examined and set aside, both posed by this book: assembling the electronic boards in-house, which would need a line the company has neither the skill nor the volume for, and buying a firm of certified installers, which would need capital and would compete with the installers of Book F9. Laurence Bergeron, the president, product and engineering, holds the technical side of the map. No result is credited to any of it.

DO THE MATH Two hundred thirty kits, 207 bought and 23 kept: the tapered stage of the case in numbers

Illustrative figures; the case is fictional. Step 1 — The forecast of the year is 230 kits, posed by this book. Step 2 — The contract manufacturer builds 207 kits and the company builds 23 in-house, and 207 plus 23 are 230. Step 3 — 207 of 230 kits is 90 percent, which is the share the company buys. Step 4 — 23 of 230 kits is 10 percent, which is the share the company keeps. Step 5 — The stage is tapered, because most of it is bought and a small part is kept in-house for repairs, pilot builds and urgent kits, and every count here is posed by this book and illustrative, never measured and never a figure of any real business. The rates of the people who work the kept part are printed by Books F9 and G7, and no other amount follows from these steps.

IN LA MACHINE La Machine can record a supplier or a molder as a company, with no field for a contract, a price or a performance

In the application copy of September 7, 2026, read in its code on October 15, 2026, CRM → Companies (business/marketing/companies) holds one record for each company, with its name, sector, size, site, status and notes, and it has no field for a supplier, a contract, a price, a tooling or a performance (La Machine, 2026). A reader can record a supplier, a molder or an installer there as a company, and it is not supplier management. There is no supplier record or scorecard, no purchase order, no bill of materials, no production order, no stock, no register of tooling and no make-or-buy model. La Machine is described and never sold, the case does not use it, and this book is valid with any tool. A business should check it in its own installation.

Costly Mistakes

• Treating the two directions as one decision. A choice made upstream does not settle the question downstream, and a business that decides once for the whole chain has decided twice without looking.
• Owning a stage for the margin and forgetting the fixed cost. The intermediary's margin is visible on every invoice and its warehouse, vehicles and slow months are not, so the arithmetic looks better than the decision is.
• Letting an ownership decision go unchallenged. A decision to own a stage should be argued against, stage by stage, before it is taken, with the three things sought set beside the three given up.
• Reading an association as a cause. A panel of product introductions in one industry, or a set of business units in one study, reports an association in its own setting; it is never a rule for another business.
• Building a stage without naming the skills it needs. Equipment can be bought in a quarter and a skill cannot, and a stage whose two trained people leave is equipment rather than a capability.
• Buying a company without knowing what a merger is. The purchase of a supplier or of a customer falls under the merger provisions of the Competition Act, called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements), and the notification thresholds decide only who must notify, never whether the transaction can be reviewed.
• Writing a decision without its trigger. A stage judged worth owning on a day when a supplier failed is a stage to be re-examined on the day the conditions change, and an undated decision is never re-examined.

Key Takeaways

• The core of a business is the activity it would not give away; upstream stages supply the core, downstream stages carry its output to the customer, and a lateral stage sits beside it, with every direction a separate decision.
• Backward integration buys control of an input and sells the freedom to change supplier, and it is worth its price when the input is specific, scarce or central to the quality the customer experiences.
• Forward integration buys the customer relationship and the margin and takes on the volume, skills and fixed cost of a channel, and the trap is to compute the gain on a volume the business does not have.
• Ownership comes in degrees — full, tapered and quasi-integration — and a business can taper one stage while owning another fully, at the cost of running two arrangements for the same output.
• There are three routes to a stage: build it, buy a company that runs it, or contract the work; the purchase of a supplier or of a customer is a merger under the Competition Act, called to Book H4, and the notification thresholds decide only who must notify.
• Ownership commits volume, skills and fixed cost, and a service stage commits people and know-how where a product stage commits equipment.

Your Action Plan

1. Write the core of the business in one sentence and place every stage around it. Name the activity the business would not give away, then list the stages that supply it, the stages that carry its output to the customer, and any lateral stage. Deliverable: a one-page chain map with the core marked and each stage labeled upstream, downstream or lateral. Verification: every stage on the map carries a label, and the core sentence is one the owner would repeat to a customer.
2. Mark each stage as owned, made, shared or bought, and say which of the two directions each decision takes. For every stage, write who does it today and whether the current answer was chosen or inherited. Deliverable: a list of stages with the current arrangement beside each one. Verification: at least one stage is marked as inherited rather than chosen, or the owner has written that none is.
3. Take the five ownership signals and date each one. For every adjacent stage, write whether a supplier has already caused a failure, whether the price can be checked, whether the know-how is the business's own, whether the volume would keep a unit busy, and whether any supplier exists at all. Deliverable: a dated signal sheet, one line per stage. Verification: every line carries either a date of an event or an explicit statement that no such event has occurred.
4. Write the two integrations you have set aside, with the reason. Name one backward stage and one forward stage that the business examined and did not take, and write the reason in one sentence each. Deliverable: two written set-asides with their reasons. Verification: each reason names a constraint — control, capability or capital — rather than a preference.
5. Write the trigger that would bring each set-aside back for review. For each stage set aside, write the event or the date that sends it to the grid of Chapter 3. Deliverable: one trigger per set-aside, in the register of decisions. Verification: every trigger can be observed by a second person, and each has a place where it will be read again.

In La Machine

In the application copy of September 7, 2026, read in its code on October 15, 2026, the page that touches this chapter is CRM → Companies (business/marketing/companies), which holds one record for each company with its name, sector, size, site, status and notes (La Machine, 2026). An owner can record a supplier, a molder or a firm of installers there as a company.

What the platform does not do for this chapter's subject is hold the decision or the evidence behind it. La Machine has no field for a supplier, a contract, a price, a tooling or a performance, and it computes no make-or-buy verdict and no cost of a stage. The chain map, the set-asides and the triggers of this chapter live in a document or a spreadsheet the business keeps, and this book is valid with any tool. La Machine is described and never sold; the case does not use it. Check it in your installation.

Self-Check

1. What is the core of a business, and how are the stages upstream, downstream and lateral of it distinguished?
2. What does backward integration buy, what does it sell, and when is it worth its price?
3. What does forward integration buy and what must a business carry, and what is the trap that makes it look cheaper than it is?
4. What are the three degrees of integration, and what are the three routes to a stage?
5. What does ownership of a stage commit a business to, and which signals say that an adjacent stage deserves ownership?

Answers. 1. The core is the activity the business would not give away; upstream stages supply the core, downstream stages carry its output to the customer, and a lateral stage is a related activity beside the core, with each direction decided separately (Williamson, 2009). 2. Backward integration buys control of an input — continuity, quality and a cost that can be seen — and sells the freedom to change supplier, the volume advantage of a specialist and the capital and skills the stage needs; it is worth its price when the input is specific, scarce or central to the quality of the product. 3. Forward integration buys the customer relationship, the margin the intermediary takes and control of the last mile, and the business must carry the volume a channel needs, the skills of a trade and the capital a channel ties up; the trap is that the intermediary's margin is visible and its fixed cost is not, which is why the cost of the stage is counted on the same basis as the price of buying it, in Chapters 3 and 7. 4. The three degrees are full integration, tapered integration and quasi-integration; the three routes are building a stage, buying a company that already runs it, and contracting the work, and the purchase of a supplier or of a customer is a merger under section 91 of the Competition Act, called to Book H4, with the notification thresholds deciding only who must notify. 5. Ownership commits volume, skills and fixed cost; the signals that a stage deserves ownership are a supplier that has already hurt the business, a price that cannot be checked, a know-how that is the business's own, a volume large enough to keep an owned unit busy, and a market in which no supplier yet exists, while an unpredictable demand, a missing skill and capital needed elsewhere argue against.

Chapter References

Anderson, E., & Coughlan, A. T. (1987). International market entry and expansion via independent or integrated channels of distribution. Journal of Marketing, 51(1), 71–82. https://doi.org/10.1177/002224298705100106

Anderson, E., & Schmittlein, D. C. (1984). Integration of the sales force: An empirical examination. The RAND Journal of Economics, 15(3), 385–395. https://doi.org/10.2307/2555446

Business Development Bank of Canada. (n.d.). Manufacturing equipment: 6 steps to plan your purchase. Retrieved October 15, 2026, from https://www.bdc.ca/en/articles-tools/money-finance/get-financing/manufacturing-equipment-steps-plan-purchase

Competition Act, R.S.C. 1985, c. C-34, ss. 45, 77, 90.1, 91, 92, 109 and 110 (Canada). Justice Laws Website, current to September 30, 2026 (last amended March 26, 2026) — read October 16, 2026. https://laws-lois.justice.gc.ca/eng/acts/C-34/

Competition Bureau. (2026, March 2). Pre-merger Competition Bureau notification threshold to remain at $93M in 2026 [News release]. Government of Canada. Retrieved October 15, 2026, from https://www.canada.ca/en/competition-bureau/news/2026/03/pre-merger-competition-bureau-notification-threshold-to-remain-at-93m-in-2026.html

Harrigan, K. R. (1984). Formulating vertical integration strategies. The Academy of Management Review, 9(4), 638–652. https://doi.org/10.2307/258487

Harrigan, K. R. (1985). Vertical integration and corporate strategy. Academy of Management Journal, 28(2), 397–425. https://doi.org/10.2307/256208

Institute for Strategy and Competitiveness. (n.d.). The value chain. Harvard Business School. Retrieved October 15, 2026, from https://www.isc.hbs.edu/strategy/business-strategy/Pages/the-value-chain.aspx

La Machine. (2026). Documentation — route reference inventory. Internal product documentation, read October 16, 2026, together with the code of the pages this book describes: business/marketing/companies.

Lieberman, M. B. (1991). Determinants of vertical integration: An empirical test. The Journal of Industrial Economics, 39(5), 451–466. https://doi.org/10.2307/2098455

Mahoney, J. T. (1992). The choice of organizational form: Vertical financial ownership versus other methods of vertical integration. Strategic Management Journal, 13(8), 559–584. https://doi.org/10.1002/smj.4250130802

Rothaermel, F. T., Hitt, M. A., & Jobe, L. A. (2006). Balancing vertical integration and strategic outsourcing: Effects on product portfolio, product success, and firm performance. Strategic Management Journal, 27(11), 1033–1056. https://doi.org/10.1002/smj.559

Stuckey, J., & White, D. (1993, Spring). When and when not to vertically integrate. Sloan Management Review. https://sloanreview.mit.edu/article/when-and-when-not-to-vertically-integrate/

Williamson, O. E. (2009, December 8). Transaction cost economics: The natural progression [Prize lecture]. The Nobel Foundation. https://www.nobelprize.org/uploads/2018/06/williamson_lecture.pdf

Chapter 3 — Make-or-Buy Decisions

A make-or-buy decision is taken one activity at a time and on more than the supplier's price, so this chapter frames the decision for one activity, counts total cost beyond the quote, reads control, capability and risk, explains specific assets, uncertainty and the hold-up problem, shows what it means to make and buy the same thing at once, scores activities on one grid of six weighted criteria, writes the decision so that it can be reviewed, reads each study in its own setting and each theory as a theory, and scores the six activities of Hydrosentinelle inc. on day 24 and signs their verdicts on day 28, all posed by this book.

What You'll Be Able to Do

• Frame a make-or-buy decision for one whole activity, or one part or one family of parts, rather than for the business as a whole, and say why the unit of decision is that small.
• Count the total cost of an activity beyond the supplier's quote, on the same basis on both sides, and name the lines that a price alone leaves out.
• Read control, capability and risk as three separate questions, and say what a business can direct, what it can only ask for and what it can verify.
• Explain a specific asset, uncertainty and the hold-up problem in plain words, and name the remedies that keep a specific investment from being held to ransom.
• Score each activity on one grid of six weighted criteria, read the total against the rule, and write the verdict with its reason, its owner and its triggers.
• Write a decision record that a second person can read a year later, and say what makes a decision reviewable rather than merely remembered.

Why It Matters

A make-or-buy decision can be taken twice: once when the business is small and buys everything, and once when it has grown and wonders whether it should make the thing itself. The first decision is taken by default, because a young business has no capacity and no choice. The second decision can be taken under pressure, for example after a late delivery or a price increase, and for the whole business at once: should the company make its own boards, run its own assembly, own its own molds? That is the wrong unit. The right unit is one activity, or one part, or one family of parts, because each activity has its own volume, its own skill and its own risk, and a business that decides all of them together decides none of them well.

The reason the unit matters is that the answer differs from activity to activity inside the same company. A business may be right to keep the firmware in-house, right to buy the boards, right to share the installation with partners and right to hold the molds it paid for. Those four verdicts sit side by side in one chain, and each rests on a different reason. A decision taken for the whole chain at once would have to average those reasons, and an average of a capability, a cost and a risk is not a decision.

The cost of getting the unit wrong is paid slowly. A business that decides its whole chain at once may keep too much, because keeping is the default and the sunk cost of the workshop argues for it, or buy too much, because a quote looks cheaper than a cost nobody has counted. Both errors are hard to see in the year they are made and easy to see three years later, when a skill has left, a supplier has raised a price, or a specific asset sits at a partner who knows the business cannot move it. This chapter exists to make the decision small enough to be written down, scored and reviewed, and to keep the reasoning beside the verdict so that the next person who reads it can see why it was taken.

In This Chapter

Section 3.1 frames the decision one activity at a time and sets the frontier between this chapter and its neighbors. Section 3.2 counts total cost beyond the supplier's price. Section 3.3 reads control, what a business can direct and what it can only ask for. Section 3.4 reads capability, what a business does better and what it could learn. Section 3.5 reads risk, supply, quality, continuity and knowledge. Section 3.6 explains specific assets, uncertainty and the hold-up problem. Section 3.7 shows what it means to make and buy the same thing at once. Section 3.8 scores activities on one grid of six weighted criteria. Section 3.9 writes the decision so that it can be reviewed. Section 3.10 scores the six activities of Hydrosentinelle inc. and signs their verdicts.

3.1 Framing the Decision One Activity at a Time

The unit of a make-or-buy decision is one whole activity, or one part, or one family of parts, and never the business as a whole. An activity is one whole piece of the chain that a business could make, buy or contract: the assembly of a kit, the tooling of an enclosure, the installation of a product in a building, the hosting of a service, the writing of firmware. Each of those activities has its own volume, its own skill and its own risk, and each can be decided on its own. A business that decides its whole chain in one meeting is deciding a dozen different questions with one answer, and the answer will fit some of them and miss the rest.

The reason the small unit is the right one is that the reasons for making and the reasons for buying do not travel together. An activity may be central to what the product is and still be cheap to buy; another may be peripheral and still be worth keeping because the know-how would leave with it. The abstract of one article read for this book reports a framework for the decision on an individual part or a family of parts, developed from a review of the principal make-or-buy approaches and illustrated with case studies (Cánez et al., 2000, abstract only). The abstract of another reports that in a sample of make-or-buy decisions on components in one division of a United States automobile company, comparative production costs between buyer and supplier were the strongest predictor of the decision, while volume uncertainty and competition in the supplier market had small but significant effects (Walker & Weber, 1984, abstract only). Neither finding is a rule for any other business: the first is a framework, and the second is one division of one company in one period. What both support is the shape of the decision, taken part by part and not company by company.

A third reading sets the expectation for what a decision can promise. The abstract of a study read for this book reports that no significant direct effect of outsourcing peripheral and near-core tasks on firm performance was found, while firm strategy and environmental dynamism moderated the relationship (Gilley & Rasheed, 2000, abstract only). That is why this chapter rests the decision on the business's own grid and its own numbers rather than on a promise that buying will improve performance. A decision to buy is a decision about who does the work, what it costs on the same basis, what control is kept and what risk is taken, and it is not a claim that the business will do better because the work left.

The frontier between this chapter and its neighbors is set plainly because the subjects touch. The core and the non-core, the full cost of bought work and the grid of seven families of work of the case are Book I8's, cost structures and break-even analysis are Book G3's, the partners block of the business model is Book A1's, and structures, the network form included, are Book I2's, each cited here and never redone. The matter of Book J5 lies outside this book.

The table below sets the frontier row by row: for each neighbor, what the neighbor holds and what this chapter holds. Read it as a map of who decides what, so that a reader who wants the full cost of bought work, the cost structures or the partners block goes to the book that owns it, and a reader who wants the decision for one activity stays here.

Table 3.1 — Where this chapter stops and its neighbors begin

DEFINITION The words this book uses for an activity, total cost, a specific asset, hold-up and concurrent sourcing

These are this book's working words, and no source read defines them in this sense. An activity is one whole piece of the chain, such as assembly, tooling or installation, that can be made, bought or contracted. Total cost is the price of a thing together with the other costs of getting it, keeping it and changing it. A specific asset is an asset or a piece of know-how that is worth much more to one partner than to any other. Hold-up is the position of a business whose specific investment cannot be moved and whose partner can therefore ask for more. Concurrent sourcing is the making and the buying of the same thing at once. The core and the non-core, and the grid of seven families of work of the case, are Book I8's (Freelancers, Agencies and Outsourcing, chapter 2) and are not redefined here.

3.2 Total Cost: Beyond the Supplier's Price

The supplier's quote is one line of the cost of buying an activity, and a decision taken on that line alone is a decision taken on the smallest part of the answer. The quote is the price of the thing; the cost of buying it includes the oversight of the supplier, the freight, the cost of a defect, the cost of changing supplier and the hours of the business that the arrangement consumes. The cost of making the same thing includes the materials, the hours, the floor space, the equipment, the rework and the hours of oversight that the workshop consumes. The two are comparable only when both are counted on the same basis, and the comparison is fair only when the capacity the business does not have is counted on the making side.

The idea that a purchase has costs beyond its price is the subject of two readings. The abstract of one article reports that total cost of ownership looks beyond the price of a purchase to many other purchase-related costs, and that case studies of 11 firms show dollar-based and value-based models, used for supplier selection or supplier evaluation, with a strong relationship between model type and use, and barriers and benefits of the approach (Ellram, 1995, abstract only). The abstract of another reports that leading companies use total cost of ownership models, that a generic model is not appropriate, and that a model built on a core set of cost drivers present in most computations, with auxiliary drivers chosen for the purchase at hand, is suggested (Ferrin & Plank, 2002, abstract only). Neither abstract gives a typical share of hidden costs, and none is printed here. What the two support is the practice of counting the lines beyond the price and of building the model for the purchase at hand rather than from a template.

The comparison also has to be made on the same basis on both sides, and that is where a comparison can fail. A quote is a number a supplier has worked out, and the cost of making is often a number the business has never checked: the hours are estimated from memory, the floor space is treated as free because the lease is already paid, and the rework is left out because it is not on any invoice. The abstract of the study recalled from section 3.1 reports that comparative production costs were the strongest predictor in one division (Walker & Weber, 1984, abstract only); that is no rule, and it points to the discipline of counting both sides before comparing them.

The method and the numbers of the comparison are Chapter 7's, and the hidden costs of a transition are Chapter 4's. This section holds only the shape of the cost: the price of the thing, the oversight of the supplier, the freight, the cost of a defect, the cost of changing supplier and the hours of the business. A business that writes those lines down for both sides has a comparison it can defend; a business that compares one price with one price has a comparison that will be overturned by the first defect or the first late delivery.

WATCH OUT A supplier's price is one line of the cost, and the comparison fails when made and bought are not costed on the same basis

The trap is a quote set against a cost of making that leaves out the hours, the floor space and the rework of the business itself. The signs are easy to recognize: one price is compared with one price, and the made cost is a number nobody has checked. The remedy has three parts. Cost both sides on the same basis, so that the hours, the floor space, the equipment and the rework appear on the making side and the oversight, the freight, the defects and the change of supplier appear on the buying side. Count the capacity the business does not have, because a bench that is already full cannot absorb the work without holding something else back. And let a second person read the comparison before the decision is taken, because a comparison built by the person who wants a particular answer is a comparison that will find it. The reasoning is this book's own, and no figure is printed here.

3.3 Control: What You Can Direct and What You Can Only Ask For

Control is the second question of a make-or-buy decision, and it can be confused with ownership. A business can own an activity and still not direct it, and it can buy an activity and still direct the part that matters. The useful question is not who owns the activity but what the business can direct, what it can only ask for and what it can verify.

Two theories read for this book help to ask the question, and both are theories. The abstract of one reports that contractual rights are specific or residual; that when it is costly to list all specific rights over assets in a contract, it may be optimal to let one party purchase all residual rights, and ownership is the purchase of these residual rights; that residual rights bought by one party are lost by the other, which creates distortions; and that one firm should buy another when its control raises the productivity of its management more than the loss of control lowers the other's (Grossman & Hart, 1986, abstract only). The abstract of the other reports that a firm is identified with the assets its owners control; that the crucial difference between owning a firm and contracting for a service from another party who owns it is that under integration the owner can selectively stop working with the people of the firm, while without integration the owner can only stop dealing with the whole firm and its assets; and that partnerships and cooperatives emerge as special cases (Hart & Moore, 1990, abstract only). Neither is measured evidence; each holds what follows from its assumptions.

What the two theories suggest for a business is a way of sorting the question. What a business can direct is the work it performs itself, the test it writes and runs, the file it holds and the content it supplies to a partner. What a business can only ask for is a supplier's schedule, a supplier's method and the order in which a supplier serves its customers. What a business can verify is a record returned, a sample inspected and a test result reported. A decision that keeps the content and the test and buys the assembly is a decision that keeps control of what matters and gives up control of what does not.

The case, posed by this book, shows the sorting. The firmware and the alert logic are directed by the company and score 5 on control in the grid of section 3.8. The contract manufacturer flashes the image that the company supplies and runs the company's end-of-line test on the company's duplicate bench, so the company directs the content and asks for the schedule. That is a division of control that a written agreement can hold, and the agreement itself is Book H1's (Contracts and Commercial Agreements) and a lawyer's.

THE RULE A theory holds, it does not measure (Grossman & Hart, 1986)

The abstract of the article read for this book reports a theory of the costs and benefits of ownership in which contractual rights are specific or residual, ownership is the purchase of residual rights of control, and residual rights bought by one party are lost by the other (Grossman & Hart, 1986, abstract only). The rule this book draws from it is a rule about reading: a theory shows what follows from its assumptions, and no measured effect follows from it. The theory helps a business ask what control it needs over an activity, what it can direct, what it can only ask for and what it can verify, and it does not tell the business what to do. The answer is scored on the grid of section 3.8, where control is one criterion of six, and the verdict is written with its reason. No figure is printed here, and no source read measures the effect of ownership on the control a small business holds.

3.4 Capability: What You Do Better, and What You Could Learn

Capability is the third question, and it has two halves. The first half is what the business does better than the market today: the skill, the tooling and the knowledge that make its own work better than a supplier's. The second half is what the business could learn in time: the skill it does not have yet but could acquire, and the time that acquisition would take. Both halves belong in the answer, because a business that buys an activity it could learn in a season may be giving up a capability it will want in three years.

The readings on capability are careful about their settings. The abstract of one article reports that a transactions cost theory of vertical integration, refined to take into account industrial know-how and the cost of transferring it, was tested with probit techniques on data from the United States automobile industry, and that the results support the view that the development and deepening of human skills bear on efficient organizational boundaries (Monteverde & Teece, 1982, abstract only). The abstract of another reports that, comparing the capabilities explanation with the asset-specificity explanation on make-or-buy decisions of one large firm, asset specificity alone decides in some cases and capabilities or combinations in others, and that the similarity of knowledge bases and the time needed to acquire knowledge indicate when capabilities matter (Argyres, 1996, abstract only). The first is one industry and one period; the second is one firm. Neither is a rule for any other business.

Two further readings widen the question. The abstract of one reports that transaction costs and capabilities are intertwined in determining vertical scope, that capability differences are a necessary condition for vertical specialization, and that lower transaction costs lead to specialization only when capabilities along the chain are heterogeneous; the abstract reports four evolutionary mechanisms that reshape vertical scope over time, illustrated by mortgage banking in the United States, which disintegrated, and Swiss watch manufacturing, which integrated (Jacobides & Winter, 2005, abstract only). The abstract of another reports that in the luxury automobile segment, initial performance is associated with a decline as the level of vertical integration rises while performance improvement over the product lifecycle is associated with an increase, a trade-off between short-term performance and the evolution of capabilities (Novak & Stern, 2008, abstract only). The first is a theory of co-evolution illustrated by two industries; the second is an association in one segment of one industry, and it is written as an association.

A fifth reading is a conceptual model. The abstract of one article reports that transaction cost theory explains outsourcing as economizing, that the resource-based view examines specialized capabilities as a source of value in relationships between firms, and that the article extends both to explain the conditions leading to strategic outsourcing (Holcomb & Hitt, 2007, abstract only). It proposes conditions; it measures nothing.

The table below sets four of those readings beside their setting, what the abstract reports and what it does not support. Read it as a guard against carrying a finding out of the industry and the sample where it was made: each row is one study or one model, and the last column says what the row cannot be used for.

Table 3.2 — Four readings on capability, at the level read

WHO DOES THIS The person who does the work rates the capability, and the owner decides

The judgment of capability is made by different people at different sizes of business. A solo owner judges alone and asks an outside peer to read the judgment, because a person who does the work is the last to see that a supplier does it better. In a small team, the person who does the work rates the capability and the owner decides, so that the rating comes from the bench and the decision from the person who carries the consequence. In an established business, the operations lead and the quality lead rate it, and Chapter 9 reads who helps with this work and when to get help. In the case, posed by this book, Laurence Bergeron rates the capability of the company on engineering and testing, and the staff installer reports the hours of the workshop. No hire is made and no dollar is printed in this rating; the rating is a score from 1 to 5 on the grid of section 3.8.

3.5 Risk: Supply, Quality, Continuity and Knowledge

Risk is the fourth question, and this book reads four risks for an activity. Supply is whether the thing arrives: whether the supplier can deliver the volume, on the date, in the condition the business needs. Quality is whether the thing is right: whether the supplier's output meets the specification the business wrote, and whether the business can tell. Continuity is whether the supplier stays: whether the supplier will still be there in three years, at the same price, with the same people. Knowledge is whether the know-how leaves with the work: whether the business will still understand the activity after it has bought it for a few years.

The opening of one article read for this book names the risks of a supply chain. The article was read at its opening only, the opening the publisher shows, published October 15, 2004, and the exhibits are not shown; nothing is printed here from the parts not visible. At that level, the article establishes that supply chain risks fall into delays, disruptions, forecast inaccuracies, systems breakdowns, intellectual property breaches, procurement failures, inventory problems and capacity issues, which can hurt sales, raise costs or both; that a manufacturer with several suppliers kept producing after a plant fire while a single-sourced competitor was disrupted for months; and that no single strategy protects every supply chain, so the mitigation must match the risk (Chopra & Sodhi, 2004, opening only). The example is a manufacturer with several suppliers and a single-sourced competitor, described by kind and never named.

The case carries one of those risks in a printed figure. Book G8 (Business Insurance and Risk Financing) prints that a supplier's component shortage that stops kit assembly is rated at $40,000 and 30 days, reduced by a second source and a safety stock (Book G8, chapter 3). The second source is Chapter 5's subject, and the reduction is Book G8's figure, cited here and never recomputed.

WHAT CHANGED In the second quarter of 2026, 54.0 percent of manufacturing businesses expected a negative impact of United States tariffs (Statistics Canada, 2026)

The dated change comes from the statistics agency's release of May 27, 2026 on the Canadian Survey on Business Conditions, second quarter 2026, read on October 15, 2026. For that quarter, the release reports that 54.0 percent of manufacturing businesses expected a negative impact of United States tariffs, against 34.0 percent of all businesses, and that 48.1 percent of manufacturing businesses expected the cost of inputs to be an obstacle (Statistics Canada, 2026). The shares are the agency's, for the businesses surveyed in that quarter, and they will change with the next release. The release gives no figure on supply chain challenges and no figure on outsourcing, and none is printed here. The date is the source's date and never a date of the case.

3.6 Specific Assets, Uncertainty and the Hold-Up Problem

A specific asset is an asset or a piece of know-how that is worth much more to one partner than to any other, and it is the reason a make-or-buy decision can be hard to reverse. A mold cut for one product, a test bench built for one assembly, a line of code written for one gateway: each of those is worth what it cost only inside the relationship that uses it, and each is worth much less to anyone else. Once a business has made such an investment, the partner on the other side knows the investment cannot be moved, and the partner can ask for more. That is the hold-up problem.

The cost of using the price mechanism is the starting point, and it was read where the text is legible. The text read for this book establishes that the main reason why it is profitable to establish a firm would seem to be that there is a cost of using the price mechanism, the most obvious being the cost of discovering what the relevant prices are, then the costs of negotiating and concluding a separate contract for each exchange (Coase, 1937). The paper's later argument on the size of the firm is not printed, because the deposited text is illegible there.

Two further readings are theories. The abstract of one reports that the paper combines institutional economics, contract law and organization theory to distinguish three generic forms of economic organization, market, hybrid and hierarchy, by their coordinating and control mechanisms and their abilities to adapt to disturbances, and that the cost-effective form varies systematically with the attributes of transactions (Williamson, 1991, abstract only). The paper argues; it does not measure. The lecture read for this book establishes that the critical dimensions of a transaction are its complexity, the condition of asset specificity and the disturbances to which it is subject, and that incomplete contracts, bilateral dependency contingent on asset specificity and defection when a contract meets significant disturbances made vertical integration a paradigm of contract and organization (Williamson, 2009). It is a theoretical account.

The article on appropriable quasi rents was read at its opening only, the opening text deposited with its reprint, because the record of the journal has no abstract; nothing is printed here from the unread parts. At that level, the article explores one particular cost of using the market system, the possibility of postcontractual opportunistic behavior, and identifies its likely source in appropriable specialized quasi rents: once a specific investment is made and such quasi rents exist, the possibility of opportunistic behavior is very real (Klein et al., 1978, opening only). The industrial examples of the article are not printed.

Two studies complete the picture. The abstract of one reports that in the electronic components industry, integration of personal selling is associated with higher asset specificity and with the difficulty of evaluating performance, while neither the frequency of transactions nor the interaction of specificity and uncertainty is significantly related to integration (Anderson & Schmittlein, 1984, abstract only). The abstract of another reports that, in models that account for firm and transaction features, neither outsourcing nor internalization per se results in superior performance, which is contingent on the alignment between the governance decision and the degree of contractual hazards (Leiblein et al., 2002, abstract only).

The remedies for a hold-up are three, and Chapter 5 reads them in full. A business can keep the specific asset, so that the investment is its own and cannot be held. It can write the terms of the relationship, so that the price and the volume are fixed for a period. Or it can keep a second source, so that a partner who asks for more is not the only partner. The case, printed by Book A5 (Financing a Business), shows why the question is real: few buyers exist for a mold made for someone else's product (Book A5, chapter 3), which is why the molds of the second-generation enclosure score 5 on specificity in section 3.10.

The table below sets four of those readings beside their setting, what each reports and what it does not support. Read it as a guard against carrying a theory or a single industry's finding into a decision for another business: the first row is an article read at its opening only, the second is a conceptual analysis, the third is one industry, and the fourth is a set of models.

Table 3.3 — Four readings on specific assets, at the level read

QUÉBEC & CANADA NOTE The agreements that guard a specific asset are read in Québec and Canada through the books that hold the law

In Québec and Canada, the agreements that guard a specific asset are read through the books that hold the law and never through this chapter. Contracts, including the agreements that fix a price and a volume for a period, are Book H1's (Contracts and Commercial Agreements, and sections 4.4 and 5.2 of this book). Competition law, including the rules on exclusive dealing, tied selling and market restriction, is Book H4's (Consumer Protection, Advertising Law, Competition and Language Requirements, and section 5.4 of this book). Intellectual property, including the firmware and the trade secrets of a business, is Book H5's (Intellectual Property). A given agreement is a lawyer's. This chapter reads no text of law and rewrites none, and the book recommends no vendor, no supplier, no contract manufacturer, no molder, no adviser, no product and no tool.

3.7 Making and Buying the Same Thing at Once

A business does not have to choose between making and buying. It can do both at once, and the practice has a name: concurrent sourcing. The abstract of one article reports that a survey of small manufacturing firms compares when firms make, buy and concurrently source, that is, make and buy the same good at once; that support is found for aspects of transaction cost economics, neoclassical economics and the capabilities literature; and that concurrent sourcing is a distinctly different choice rather than a point on a make-or-buy continuum (Parmigiani, 2007, abstract only). The abstract reports no measured effect on performance, and none is printed here.

The reasons a business does both are practical. It keeps a skill alive, so that the know-how does not leave with the work and the business can still make the thing if the supplier fails. It keeps a second supplier ready, so that a partner who raises a price is not the only partner. It absorbs a peak, so that the busy season does not have to be bought at a premium or refused. And it keeps control of the urgent and the repaired, so that a customer who needs a kit tomorrow is served by the business itself. Each of those reasons is a reason to keep a part of the work, and none of them requires keeping all of it.

The cost of doing both is real and is worth stating. The volume of each side is smaller, so neither the workshop nor the supplier reaches the volume that would make its own side cheapest. Two routines must be managed, one for the workshop and one for the supplier, and each routine needs its own records, its own quality checks and its own review. And the business must decide which orders go to which side, because a rule that is not written will be decided by whoever is closest to the order.

The case, posed by this book, does both. The company buys 207 kits and keeps 23 in-house, for repairs, pilot builds and urgent kits. Chapter 2 reads that arrangement as a tapered stage, and Chapter 7 costs it. The arrangement is not a compromise between making and buying; it is a choice of its own.

The abstract of one study reports that on a panel of more than 3,500 product introductions in the global microcomputer industry, balancing vertical integration and strategic outsourcing, called taper integration, is associated with a richer product portfolio, product success and superior firm performance (Rothaermel et al., 2006, abstract only). That is an association in one industry, recalled from section 2.5, and it is not a cause and not a promise for any other business.

KEY CONCEPT Making and buying the same thing at once keeps a skill alive and a second supplier ready, and costs some of the volume of each

The idea of this section is this book's own reading and never a finding of a source. A business may keep a part of the work for repairs, pilot builds and urgent orders and buy the rest, and the arrangement serves four purposes at once: it keeps the skill alive inside the business, it keeps a second supplier ready, it absorbs a peak without buying all of it, and it keeps control of the urgent and the repaired. two routines and a smaller volume on each side, and the business must write which orders go to which side. No figure is printed here, and no source read measures what the arrangement does to performance.

3.8 Scoring Activities on One Grid

The decision for one activity is made on six criteria, and the criteria are scored on one grid so that two activities can be compared and a verdict can be defended. The six criteria are how central the activity is, the company's capability against the suppliers', how specific the assets and the knowledge are, the cost of making against buying, the risk of relying on the market, and the control the business needs. Each is scored from 1 to 5, where 5 favors making, and each carries a weight. The points of a criterion are its score times its weight divided by five, so that the total runs from 20 to 100. The rule is make at 75 or more, hybrid from 55 to 75, buy below 55.

The grid is this book's own and is posed. Two frameworks read for this book propose approaches to the same question, and neither tests this grid. The abstract of one article reports that neither transaction cost economics nor the resource-based view alone explains the complexities of outsourcing; that a prescriptive framework for outsourcing evaluation, developed from both theories and in-depth case studies, highlights performance management, operations strategy and process redesign; and that the theories should be applied with caution because their prescriptions sometimes contradict each other (McIvor, 2009, abstract only). The abstract of the other reports a framework for the decision on an individual part or a family of parts, developed from a review of the principal make-or-buy approaches and illustrated with case studies (Cánez et al., 2000, abstract only). Both propose; neither measures.

The weights are the part of the grid that a business must write for itself, and the reason is that the weights carry the judgment. A business whose product is its firmware will weight centrality and capability heavily. A business whose product is a commodity will weight cost heavily. The weights must sum to 100, and the reason for each weight should be written beside it, because a weight that nobody can explain is a weight that will be changed to produce a wanted answer.

The table below sets the six criteria beside their weights, what a score of 5 means for each and the question each criterion asks. Read it as the skeleton of the grid: the weights are this book's own and posed, a business writes its own, and the questions are the ones the grid is meant to make a business answer before it decides.

Table 3.4 — The six criteria of the grid, their weights and what a score of 5 means

METHOD Scoring an activity on one grid of six criteria in five steps

First, list the whole activities of the chain, one by one, so that each can be scored on its own. Second, weigh the six criteria so that the weights sum to 100 and write why each weight was chosen. Third, score each activity from 1 to 5 on each criterion, where 5 favors making. Fourth, turn the scores into points, each score times its weight divided by five, and read the total against the rule: make at 75 or more, hybrid from 55 to 75, buy below 55. Fifth, write the verdict, the reason, the person who disagrees and the date when it will be read again. The sequence is this book's own and is untested; no source tests it. No figure beyond the rule is printed here.

3.9 Writing the Decision So It Can Be Reviewed

A decision that is not written is made again by the next crisis. The verdict, the reason, the scores, the person who decides, the triggers that would bring the decision back to the table and the date of the next review belong in one record, and the record belongs in a register that a business keeps whatever tools it uses. The register is the memory of the boundary, and it is what lets a business read its own decisions a year later without relying on the recollection of the people who took them.

The record has six parts, and each has a purpose. The activity names what was decided, in the words the business uses for its own chain. The verdict says make, hybrid or buy. The reason says why, in a sentence or two, so that a reader who was not in the room can follow the argument. The scores give the grid's numbers, so that a later reader can see what the decision rested on and whether the scores have changed. The person who decides is named, so that the decision has an owner. The triggers say what would bring the decision back, and the date of the next review says when it will be read even if no trigger fires.

The triggers are the part that keeps the register alive. A trigger is an observable event that says the decision should be read again: a defect rate above a threshold for two months running, a price above a level, three late releases in a year. A trigger is not a wish and not a worry; it is a condition that a person can check, and it is written beside the decision so that the check happens without a meeting. The cadence that reads the register, the weekly release, the monthly scorecard, the quarterly review and the yearly review of every activity, is Chapter 8's.

The case, posed by this book, writes its verdicts and signs them. Laurence Bergeron and Karim Haddad sign the six verdicts together on day 28, and the triggers that would bring assembly back in-house are written at the third gate, told in Chapter 11. The signature is not a formality: it is the moment at which the decision becomes the company's rather than one person's, and it is what makes the register a record rather than a set of notes.

TOOLS OF THE TRADE What a business needs to keep a decision alive: a grid, a place for notes, a register and a calendar

Four functions carry a decision record, and a business should look for them rather than for a product. The first is a grid with weights that a person can change, so that the criteria and their weights can be written for the business rather than taken from a template. The second is a place to keep the scores and the notes of each activity, so that the reasoning sits beside the numbers. The third is a register of decisions with the reason, the owner, the triggers and the next review date, so that a reader can find the decision and see why it was taken. The fourth is a calendar of review dates, so that the register is read even when no trigger fires. Chapter 10 gives the functions in full. No brand name belongs on this list and no price; the requirement is the function.

3.10 The Case's Activity Grid and Its Verdicts

Hydrosentinelle inc. is a fictional company in Sherbrooke that designs and sells connected water-leak detectors for commercial and rental buildings, with two founders, Laurence Bergeron, the president, product and engineering, and Karim Haddad, the head of sales and operations (Book A5, chapter 1). The company writes its kit forecast on day 18, scores its six activities on day 24 and signs its verdicts on day 28, which is a condition of the first gate on day 30; the accountant reads the cost comparison at that gate, two hours at the printed illustrative rate of $120 an hour (Book G7, chapter 6). Every score, every total and every verdict in this section is posed by this book and illustrative.

The six activities are scored on the grid of section 3.8, with the weights of 25, 20, 20, 15, 10 and 10. Table 3.5 gives the scores in order, the totals and the verdicts: one make at 92, three hybrid at 68, 57 and 71, and two buy at 35 each.

The reasons are told in words and without a new figure. The firmware and the alert logic are what makes the product, they are assets of the company (Book H5, chapter 1), and they must be directed by the company. The molds are specific to the product, and the company is not a molder. The boards and the hosting are standard things that suppliers do better. The assembly scores low on cost once the capacity the company does not have is counted, which Chapter 7 reads. And installation is shared with the certified installers of Book F9 (Building and Running a Partner Program, chapter 4).

The cost comparison behind the assembly verdict is posed and told in full in Chapter 7: the contract manufacturer's quote, with freight and the company's oversight, is $202 a kit against $175.50 made in-house, so buying costs $26.50 more per kit. The verdict is hybrid anyway, because the bench needs 713 staff installer hours in the year and the staff installer has 600 free after installations, and the capacity the company does not have is the reason to buy a part of the work.

Figure 3.1 scores the case's six activities on one grid of six weighted criteria, against the thresholds of the grid's rule.

Figure 3.1 — One grid, six verdicts. One make, three hybrid and two buy verdicts; make at 75, a hybrid from 55 to 75 (posed).

The figure is read activity by activity. The firmware and the alert logic stand in the make zone, above the line at 75. The enclosure tooling, the kit assembly and the installation stand in the hybrid zone, between the lines at 55 and 75, and the boards and the hosting stand in the buy zone, below the line at 55.

The table below sets the six activities beside their scores in order, their totals and their verdicts. Read it as the case's grid in one place: the scores run in the order of the criteria, how central the activity is, the company's capability, how specific the assets and the knowledge are, the cost of making against buying, the risk of relying on the market and the control the business needs, and every score and total is posed by this book and illustrative.

Table 3.5 — The case's six activities, their scores and their verdicts, posed by this book

In La Machine, as of the application copy of September 7, 2026, the page that touches this chapter is CRM → Companies (business/marketing/companies), where a supplier, a molder or an installer can be recorded as a company with its sector, size, site, status and notes. It holds no contract, no price and no make-or-buy model; check it in your installation. The case does not use La Machine, and La Machine is described and never sold.

CASE IN POINT Hydrosentinelle inc. scores six activities on day 24 and signs six verdicts on day 28

Hydrosentinelle inc. is a fictional business created for this encyclopedia; its founders, its employees, its customers, its partners, its suppliers, its figures and the events described are invented, and any resemblance to a real business is coincidental. The company scores six activities on day 24 and signs six verdicts on day 28: one make, three hybrid and two buy. The firmware and the alert logic are made, at 92. The enclosure tooling for the second generation, the final kit assembly and end-of-line test, and installation in buildings are hybrid, at 68, 57 and 71. The electronic boards and the cloud hosting of the monitoring service are bought, at 35 each. Laurence Bergeron and Karim Haddad sign the six verdicts together on day 28, and the first gate on day 30 requires the verdicts written and signed. No result is credited to any verdict, and every score is posed by this book and illustrative.

DO THE MATH Two activities, six scores each: how the grid gives 92 and 68

Illustrative figures; the case is fictional. Step 1 — The six weights are 25, 20, 20, 15, 10 and 10, which sum to 100, posed. Step 2 — The firmware and the alert logic score 5, 5, 5, 3, 4 and 5. Step 3 — Each score times its weight divided by five gives 25, 20, 20, 9, 8 and 10 points. Step 4 — Those points sum to 92, which is above 75, so the verdict is make. Step 5 — The enclosure tooling scores 3, 2, 5, 3, 4 and 4, which give 15, 8, 20, 9, 8 and 8 points, a total of 68, between 55 and 75, so the verdict is hybrid, and every score is posed by this book and illustrative.

IN LA MACHINE La Machine can record a supplier as a company and computes no total cost and no make-or-buy verdict

In the application copy of September 7, 2026, CRM → Companies (business/marketing/companies) holds one record for each company, with its name, sector, size, site, status and notes, and it has no field for a supplier, a contract, a price, a tooling or a performance (La Machine, 2026). What does not exist is worth saying plainly: there is no supplier scorecard, no purchase order, no bill of materials, no total cost, break-even, present value or make-or-buy model, and no register of sourcing decisions and their triggers. The page computes no total cost and no make-or-buy verdict, and a business should check it in its own installation. The case does not use La Machine, and La Machine is described and never sold.

Costly Mistakes

• Deciding the whole chain at once. A verdict taken for the business as a whole averages a capability, a cost and a risk, and an average of three different questions is not a decision for any one activity.
• Comparing one price with one price. A quote set against a cost of making that leaves out the hours, the floor space and the rework is a comparison that will be overturned by the first defect or the first late delivery.
• Counting the cost of buying and forgetting the capacity of making. A bench that is already full cannot absorb the work without holding something else back, and the capacity the business does not have belongs on the making side of the comparison.
• Confusing ownership with control. A business can own an activity and still not direct it, and it can buy an activity and still direct the content and the test; the useful question is what it can direct, what it can only ask for and what it can verify.
• Buying an activity the business could learn in a season. A capability that would take one season to acquire is a capability the business may want in three years, and the time to learn belongs in the answer.
• Making a specific investment without a remedy for the hold-up. A mold, a bench or a line of code that is worth much more inside one relationship than outside it can be held to ransom, and the remedies are to keep the asset, to write the terms or to keep a second source.
• Leaving the decision unwritten. A verdict that lives in the memory of the people who took it will be taken again by the next crisis, and the reason, the scores, the owner and the triggers will be lost with it.

Key Takeaways

• A make-or-buy decision is taken one activity at a time, or one part or one family of parts, because each activity has its own volume, its own skill and its own risk, and a decision taken for the whole chain at once decides none of them well.
• The supplier's price is one line of the cost of buying; the comparison is fair only when made and bought are costed on the same basis, with the capacity the business does not have counted on the making side.
• Control, capability and risk are three separate questions: what a business can direct, what it can only ask for and what it can verify; what it does better today and what it could learn in time; and whether supply, quality, continuity and knowledge are safe.
• A specific asset is worth much more to one partner than to any other, and once such an investment is made the partner can ask for more; the remedies are to keep the asset, to write the terms or to keep a second source.
• A business can make and buy the same thing at once, which the abstract of the survey read for this book calls a distinctly different choice rather than a point on a make-or-buy continuum; the cost is two routines and a smaller volume on each side.
• The decision is scored on one grid of six weighted criteria, written with its reason, its owner, its triggers and its next review date, and signed; the grid, its weights and its thresholds are this book's own and posed, and no source read tests them.

Your Action Plan

1. List the whole activities of your chain and stop deciding them together. Write each activity, or each part or family of parts, on its own line, so that each can be scored and decided on its own. Deliverable: a written list of the activities of the chain. Verification: every line names one whole activity, and no line covers the business as a whole.
2. Count both sides of one activity on the same basis. For one activity, write the price of the thing, the oversight of the supplier, the freight, the cost of a defect, the cost of changing supplier and the hours of the business on the buying side, and the materials, the hours, the floor space, the equipment and the rework on the making side. Deliverable: a two-sided cost sheet for one activity. Verification: both sides carry the same categories, and the capacity the business does not have is counted on the making side.
3. Weigh the six criteria and score each activity. Write the six weights so that they sum to 100, write why each weight was chosen, and score each activity from 1 to 5 on each criterion, where 5 favors making. Deliverable: a grid with the weights, the scores and the totals. Verification: the weights sum to 100, every score has a reason, and each total is read against the rule of make at 75 or more, hybrid from 55 to 75 and buy below 55.
4. Write the decision record and sign it. For each activity, write the verdict, the reason, the scores, the person who decides, the triggers that would bring the decision back and the date of the next review, and have the person who decides sign it. Deliverable: a register of decisions with one record per activity. Verification: every record carries a verdict, a reason, an owner, at least one trigger and a next review date, and the owner has signed it.
5. Set the triggers and the review date, and read the register on the cadence. Write each trigger as an observable condition, put the review dates on a calendar, and read the register on the cadence of Chapter 8. Deliverable: a calendar of review dates with the triggers beside each decision. Verification: every trigger is a condition a person can check, and the first review has been held on its date.

In La Machine

In the application copy of September 7, 2026, the page that touches this chapter is CRM → Companies (business/marketing/companies), where a supplier, a molder or an installer can be recorded as a company with its sector, size, site, status and notes (La Machine, 2026). It is the only page of the platform that this chapter names, and a business should check it in its own installation.

What a business needs for this chapter lives elsewhere. La Machine holds no supplier scorecard, no purchase order, no bill of materials, no total cost, break-even, present value or make-or-buy model, and no register of sourcing decisions and their triggers, and a company record holds no contract, no price, no tooling and no performance. The grid, the two-sided cost sheet, the decision record and the calendar of review dates are kept by the business in a spreadsheet, a document or a tool it chooses by function, and this book is valid with any tool.

Self-Check

1. Why is the unit of a make-or-buy decision one activity, or one part or one family of parts, rather than the business as a whole?
2. What does the total cost of an activity include beyond the supplier's price, and why must made and bought be costed on the same basis?
3. What is the difference between what a business can direct, what it can only ask for and what it can verify?
4. What is a specific asset, what is the hold-up problem, and what are the three remedies?
5. What are the six criteria of the grid, and what is the rule that reads a total?

Answers. 1. Each activity has its own volume, its own skill and its own risk, so a decision taken for the whole chain at once averages a capability, a cost and a risk and fits none of them. 2. The total cost includes the price of the thing, the oversight of the supplier, the freight, the cost of a defect, the cost of changing supplier and the hours of the business, and the comparison is fair only when both sides carry the same categories and the capacity the business does not have is counted on the making side. 3. What a business can direct is the work it performs itself, the test it writes and runs, the file it holds and the content it supplies; what it can only ask for is a supplier's schedule, a supplier's method and the order in which a supplier serves its customers; and what it can verify is a record returned, a sample inspected and a test result reported. 4. A specific asset is an asset or a piece of know-how that is worth much more to one partner than to any other, and the hold-up problem is the position of a business whose specific investment cannot be moved and whose partner can therefore ask for more; the three remedies are to keep the specific asset, to write the terms of the relationship and to keep a second source. 5. The six criteria are how central the activity is, the company's capability against the suppliers', how specific the assets and the knowledge are, the cost of making against buying, the risk of relying on the market and the control the business needs, with the weights of this book's grid at 25, 20, 20, 15, 10 and 10; the rule is make at 75 or more, hybrid from 55 to 75 and buy below 55, and the grid, its weights and its thresholds are this book's own and posed, with no source read testing them.

Chapter References

Anderson, E., & Schmittlein, D. C. (1984). Integration of the sales force: An empirical examination. The RAND Journal of Economics, 15(3), 385–395. https://doi.org/10.2307/2555446

Argyres, N. (1996). Evidence on the role of firm capabilities in vertical integration decisions. Strategic Management Journal, 17(2), 129–150. https://doi.org/10.1002/(SICI)1097-0266(199602)17:2<129::AID-SMJ798>3.0.CO;2-H

Chopra, S., & Sodhi, M. S. (2004, Fall). Managing risk to avoid supply-chain breakdown. MIT Sloan Management Review. https://sloanreview.mit.edu/article/managing-risk-to-avoid-supplychain-breakdown/

Coase, R. H. (1937). The nature of the firm. Economica, 4(16), 386–405. https://doi.org/10.1111/j.1468-0335.1937.tb00002.x

Cánez, L. E., Platts, K. W., & Probert, D. R. (2000). Developing a framework for make-or-buy decisions. International Journal of Operations & Production Management, 20(11), 1313–1330. https://doi.org/10.1108/01443570010348271

Ellram, L. M. (1995). Total cost of ownership: An analysis approach for purchasing. International Journal of Physical Distribution & Logistics Management, 25(8), 4–23. https://doi.org/10.1108/09600039510099928

Ferrin, B. G., & Plank, R. E. (2002). Total cost of ownership models: An exploratory study. Journal of Supply Chain Management, 38(2), 18–29. https://doi.org/10.1111/j.1745-493X.2002.tb00132.x

Gilley, K. M., & Rasheed, A. (2000). Making more by doing less: An analysis of outsourcing and its effects on firm performance. Journal of Management, 26(4), 763–790. https://doi.org/10.1177/014920630002600408

Grossman, S. J., & Hart, O. D. (1986). The costs and benefits of ownership: A theory of vertical and lateral integration. Journal of Political Economy, 94(4), 691–719. https://doi.org/10.1086/261404

Hart, O., & Moore, J. (1990). Property rights and the nature of the firm. Journal of Political Economy, 98(6), 1119–1158. https://doi.org/10.1086/261729

Holcomb, T. R., & Hitt, M. A. (2007). Toward a model of strategic outsourcing. Journal of Operations Management, 25(2), 464–481. https://doi.org/10.1016/j.jom.2006.05.003

Jacobides, M. G., & Winter, S. G. (2005). The co-evolution of capabilities and transaction costs: Explaining the institutional structure of production. Strategic Management Journal, 26(5), 395–413. https://doi.org/10.1002/smj.460

Klein, B., Crawford, R. G., & Alchian, A. A. (1978). Vertical integration, appropriable rents, and the competitive contracting process. The Journal of Law and Economics, 21(2), 297–326. https://doi.org/10.1086/466922

La Machine. (2026). Documentation — route reference inventory. Internal product documentation, read October 16, 2026, together with the code of the pages this book describes: business/marketing/companies.

Leiblein, M. J., Reuer, J. J., & Dalsace, F. (2002). Do make or buy decisions matter? The influence of organizational governance on technological performance. Strategic Management Journal, 23(9), 817–833. https://doi.org/10.1002/smj.259

McIvor, R. (2009). How the transaction cost and resource-based theories of the firm inform outsourcing evaluation. Journal of Operations Management, 27(1), 45–63. https://doi.org/10.1016/j.jom.2008.03.004

Monteverde, K., & Teece, D. J. (1982). Supplier switching costs and vertical integration in the automobile industry. The Bell Journal of Economics, 13(1), 206–213. https://doi.org/10.2307/3003441

Novak, S., & Stern, S. (2008). How does outsourcing affect performance dynamics? Evidence from the automobile industry. Management Science, 54(12), 1963–1979. https://doi.org/10.1287/mnsc.1080.0922

Parmigiani, A. (2007). Why do firms both make and buy? An investigation of concurrent sourcing. Strategic Management Journal, 28(3), 285–311. https://doi.org/10.1002/smj.580

Rothaermel, F. T., Hitt, M. A., & Jobe, L. A. (2006). Balancing vertical integration and strategic outsourcing: Effects on product portfolio, product success, and firm performance. Strategic Management Journal, 27(11), 1033–1056. https://doi.org/10.1002/smj.559

Statistics Canada. (2026, May 27). Canadian Survey on Business Conditions, second quarter 2026. The Daily. Retrieved October 15, 2026, from https://www150.statcan.gc.ca/n1/daily-quotidien/260527/dq260527a-eng.htm

Walker, G., & Weber, D. (1984). A transaction cost approach to make-or-buy decisions. Administrative Science Quarterly, 29(3), 373–391. https://doi.org/10.2307/2393030

Williamson, O. E. (1991). Comparative economic organization: The analysis of discrete structural alternatives. Administrative Science Quarterly, 36(2), 269–296. https://doi.org/10.2307/2393356

Williamson, O. E. (2009, December 8). Transaction cost economics: The natural progression [Prize lecture]. The Nobel Foundation. https://www.nobelprize.org/uploads/2018/06/williamson_lecture.pdf

Chapter 4 — Outsourcing and Insourcing

Moving an activity out of a business, or back into it, is a project with a scope, an agreement, a pilot and a plan, so this chapter says what moves when an activity leaves, how to scope, select and contract, how pilot builds and a transition plan lower the risk, where the hidden costs sit, how quality and traceability are kept across two companies, when and how an activity comes back, and what happens to employees when work moves, reads the practice guidance of a Crown corporation as guidance and each study in its own setting, calls the law to the books that hold it, and tells the move of kit assembly of Hydrosentinelle inc. in ninety days, posed by this book.

What You'll Be Able to Do

• Describe what actually moves when a whole activity leaves a business — the work, the know-how, the records, the risk and the oversight — and what never moves, which is the duty of the business to its own customers.
• Scope an activity before any quote is sought, with a written specification, the volumes and their range, an acceptance test, the records to be returned and a schedule.
• Select a contract manufacturer or a service provider on several criteria and on one scope, comparing total cost rather than unit price.
• Settle the five parts of an agreement for a transfer — price, quality, capacity, confidentiality and exit — and call the law that governs each to the book that holds it.
• Run a pilot build and a transition plan, keeping the old way running until the new one has passed its acceptance, and reading the estimate again after the pilot.
• Recognize the signals that bring an activity back in-house, the costs of coming back, and the means a business keeps on hand so that the road back stays open.

Why It Matters

An activity that leaves a business does not leave empty-handed. It carries with it the work, the know-how that went with the work, the records that describe the work, the risk of a delay or a defect, and the oversight that the business now owes to whoever does it. A business that hands over a stage and thinks only about the unit price has handed over five things and priced one. The gap between those two numbers is where the surprises live, and the surprises arrive slowly: a first shipment that passes, a second that passes, and a third where a lot is held because a test that was never written down was never run.

The reason this matters for a small business is that the decision to move an activity can be taken under pressure. The bench is full, the season is coming, the owner has read that a specialist can do the work more cheaply, and the quote looks simple. None of that is unreasonable. What is missing is a project: a scope written before quotes are sought, so that two offers answer the same question; a selection on several criteria; an agreement that settles more than price; a pilot that tests the new way on a small number of units while the old way still runs; and a plan that says who does what on which day. A transfer run as a project is slower at the start and leaves fewer decisions to be made in a panic.

The third reason is that the move runs in both directions. Outsourcing and insourcing are the same decision seen from opposite ends, and a business that moves an activity out without keeping the means to bring it back has decided something it may not be able to decide again. Keeping the test protocol, a duplicate of the bench, the records and a small build cell is what makes the road back real rather than theoretical. The two directions from one core belong to Chapter 2, the grid to Chapter 3, the hybrid forms to Chapter 5, the digital layer to Chapter 6 and the numbers to Chapter 7.

In This Chapter

Section 4.1 says what moves when an activity leaves the business and opens the chapter's frontier table. Section 4.2 covers scoping and section 4.3 the selection of a provider on five criteria. Section 4.4 covers the agreement and its five parts, and section 4.5 pilot builds and the transition plan. Section 4.6 covers the hidden costs, section 4.7 quality and traceability across two companies, and section 4.8 insourcing. Section 4.9 covers employees, unions and the move of work, and section 4.10 tells the case's move of kit assembly in ninety days.

4.1 What Moves When an Activity Leaves the Business

When an activity leaves a business, the work itself is only the first thing to go. The know-how that went with the work goes too, because a supplier that makes the thing must be able to make it, and the knowledge of how it is made moves into the supplier's hands. The records that describe the work go, or should: the specification, the test, the drawing, the version of the firmware. The risk of a delay or a defect becomes a risk the business shares with a supplier it does not control. And the oversight that the business now owes arrives with all of it, because a business that has someone else make its product still answers to its own customers for that product. Five things move, and only one of them appears on a quote.

A page of a federal Crown corporation read on October 15, 2026, updated on October 4, 2022, and described here as that corporation's practice guidance and never as an obligation, sets out how that author frames the decision. The page states that a business can subcontract the production of a good, one manufacturing step, a service or storage; that a supplier may produce more cheaply, mostly as a matter of volume, so the decision should be revisited as the company grows; and that the risks to watch are dependence, the loss of in-house expertise and the sharing of inside information (Business Development Bank of Canada, 2022). Those are guidance rather than a rule, and the page gives no figure.

What does not move is the duty of the business to its own customers. A customer who buys a kit from the business has one supplier of record, and that supplier is the business, not the contract manufacturer behind it. This is why the oversight that a transfer creates is not optional: the business has moved the work and kept the promise. The five things that move and the one thing that does not are the shape of everything that follows in this chapter, and they are the reason a transfer is written as a project rather than approved as a price.

The table below names six neighbors, what each holds, and what this chapter holds in place of it. Read it row by row: each row is a boundary, and the boundary is a place where a business should stop and open the other book rather than improvise.

Table 4.1 — Where this chapter stops and its neighbors begin

The same page names steps: analyze operating costs first, define the work, prepare specifications and choose a supplier on several criteria and not price alone, start with a pilot and a transition plan, and fix one's own inefficiencies before handing work over (Business Development Bank of Canada, 2022). It also recommends being transparent with the subcontractor, in the page's own terms, by giving it a solid overview of the work ahead and sharing the right information. The page does not say to share forecasts or production data, and it does not say to write the specifications.

The last thing to say about what moves is that the records are the part that is expensive to rebuild. A specification that stays in a drawer at the company is not a scope for a supplier. The case's scope is built on three documents — a forecast, a protocol and an image — and each of them is the company's own, which is the subject of section 4.2.

DEFINITION The words this book uses for a transfer, a specification, acceptance, a pilot build and a transition plan

These are this book's working words, and no source read defines them in this sense. A transfer is the move of a whole activity from the business to a supplier, or back into the business. A specification is the written description of what is to be made or done, precise enough that two suppliers could quote on it. Acceptance is the test that decides whether a unit or a lot is good, written before the first unit is made. A pilot build is a small run made under the agreed method and inspected unit by unit before volume moves. A transition plan is the list of the steps of a transfer, each with a date and a person. The words are this book's own, and the case's figures are posed by this book. Scoping, contracts and managing external work are Book I8's (chapters 5 to 7), and this chapter cites them and does not redefine them.

4.2 Scoping the Activity: Specifications, Volumes and Acceptance

A scope is written before a supplier is chosen, and it holds five things. It holds the specification, which is the written description of what is to be made or done and precise enough that two suppliers could quote on it. It holds the volumes expected and their range, because a supplier prices a volume and a range, not a hope. It holds the acceptance test, which decides whether a unit or a lot is good. It holds the records to be returned, because the business will want to know what each unit's test result was. And it holds the schedule, which says when the first release is due and how often releases follow.

The company prepares the specification and the acceptance test, and it prepares them because they are its own know-how. A specification written by the supplier describes what the supplier already knows how to make; a specification written by the company describes what the company wants made, and it is the instrument that makes two quotes comparable. The Crown corporation page read for this book puts preparing specifications among its steps, and it puts it before the choice of a supplier (Business Development Bank of Canada, 2022). The same page advises analyzing operating costs first and fixing one's own inefficiencies before handing work over, which is the reason the specification is written by the business and not borrowed. The page is a practice and not an obligation, it is that author's own text read on the day of its entry, and no figure from it is printed here.

The case shows the shape of a scope, and every element of it is posed by this book. The kit forecast of 230 kits is written on day 18 of the plan by Karim Haddad, the head of sales and operations. The end-of-line test protocol is the company's own, written by Laurence Bergeron, the president, product and engineering. The firmware image that the contract manufacturer flashes is supplied by the company, so the supplier handles an image and never a design. Those three elements — a volume, a test and an image — are the smallest scope that lets two providers be compared, and each of them is an example of a scope element rather than a description of a real arrangement.

The volumes are two numbers and not one: the volume expected for the year and the range around it, because a supplier sizes a process and staffs a line for a volume and a peak. The year's number is 230 kits, posed by this book, and the range around it is the season. The page read for this book names the decision to subcontract as one to revisit as the company grows (Business Development Bank of Canada, 2022).

The acceptance test is the part of a scope that decides whether a unit is good. A test written before the first unit is made is a test that both sides agreed to; a test written after a defect appears is a negotiation. The test says what is measured, on which bench, against which criterion, and what happens to a unit that fails. The Crown corporation page names performance indicators on quality among what an agreement should carry, and this chapter treats the acceptance test as the thing those indicators are built on (Business Development Bank of Canada, 2022).

The records to be returned are the quiet element of a scope. A business that receives kits but not test results does not know whether its supplier's process is drifting, and it learns about drift through its customers instead. Asking for the test result of every unit is a small request that keeps the business's own records whole, and it is the element that section 4.7 builds on. The schedule closes the scope: it says when the first release is due, how often releases follow, and what notice a change of volume needs.

KEY CONCEPT Scoping comes before selecting, because two suppliers can only be compared on the same scope

This is the chapter's central idea for the sections that follow, and it is this book's own reading and never a finding of a source. A scope holds five things: the specification, the volumes and their range, the acceptance test, the records to be returned, and the schedule. Without a scope, each quote answers a different question, and a comparison between two quotes is a comparison between two different activities. The company prepares the specification and the acceptance test because they are its own know-how, and the Crown corporation's page read for this book puts preparing specifications before the choice of a supplier, as practice guidance and never as an obligation (Business Development Bank of Canada, 2022). The scoping of work and services belongs to Book I8 (chapter 5 of Book I8), and this chapter holds the scoping of a whole activity or a stage.

4.3 Selecting a Contract Manufacturer or Service Provider

A selection begins with one scope sent to more than one provider, because a quote is only a quote against something. Two offers that answer the same specification, the same volume and the same acceptance test can be compared; two offers that answer different questions cannot. The Crown corporation page read for this book advises choosing a supplier on several criteria and not on price alone, and it names no criterion (Business Development Bank of Canada, 2022). The criteria below are this book's own, said to be its own, and they are five.

The first criterion is capability: can the provider do this work at this quality? The way to check it is a sample and a first-article inspection, which is a full inspection of the first unit produced against the specification. A provider that cannot produce an acceptable first article will not produce acceptable volume, and finding that out on one unit is cheaper than finding it out on a release.

The second criterion is capacity: can the provider carry the volume and the peaks? The way to check it is to ask for a stated capacity and a schedule, and to read the two against the business's own peak.

The third criterion is quality: how does the provider find and correct a defect? The way to check it is to ask for its test records and to visit when the work matters, because a process for finding a defect is more informative than a promise that there will be none. A visit is not an audit in the sense of a certification scheme; it is a look at how the work is done, and what it shows is whether the test records are kept because they are used or because they were requested.

The fourth criterion is total cost: what does the thing cost with freight, oversight and rework? The way to check it is a quote set beside the company's own cost on one basis, which is the comparison Chapter 7 builds in full. A unit price compared with a fully loaded internal cost is not a comparison; a favorable quote can be a quote for something else, such as a different volume, a narrower specification or a test that the supplier expects the company to run.

The fifth criterion is continuity: will the provider be there in a year, and what happens if it fails? The way to check it is a second source and the terms of exit. Continuity decides what happens on the day a supplier stops, a risk that Book G8 (Business Insurance and Risk Financing) covers from the side of the exposures. A second source is also what reduced the shortage exposure that Book G8 rates at $40,000 and 30 days for the case, and that rating is cited as Book G8's and never recomputed here.

The table below sets the five criteria beside the question each one asks and the way to check it. Read it as a checklist for a selection: every row can be answered before the choice, and a row that cannot be answered is an open question rather than a reason to proceed.

Table 4.2 — Five criteria for comparing two offers on one scope

The case's selection is posed by this book. Two quotes for the final assembly and test of the kits are received by day 30, and receiving them is a condition of the first gate of the plan of Chapter 11. The contract manufacturer is the electronics supplier of Book A5, never named, and the second quote is from a provider never named. Karim Haddad collects the quotes and Laurence Bergeron assesses the technical side, and the assessment includes the sample and the first-article inspection rather than a reading of the quote alone. No hire and no dollar figure belongs to this selection; the cost comparison is Chapter 7's.

WHO DOES THIS The owner decides, the person who knows the work assesses the sample, and a second person reads the cost

The distribution below is this book's suggestion, and the case's part is posed by this book. A solo owner runs the selection alone and asks for a second opinion on the sample and the numbers, because a single reader of a quote is a reader who can be persuaded by it. In a small team the owner decides, and the person who knows the work assesses the sample and the first article, because the person who knows the work can see what a quote cannot show. In an established business a purchasing manager runs the selection, an operations lead assesses it and a finance lead reads the cost, and Chapter 9 develops those roles and the help around them. In the case, Karim Haddad collects the quotes and Laurence Bergeron assesses the technical side. No hire is made and no dollar figure is printed here.

4.4 The Agreement: Price, Quality, Capacity, Confidentiality and Exit

An agreement for a transfer settles five things, and the five are the parts a business should be able to find in the document it signs. It settles price and payment, quality and acceptance, capacity and schedule, confidentiality and ownership of what the company supplies, and exit and return. The Crown corporation page read for this book says that an agreement, signed by both parties, should carry performance indicators on quality, production capacity, stock levels and delivery times and confidentiality clauses (Business Development Bank of Canada, 2022). That is that author's guidance and not an obligation, and it maps onto four of the five parts below.

Price and payment settle the price of a unit or a lot, the terms of payment, and what a change of volume does to the price. The last item can be left out, and it is needed, because a price set for one volume is not a price for another, and the Crown corporation page notes that the decision should be revisited as the company grows (Business Development Bank of Canada, 2022). A clause that says what a second volume costs turns a future negotiation into a reading of the agreement.

Quality and acceptance settle the specification, the test, the indicator of quality and what happens to a rejected unit. This is the part of the agreement that turns the acceptance test of section 4.2 into something both sides have agreed to, and it is where the performance indicators the page names on quality belong. Quality management as a discipline belongs to Book J4 (Quality Management and Continuous Improvement), and this chapter cites it rather than restating it.

Capacity and schedule settle the volume the supplier holds for the business, the lead time, and the indicators of capacity and delivery the Crown corporation page names (Business Development Bank of Canada, 2022). Confidentiality and ownership settle what the supplier may not share and what stays the company's property, such as the firmware image, the test protocol and the records; the firmware as an asset belongs to Book H5 (Intellectual Property), chapters 1 and 5 of that book, and this chapter cites it rather than redefining it. Exit and return settle how either side ends the agreement, what the company pays on resiliation and how the work returns.

The law that governs the agreement is called, never rewritten. The Civil Code of Québec, LégisQuébec, up to date to August 12, 2026, read on October 15, 2026, provides at article 1708 that a sale transfers ownership of property for a price in money, at article 1726 that the seller warrants the property free of latent defects, and at article 1729 that in a sale by a professional seller a defect is presumed to have existed at the sale if the property malfunctions or deteriorates prematurely compared with property of the same type; at article 2098 that a contract of enterprise or for services is one by which a contractor or provider undertakes to carry out physical or intellectual work or to supply a service for a price; at article 2103 that the contract is a sale, not a contract of enterprise, where the work is merely accessory to the value of the property supplied; at article 2125 that the client may unilaterally resiliate the contract even while the work is in progress; at article 2126 that the contractor may resiliate only for a serious reason and never at an inopportune moment; and at article 2129 that on resiliation the client pays, in proportion to the agreed price, the actual costs, the value of the work performed and of the property supplied that can be used. Those provisions are called to Book H1 (Contracts and Commercial Agreements) and to a lawyer, and none of them is applied to a real agreement here. Only a lawyer reads the line between a contract of enterprise and a sale for a given agreement.

The table below sets the five parts of an agreement for a transfer beside what each part settles in this book's words and where it comes from. Read the last column as a map: two parts rest on this book's own reasoning together with the contract book, two rest on the Crown corporation's page of 2022 with a neighbor, and the exit rests on the articles called above.

Table 4.3 — The five parts of an agreement for a transfer and where each comes from

The case's agreement is posed by this book. The supply and quality agreement is signed on day 57 after the outside lawyer's reading, which is 4 hours in days 31 to 60 and 1 hour in days 61 to 90 of the plan of Chapter 11; the outside lawyer's illustrative rate is $300 an hour (Book F9). The agreement is signed after the two pilot builds of section 4.5 have been accepted, which is the order this chapter recommends: the test first, the signature second. A signature that precedes the first acceptable unit is a signature that buys an intention.

QUÉBEC & CANADA NOTE Contracts of enterprise and of sale are held by Book H1, and this book explains the law as published and is not legal advice

This book explains the law as published and is not legal advice. In Québec, an agreement that has someone make or do something for a business may be a contract of enterprise or for services or a sale, and the Civil Code of Québec, LégisQuébec, up to date to August 12, 2026, read on October 15, 2026, holds the relevant rules: the sale and the seller's warranty (articles 1708, 1726 and 1729); the contract of enterprise or for services, in which the contractor chooses the means and acts in the client's best interests, and its line with a sale (articles 2098, 2099, 2100 and 2103); and the client's unilateral resiliation, the contractor's resiliation only for a serious reason, and the price of resiliation (articles 2125, 2126 and 2129). Book H1 holds contracts, service agreements and supplier agreements, and a lawyer reads a given agreement. This book recommends no vendor, no supplier and no contract manufacturer, and it is not legal advice.

4.5 Pilot Builds and the Transition Plan

A pilot build is a small run made under the agreed method and inspected unit by unit before volume moves. It is the step that puts the specification and the acceptance test to work on real units, and it is cheap by design: a handful of units made under the agreement's method, inspected one by one, before a release of any size depends on the supplier's process. The Crown corporation page read for this book names a pilot and a transition plan among its steps (Business Development Bank of Canada, 2022), and it gives no size and no duration for a pilot, which is why this chapter sets none either.

The transition plan is the list of the steps of a transfer, each with a date and a person. It is written before the first step is taken, and it names who hands over the specification, who delivers the test equipment, who runs the first inspection and who signs the acceptance. A transition plan that names a team and no person is a plan that will be read and not run. The Crown corporation page names a transition plan as a step and does not describe its contents (Business Development Bank of Canada, 2022); the contents described here are this book's own.

The old way keeps running until the new one has passed its acceptance. This is the rule that makes a pilot useful, and it is also the rule that a business under pressure is tempted to break. If the bench is dismantled the day the agreement is signed, the pilot's failure is a stoppage rather than a lesson. Keeping the old way running costs hours and space for a few weeks, and it is what allows the first release to be a test rather than a bet. The hours belong in the plan, which is the point of section 4.6.

The first pilot and the second pilot answer two different questions. A first pilot of a handful of units answers whether the supplier can make the thing at all; a second pilot of a larger handful answers whether the supplier can make it repeatably, which is the question that volume will ask. The case runs both, on 10 kits and then on 25, and the second is accepted three days before the agreement is signed.

The case shows the sequence, all posed by this book. The duplicate test bench is delivered to the contract manufacturer on day 41, so the supplier tests on the company's bench and not on its own. A pilot build of 10 kits is accepted on day 47, and a pilot build of 25 kits is accepted on day 54, each after a first-article inspection by Laurence Bergeron. The supply and quality agreement is signed on day 57, three days after the second pilot build, and the first release of 25 supplier-built kits is delivered on day 84. The order of those days is the method: equipment, small pilot, larger pilot, signature, first release.

What the pilot produces besides an answer is a record. The inspection of each unit in the pilot writes down what was measured and what was found, and those records become the baseline against which the supplier's later releases are read. A pilot that produces units and no records has tested the supplier's process and not the company's test. The case's pilot records are the company's, kept with the unit's identity, and they are the first entries of the flow of records that section 4.7 describes.

METHOD Moving an activity to a supplier in five steps

First, write the transition plan with a date and a person for each step, because a plan that names a team and no person will be read and not run. Second, hand over the specification, the acceptance test and the firmware image, so that the supplier works from the company's documents and not from its own recollection. Third, run a pilot build and inspect every unit, because a defect found on a handful of units is a lesson and the same defect found on a release is a stoppage. Fourth, accept the pilot against the written test before volume moves, so that acceptance is a decision and not a gradual drift into production. Fifth, keep the old way running until the first release has passed acceptance, because the means to keep producing is the only insurance a transfer has. The five steps are this book's own sequence, and no source read tests it.

4.6 Hidden Costs: Searching, Transferring, Monitoring and Switching

The quote covers a unit, and the transfer costs more than the unit. This section is about the difference, and it draws on two readings that are each bounded in a way worth stating before the list. The first is an article read at its opening only, the opening the publisher shows, published on April 15, 2001, with the rest behind a subscription; the chapter says so once and prints nothing from the parts not visible, and it lists no hidden cost taken from the article. At that level, the study combined a literature analysis with 50 cases of information technology outsourcing, with data gathered in 1998, and about 14 percent of the operations were deemed failures; benefits of scale and expertise are often eroded by costs managers fail to identify, although companies said they believed they understood the major costs (Barthélemy, 2001, read at its opening only). That is one study of one kind of outsourcing over one period, and the share belongs to that sample and is never a rate for a reader.

The second reading is an abstract-only study of services offshoring. The abstract reports that hidden costs are the costs of implementation neglected in strategic decisions; that in services offshoring, decision makers are more likely to make cost-estimation errors as configuration and task complexity rise; and that experience and a strong orientation toward organizational design reduce those errors (Larsen et al., 2013, abstract only). The abstract gives no size of the error and no finding on manufacturing, and neither is printed here. What the two readings have in common is a single idea: the costs that decide a transfer are not all in the quote, and the ones that are missing are the ones the business itself controls.

The four places where cost hides are this book's own list, and they are not the list of any source. The first is searching: the hours spent finding providers, comparing quotes and checking references, which are hours that no quote contains and that fall before the first unit is made. The second is transferring: the hours and the equipment spent moving the know-how, such as writing the specification, building and delivering a duplicate bench, running the pilot builds and training the supplier's people. The third is monitoring: the oversight owed to the supplier every week, which continues for as long as the transfer lasts and which section 4.7 and the routine of Chapter 8 make concrete. The fourth is switching: the cost of changing supplier or of bringing the work back, which is the price of undoing the decision and which is smallest when the road back has been kept open, as section 4.8 describes.

Each of the four has a place in the plan. Searching and transferring fall in the first slice, because they happen before volume moves. Monitoring is a recurring line and belongs to the routine rather than to the plan, since it continues every week for as long as the transfer lasts. Switching is a contingency that is not spent unless it is needed, and the means it requires — a protocol, a bench, a record and a cell — are bought once and kept.

WATCH OUT A quote covers the unit and forgets the places where cost hides when work moves

The trap is a transition that is priced on the unit and forgotten everywhere else. The quote covers the unit, and the business forgets searching, transferring, monitoring and switching, so the estimate that justified the move is an estimate of one line of a larger account. The signs are a plan with no hours for the transfer and no hours for oversight, a transition whose steps have no dates, and an estimate that was never read again after the pilot. The remedy is to put hours in the plan for each of the four places and to read the estimate again after the pilot, when the hours actually spent on the transfer are known. The four places are this book's own list, the article read at its opening gives none, and no saving, premium or failure rate of outsourcing appears here.

4.7 Quality and Traceability Across Two Companies

Quality across two companies rests on three things, and none of them is a certificate. The first is a written specification and a written acceptance test, which is the scope of section 4.2 doing its work at a distance. The second is a first-article inspection, which is a full inspection of the first unit against the specification, and which the case runs on every pilot build. The third is a flow of records back to the owner of the product, because the company that sold the kit is the company that must know how the kit was made and how it tested. A certificate says that a system was audited on a day; these three things say what happened to a unit.

A study of one industry is worth reading here for its setting as much as for its result. The abstract reports that in 30 pairs of regulated drug plants in the United States mainland and Puerto Rico, matched by parent firm and product, the Puerto Rican plants operated with a significantly higher quality risk on average, and that the transfer and maintenance of the knowledge needed to operate with low quality risk are left as the most plausible explanation (Gray et al., 2011, abstract only). The plants belong to the same firms, which is the fact that limits the reading: the study compares plants of one owner and says nothing on contract manufacturers, and it is one industry over one period. What it points to is the transfer of knowledge as the plausible explanation of a quality risk, and it is an association in its sample and not a cause.

The implication for a transfer is not that distance creates defects; it is that knowledge is the thing that must cross. A specification travels as a document and a test as a protocol, and the reason a measurement is taken travels only if somebody explains it. This is why the pilot builds of section 4.5 inspect every unit rather than a sample, and why the company runs its own end-of-line test rather than accepting a supplier's statement.

Traceability is the part of quality that answers a question after the fact. A unit is traceable when its serial number, its firmware version and its test result can be found, so that a question about one unit can be answered without inspecting a warehouse. The records of the case are the subject of Chapter 6, and this chapter says only what the transfer requires: that the supplier return the test result of each unit, that the company keep those results with the unit's identity, and that the end-of-line test run on the company's own bench rather than on a bench the supplier owns.

The case does both, all posed by this book. Each pilot build is accepted after a first-article inspection by Laurence Bergeron, and the end-of-line test of every supplier-built kit runs on the duplicate bench that the company owns and delivered on day 41, so that the result is a company record rather than a supplier statement. The device records come back every day.

THE RULE A study is printed in its setting, and plants of the same firms say nothing on contract manufacturers (Gray et al., 2011)

The rule this book applies is that a study is printed in its setting. The abstract reports that in 30 pairs of regulated drug plants in the United States mainland and Puerto Rico, matched by parent firm and product, the Puerto Rican plants operated with a significantly higher quality risk on average, and that the transfer and maintenance of the knowledge needed to operate with low quality risk are left as the most plausible explanation (Gray et al., 2011, abstract only). Two limits are said with the result. The first is that the plants belong to the same firms, so the study compares plants of one owner and says nothing on contract manufacturers. The second is that it is one industry over one period. What the study points to is the transfer of knowledge as a plausible explanation of a quality risk, and it is an association in its sample and never a cause, a rate or a forecast.

4.8 Insourcing: Signals, Costs and the Road Back

Work comes back for reasons that can be read in advance, and the signals are few. The first is a defect rate that stays above the line the business set, because a quality problem that repeats is a process problem and not a bad lot. The second is a price that passes the line, because a supplier's price can rise with its own costs and the business's comparison can change without anything else changing. The third is late releases that repeat, because a delivery problem that recurs is a capacity problem at the supplier. Each signal is a threshold, and a threshold written down before the event is a decision; a threshold discovered during the event is a quarrel. Writing the three thresholds is what the third gate of the plan of Chapter 11 requires.

A study of one kind of outsourcing is worth reading for the shape of the decision. The abstract reports that in a sample of 160 information technology managers involved with application development, product quality, service quality, relationship quality and switching costs are related to the decision to bring outsourced work back in-house, while firms that switched vendors reported high product and service quality but low relationship quality and switching costs (Whitten & Leidner, 2006, abstract only). That is one kind of outsourcing and one sample, and its findings are associations in that sample, never a rule for manufacturing or for a reader. What it suggests is that the decision to come back is associated with more than a defect count, and the price of switching is one of the things associated with it.

The costs of coming back are the mirror of the costs of going out. Coming back needs a bench, the people to run it, the time to reconstitute a process the business no longer runs, and the price of ending the agreement, which in Québec is set by articles 2125, 2126 and 2129 of the Civil Code of Québec as called to Book H1 and to a lawyer in section 4.4. A resiliation is not free and its price is a matter of the agreement and of the law, not of goodwill. The means to keep the road open are decided before the move, not after it: the test protocol stays the company's, the duplicate bench stays the company's, the records come back, and a small build cell keeps the muscle.

A survey of small manufacturing firms compared making, buying and doing both at once, and the abstract reports that concurrent sourcing is a distinctly different choice rather than a point on a continuum (Parmigiani, 2007, abstract only). That is the form the case keeps, and it is a survey of small manufacturing firms rather than a rule for any business.

The case's triggers are written at the third gate, all posed by this book. Assembly comes back in-house if the end-of-line defect rate exceeds 2 percent for 2 months running, if the price per kit exceeds $215, or if 3 releases in a year arrive late. The company keeps a build cell for 23 kits a year, which is the concurrent form of the survey above and the means of the road back. No trigger is reached on day 365, when the case has shipped 236 kits with a first-pass yield of 98.6 percent; every count in that sentence is posed by this book and illustrative, and no result is credited to the move.

TOOLS OF THE TRADE What a business needs to keep the road back open: a protocol record, test results, a trigger calendar and one register

Five functions keep the road back open, and a business should look for them by function rather than by product. The first is a record of the test protocol and its versions, so that the company's test is the company's own document even while a supplier runs it. The second is a record of each unit's test result, so that a question about one kit can be answered long after it shipped. The third is a calendar of the dates when a trigger is read, so that a threshold is looked at on a day rather than remembered in a crisis. The fourth is a list of the means kept to bring a stage back: the bench, the cell, the protocol and the trained person. The fifth is one register of decisions with the triggers of each, so that the reasons are written where they can be read again. Chapter 10 gives the functions in full. No brand name and no price belongs on this list, and the requirement is the function.

4.9 Employees, Unions and the Move of Work

Moving work can touch the people who do it. A business that transfers a stage may change what its own employees do, may reduce the hours of a role, or may move people to other work, and the rules that govern those changes are not this chapter's. The Labour Code, CQLR c C-27, s. 45, LégisQuébec, up to date to August 12, 2026, read on October 15, 2026, provides that the alienation or operation by another, in whole or in part, of an undertaking does not invalidate a certification or a collective agreement, and that the new employer is bound by them; its second paragraph provides that the first does not apply to the transfer of part of the operation where the transfer does not entail, besides functions or the right to operate, most of the elements that characterize the part of the undertaking involved. That provision is called to Book H7 (Employment Law Essentials (Québec and Canada)), and no conclusion is drawn from it for any real transfer here.

Whether a union is present, what a collective agreement says, and what the text means for a given transfer are questions for Book H7 and for a lawyer. A business that moves work and changes its employees' duties should treat the question as a legal one from the start rather than at the point where a dispute appears, because the answer can change the plan's dates and not only its paperwork. This chapter says what a move involves so that the legal question can be asked early, and it decides nothing about it.

The people side of a transfer is larger than the legal question. The person who ran the bench that leaves may be the person who holds knowledge that the specification does not capture, and that knowledge is worth an hour of writing down before the bench goes quiet. Moving those hours into the specification and the test protocol is how a transfer keeps its in-house expertise instead of losing it, which is one of the risks the Crown corporation page names (Business Development Bank of Canada, 2022). The other risk it names, the sharing of inside information, is the reason the agreement of section 4.4 carries a confidentiality part.

One further limit belongs here. The certification of a radio device built by a third party is not read for this book, and no rule on it is printed; a business that has such a device made should ask the certification body and a qualified professional.

The case's situation is simple and posed by this book. The case has no certified union and makes no acquisition, so no question of certification is raised for it, and nobody is hired or dismissed in the move. The hours of the staff installer that the transfer frees return to installations, which is told in Chapter 7. The build cell that remains keeps 23 kits a year of work in the workshop, so the transfer does not empty the bench and the workshop keeps a task.

WHAT CHANGED The Labour Code is read up to date to August 12, 2026, and a later amendment would change what is printed (Labour Code, s. 45)

The dated state of the rule is its source's. The Labour Code, CQLR c C-27, s. 45, was read in LégisQuébec on October 15, 2026, up to date to August 12, 2026, and an amendment after that day would change what the section says; Book H7 holds the current text of the employment rules and of the move of an undertaking. The move of an undertaking and the collective agreement are Book H7's, and this chapter only calls them. No real transfer is judged here, and nothing here says whether any agreement, clause or move complies with the rule. The case has no certified union and makes no acquisition, so the question of certification is not raised for it, and the case's own facts in this section are posed by this book.

4.10 The Case's Move of Kit Assembly

The move of kit assembly at Hydrosentinelle inc. is told in this section, and every figure of it is posed by this book and illustrative. Hydrosentinelle inc. is a fictional business created for this encyclopedia. What moves out is the final assembly of the kits, the flashing of the firmware image that the company supplies, the end-of-line test of 207 kits a year on a duplicate test bench the company owns, and the packing. What stays in-house is the firmware image and the alert logic, the end-of-line test protocol, the device records sent back every day, and a build cell for 23 kits a year for repairs, pilot builds and urgent kits. The duplicate test bench is $9,200, posed.

The split is the whole decision in one sentence: the company buys the assembly and the bench time, and it keeps the image, the test, the records and a cell. The firmware and the alert logic stay in-house because the grid of Chapter 3 scores them 92 and the verdict is to make them, and no figure of that grid is recomputed here. The contract manufacturer is the electronics supplier of Book A5, never named, and the molder and the second source of the sensor component are never named either. The year's forecast is 230 kits, of which 207 are bought and 23 are kept in-house, which is the concurrent form of section 4.8 and the reason the build cell exists.

The plan of the move is written in days, because the case is counted in days and never in calendar years or months. The table below gives the seven steps of the move with the day and the person for each. Every day and every figure in it is posed by this book; the hours behind the plan are valued at the illustrative rates printed by Books F9 and G7, and no calendar date belongs to any row.

Table 4.4 — The case's move of kit assembly: seven steps, posed by this book

Figure 4.1 follows the move of kit assembly to the contract manufacturer day by day, and shows what stays in-house.

Figure 4.1 — Moving kit assembly in ninety days. Quotes, a duplicate test bench, two pilot builds, a signed agreement and a first release of 25 kits by day 84; the firmware image, the test protocol, the device records and a small build cell stay in-house (posed).

Read the line of the figure step by step and it tells the chapter's method in seven steps. The quotes on day 30 give two offers on one scope, which is section 4.3. The duplicate test bench on day 41 puts the company's acceptance test at the supplier, which is sections 4.2 and 4.7. The pilot builds on days 47 and 54 test the method on 10 units and then on 25, which is section 4.5. The agreement on day 57 comes after the pilots, not before them, and the outside lawyer's reading is the hours of section 4.4. The first release of 25 kits on day 84 is the first volume that depends on the supplier's process, and the scorecard on day 88 is the routine that keeps reading it, which is Chapter 8.

The plan of Chapter 11 carries this move across three slices with gates on days 30, 60 and 90, and a gate that is missed is read again 7 days later, with nothing new started until it passes. The second gate's conditions include the duplicate bench delivered on day 41, the two pilot builds accepted on days 47 and 54, the agreement signed on day 57 and the tooling agreement signed with the second-generation molds ordered on day 58. The third gate's conditions include the first release of 25 supplier-built kits on day 84, the scorecard live on day 88, and the triggers of section 4.8 written beside the decision. The routine that the scorecard begins is 89 hours a year in the case, posed by this book, and the ninety-day plan of Chapter 11 is 121 hours in all.

In La Machine, as of the application copy of September 7, 2026, the pages that touch this chapter are Control center → Plan tasks (/taches), where a transition step can carry a "Done when" test that the Mark as done button does not verify; Control center → Projects (chains) (/projets), which shows plan projects read only; and Control center → Audit log (/journal), which traces what happened to a record, never why. It holds no purchase order, no supplier record or scorecard and no traceability of lots; check it in your installation. The case does not use La Machine.

CASE IN POINT Hydrosentinelle inc. moves kit assembly in ninety days and keeps the firmware, the test and the records

Hydrosentinelle inc. is a fictional business created for this encyclopedia; its founders, its employees, its customers, its partners, its suppliers, its figures and the events described are invented, and any resemblance to a real business is coincidental. Karim Haddad, the head of sales and operations, receives two quotes for the final assembly and test of the kits on day 30. Laurence Bergeron, the president, product and engineering, delivers the duplicate test bench to the contract manufacturer on day 41 and accepts a pilot build of 10 kits on day 47 and a pilot build of 25 kits on day 54. The supply and quality agreement is signed on day 57 after the outside lawyer's reading, and Karim Haddad receives the first release of 25 supplier-built kits on day 84 and starts the monthly scorecard on day 88. What stays in-house is the firmware image and the alert logic, the end-of-line test protocol, the device records sent back every day, and a build cell for 23 kits a year. Every day and every figure is posed by this book, no result is credited to the move, and the company has no union and makes no acquisition.

DO THE MATH Seven steps, ninety days: the transfer of the case in days and kits

Illustrative figures; the case is fictional. Step 1 — The two pilot builds of 10 kits and of 25 kits are 35 kits built before the first release. Step 2 — The pilot builds are accepted on days 47 and 54, which are 7 days apart. Step 3 — The agreement is signed on day 57 and the first release of 25 supplier-built kits arrives on day 84, which is 27 days after the signature. Step 4 — The scorecard goes live on day 88, 4 days after the first release arrives. Step 5 — The year's forecast of 230 kits is 207 bought and 23 kept in-house, and every count and every day in these steps is posed by this book and illustrative. No step is a rate, a benchmark or a forecast for any reader.

IN LA MACHINE La Machine can hold a transition step as a task with a 'Done when' test, and it traces no lot and no supplier

In the application copy of September 7, 2026, read in its code on October 15, 2026, Control center → Plan tasks (/taches) holds recurring or one-off tasks with a "Done when" criterion required when the task is created, six statuses and a next due date, and the Mark as done button does not check that criterion; a transition step can live there as a task whose test a second person reads (La Machine, 2026). Control center → Projects (chains) (/projets) shows plan projects as chains of tasks, in read-only mode, and no page creates one. Control center → Audit log (/journal) traces what happened to a record, never why, and it is never a register of lots. There is no supplier record, no supplier scorecard, no purchase order, no bill of materials and no traceability of lots or serial numbers. Check it in your installation; the case does not use La Machine, and the platform is described and never sold.

Costly Mistakes

• Handing over the work and pricing only the unit. The work, the know-how, the records, the risk and the oversight all move, and a quote covers one of the five.
• Seeking quotes before writing the scope. Two offers that answer different questions cannot be compared, and the cheaper of them will look like the better one.
• Choosing on price alone and calling it a saving. A unit price compared with a fully loaded internal cost is not a comparison, and the criteria of capability, capacity, quality and continuity are where a selection is decided.
• Signing the agreement before the pilot builds. A test run after the signature is a negotiation and a test run before it is a decision; the signature belongs after the second pilot accepts.
• Dismantling the old way on the day the agreement is signed. Keeping the old way running until the first release has passed acceptance is the only insurance a transfer has.
• Forgetting the hours of searching, transferring, monitoring and switching. The quote covers the unit, and a plan with no hours for the transfer and no hours for oversight will be read again after the pilot with a different answer.
• Moving an activity out without keeping the means to bring it back. The test protocol, the duplicate bench, the records and a small build cell are what make the road back real rather than theoretical.

Key Takeaways

• When an activity leaves a business, five things move — the work, the know-how, the records, the risk of a delay or a defect, and the oversight the business now owes — and what never moves is the duty of the business to its own customers.
• Scoping comes before selecting: a scope holds the specification, the volumes and their range, the acceptance test, the records to be returned and the schedule, and the company prepares the specification and the test because they are its own know-how.
• A provider is compared on capability, capacity, quality, total cost and continuity, on one scope, and the choice is made on several criteria rather than on price alone.
• An agreement for a transfer settles five parts — price and payment, quality and acceptance, capacity and schedule, confidentiality and ownership, and exit and return — and in Québec the rules on the contract of enterprise or for services, the sale and the client's unilateral resiliation are held by Book H1 and read by a lawyer.
• Pilot builds and a transition plan lower the risk of a transfer: the plan puts a date and a person on every step, the pilots test the method on a few units before volume moves, and the old way keeps running until the first release has passed its acceptance.
• The road back is kept open before it is needed, by a test protocol that stays the company's, a duplicate bench the company owns, records that come back and a small build cell, and the signals that bring work home are a defect rate above the line, a price above the line, or late releases that repeat.

Your Action Plan

1. Write the scope before you ask for a quote. Set down the specification, the volumes and their range, the acceptance test, the records to be returned and the schedule, and keep the specification and the test in the company's own name. Deliverable: a written scope of one activity, five elements included. Verification: two providers could quote on the document without asking a question that the document should answer.
2. Send one scope to at least two providers and compare them on five criteria. Ask for a quote on the same specification and volume, ask for a sample, and compare capability, capacity, quality, total cost and continuity rather than unit price. Deliverable: at least two quotes on one scope with a written comparison. Verification: every quote answers the same specification, and the comparison names the criterion that decides it.
3. Write the agreement's five parts and have a lawyer read it. Put price and payment, quality and acceptance, capacity and schedule, confidentiality and ownership, and exit and return into the document, with the performance indicators the Crown corporation's page names, and have a lawyer read the agreement. Deliverable: a signed agreement with the five parts present. Verification: each of the five parts can be found in the document by a reader who did not write it, and the lawyer's reading is recorded.
4. Run the pilot builds and keep the old way running until acceptance. Make a small run under the agreed method, inspect every unit, run a first-article inspection, and accept the pilot against the written test before any volume moves; keep the old way running until the first release has passed acceptance. Deliverable: pilot records with a first-article inspection and a written acceptance. Verification: the first release was delivered after the written acceptance, and the old way was still running on the day the pilot was accepted.
5. Write the triggers that bring the activity back, and keep the means to return. Set down the defect rate, the price and the late-release thresholds that would bring the work home, name the means kept for the return, and put a date on which each trigger is read. Deliverable: a written set of triggers with a reading calendar and a list of the means kept. Verification: each trigger names a threshold and a date it is read, and the means kept — protocol, bench, records and build cell — exist and are the company's.

In La Machine

In the application copy of September 7, 2026, read in its code on October 15, 2026, the pages that touch this chapter are three (La Machine, 2026). Control center → Plan tasks (/taches) holds tasks with a "Done when" criterion, and the steps of a transition can live there as tasks whose criterion a second person reads, because the Mark as done button does not check that criterion. Control center → Projects (chains) (/projets) shows the projects of a plan as chains of tasks, in read-only mode. Control center → Audit log (/journal) traces what happened to a record, never why. Only the administrator and agent roles open the Control center, and a business should check it in its own installation.

What a business keeps outside the product is the substance of a transfer. La Machine holds no supplier record or scorecard, no purchase order, no bill of materials, no production order and no traceability of lots or serial numbers, and no total cost, break-even volume or make-or-buy verdict. The scope, the quotes, the comparison, the agreement, the pilot records, the trigger calendar and the register of decisions live in the business's own documents, and this book is valid with any tool. The case does not use La Machine (Books G8, I8 and L5).

Self-Check

1. What moves when an activity leaves a business, and what never moves?
2. What does a scope hold, and why does scoping come before selecting a supplier?
3. On which five criteria is a contract manufacturer or service provider compared, and how is each one checked?
4. What are the five parts of an agreement for a transfer, and which book holds the Québec rules on the contract of enterprise or for services and on the sale?
5. What are the signals that bring an activity back in-house, and what means does a business keep so that the road back stays open?

Answers. 1. The work, the know-how that went with it, the records that describe it, the risk of a delay or a defect, and the oversight the business now owes all move; what never moves is the duty of the business to its own customers, who have one supplier of record and that supplier is the business. 2. A scope holds the specification, the volumes and their range, the acceptance test, the records to be returned and the schedule; scoping comes first because two suppliers can only be compared on the same scope. 3. Capability, checked by a sample and a first-article inspection; capacity, checked by a stated capacity and a schedule; quality, checked by test records and a visit; total cost, checked by a quote set beside the company's own cost on one basis; and continuity, checked by a second source and the terms of exit. 4. Price and payment; quality and acceptance; capacity and schedule; confidentiality and ownership; and exit and return. In Québec, the rules on the sale and the seller's warranty, the contract of enterprise or for services and the client's unilateral resiliation are in the Civil Code of Québec, up to date to August 12, 2026, and they are called to Book H1 (Contracts and Commercial Agreements) and to a lawyer; none of them is applied here. 5. The signals are a defect rate that stays above the line, a price that passes the line, and late releases that repeat; the means kept are the company's test protocol, a duplicate bench the company owns, records that come back, and a small build cell, and the case sets its triggers at a defect rate above 2 percent for 2 months running, a price above $215 a kit, or 3 late releases in a year, all posed by this book.

Chapter References

Barthélemy, J. (2001, Spring). The hidden costs of IT outsourcing. MIT Sloan Management Review. https://sloanreview.mit.edu/article/the-hidden-costs-of-it-outsourcing/

Business Development Bank of Canada. (2022). How to outsource: A decision-making guide for entrepreneurs (updated October 4, 2022). Retrieved October 15, 2026, from https://www.bdc.ca/en/articles-tools/operations/operational-efficiency/outsourcing-pros-cons

Civil Code of Québec, CQLR c CCQ-1991, arts. 1708, 1726, 1729, 2098, 2099, 2100, 2103, 2125, 2126 and 2129 (Québec). LégisQuébec, up to date to August 12, 2026 — read October 15, 2026. https://www.legisquebec.gouv.qc.ca/en/document/cs/CCQ-1991

Gray, J. V., Roth, A. V., & Leiblein, M. J. (2011). Quality risk in offshore manufacturing: Evidence from the pharmaceutical industry. Journal of Operations Management, 29(7–8), 737–752. https://doi.org/10.1016/j.jom.2011.06.004

La Machine. (2026). Documentation — route reference inventory. Internal product documentation, read October 16, 2026, together with the code of the pages this book describes: /taches, /projets and /journal.

Labour Code, CQLR c C-27, s. 45 (Québec). LégisQuébec, up to date to August 12, 2026 — read October 15, 2026. https://www.legisquebec.gouv.qc.ca/en/document/cs/C-27

Larsen, M. M., Manning, S., & Pedersen, T. (2013). Uncovering the hidden costs of offshoring: The interplay of complexity, organizational design, and experience. Strategic Management Journal, 34(5), 533–552. https://doi.org/10.1002/smj.2023

Parmigiani, A. (2007). Why do firms both make and buy? An investigation of concurrent sourcing. Strategic Management Journal, 28(3), 285–311. https://doi.org/10.1002/smj.580

Whitten, D., & Leidner, D. (2006). Bringing IT back: An analysis of the decision to backsource or switch vendors. Decision Sciences, 37(4), 605–621. https://doi.org/10.1111/j.1540-5414.2006.00140.x

Chapter 5 — Partnerships as an Alternative to Integration

Between buying on the market and owning a stage lies a middle ground of agreements that can secure supply, tooling and knowledge without a purchase of a company, so this chapter says what the hybrid middle is, what long-term supply agreements and flexible quantities do, how dedicated capacity and tooling held at a supplier work, what exclusive agreements secure and cost, what a joint investment is and why it is only named here, which safeguards answer the hold-up problem, how to work with a supplier that also competes, what the relational view claims, and when a partnership should become ownership or end, reads each study at its abstract, each model as a model and each theory as a theory, calls the law to the books that hold it, and tells the supply and tooling agreements of Hydrosentinelle inc. on days 57 and 58 and the second source of the sensor component, posed by this book.

What You'll Be Able to Do

• Describe the hybrid middle between a spot purchase and the ownership of a stage, and place a given supply relationship on one of its five rungs.
• Write a long-term supply agreement that holds a price or the way it is set, a quality and an acceptance, a forecast and the quantities each side commits to, a term and an exit.
• Decide who owns a tool made for one product and where that tool sits, and say what each arrangement lets each side do when the relationship ends.
• Read an exclusive agreement for what it secures and what it costs, and know which provisions of the Competition Act are called to Book H4 rather than applied here.
• Build the three safeguards against hold-up — a second source, a pledge and the ownership of the tool — and tell a second source that exists from one that exists only on a list.
• Decide when a partnership should become ownership or end, on a written trigger, and write the decision, the date and the exit.

Why It Matters

Many supply relationships in a small business are neither a market purchase nor an acquisition. A business that buys its boards order by order has a price and a delivery date and nothing else; a business that owns a board plant has capital, skills and volume to carry. Between those two sits an agreement that lasts: a price that is set for a term, a forecast that both sides see, a quantity that can move inside a band, a tool that one side owns and the other holds, a second supplier that is qualified and takes a share. Each of those arrangements buys something the market does not give and gives up something ownership would keep, and the trade is worth making deliberately rather than by drift.

The reason the middle deserves its own chapter is that it is where the hold-up problem lives. A tool made for one product is worth little to anyone else; a supplier that has bought capacity for one buyer has a stake in that buyer; a buyer that has qualified one source for a component has a stake in that source. Once a specific investment is made, the party that did not make it can behave opportunistically, and the theory of appropriable quasi rents says so. The answer is not always to own the stage. It can be to write the agreement so that leaving costs both sides something, to keep a second source alive, and to hold the tool.

The cost of the middle is freedom, and the chapter says so plainly. An exclusive agreement reserves a supply and gives up the freedom to buy elsewhere. A tool held at a supplier ties a business to a place. A joint investment ties two firms to one asset. None of those is wrong, and none of them is free. A business needs a way to see the rung it is standing on, the safeguard it holds, and the trigger that would move it up or down the ladder, and this chapter gives it without promising a saving, a premium or a result.

In This Chapter

Section 5.1 sets out the hybrid middle between the market and ownership, with a table of where this chapter stops and its neighbors begin and a table of the five rungs. Section 5.2 covers long-term supply agreements and flexible quantities, and calls the sale and the contract of enterprise to Book H1. Section 5.3 covers dedicated capacity and tooling held at a supplier, and the specific asset a mold is. Section 5.4 covers exclusive agreements, what they secure and what they cost, and calls section 77 of the Competition Act to Book H4. Section 5.5 covers joint investment without merging, and names it only. Section 5.6 covers the safeguards against hold-up: second sources, pledges and the ownership of tools. Section 5.7 covers working with a supplier that also competes with the business, and calls sections 45 and 90.1 of the Competition Act to Book H4. Section 5.8 covers relational advantage, shared knowledge and trust. Section 5.9 covers when a partnership should become ownership or end. Section 5.10 places the case's supply relationships on the ladder, with the supply and quality agreement of day 57, the tooling agreement of day 58 and the second source of the sensor component.

5.1 The Hybrid Middle: Neither Market Nor Ownership

A business has more than two choices. Between a purchase priced order by order and the ownership of a stage there are agreements that last, tooling made for one buyer, and investments made together, and each of them has a price, which is the freedom that it gives up.

The vocabulary for the middle comes from a conceptual analysis. The abstract of the paper read for this book reports that it combines institutional economics, contract law and organization theory to distinguish three generic forms of economic organization — market, hybrid and hierarchy — by their coordinating and control mechanisms and their abilities to adapt to disturbances, that each form is supported by a distinctive type of contract law, and that the cost-effective form varies systematically with the attributes of transactions (Williamson, 1991, abstract only). The paper argues rather than shows; it is a conceptual analysis and never measured evidence, and no test for a small firm follows from it. What it gives a business is a name for the middle and a reason to expect that the right form depends on the transaction rather than on a preference.

The hybrid mode was introduced later than the first two, and the lecture read for this book says so. The lecture states that taking the technological core as given, attention focuses on a series of separable make-or-buy decisions, backward, forward and lateral, to decide which transactions stay within the ownership boundary of the firm, and that the hybrid mode was introduced later (Williamson, 2009). That is a theoretical account, recalled here from section 1.1, and it is not a rule for any business.

The question of when an intermediate form becomes necessary has been synthesized. The abstract of the synthesis read for this book reports a review of theory and evidence on the advantages and disadvantages of vertical financial ownership relative to vertical contracts, and it proposes that without agency and transaction costs the two are equivalent, and that with them the question becomes when market mechanisms suffice, when intermediate forms of vertical contracting become necessary and when ownership is preferred (Mahoney, 1992, abstract only). The synthesis proposes a framework; it measures no effect, and it is recalled here from section 2.5.

Two consultants have taken a position on the same ground. The article, read for this book at its opening only and published in 1993, holds that vertical integration can be important but is hard to implement and costly to reverse (Stuckey & White, 1993). That is the position of two consultants, recalled from section 2.8; the quasi-integration options of the article are not visible on the page read and are not printed here.

What buyers and suppliers actually do together has been studied. The abstract of the study read for this book reports that closer buyer-supplier ties are distinguished by joint action, with roles integrated by undertaking activities together, that drawing on transaction costs such ties are useful to safeguard relationship-specific investments and to adapt to uncertainty, and that data from industrial firms and their suppliers support the model (Heide & John, 1990, abstract only). The abstract reports an association and no effect size, and it is recalled again in section 5.8.

Alliances themselves have been reviewed. The abstract of the review read for this book reports that alliances are shaped by the social networks in which firms are embedded, and it identifies five issues — formation, choice of governance structure, dynamic evolution, performance of alliances and the performance consequences for the firms entering them — with research and an agenda for each (Gulati, 1998, abstract only). A review measures no effect, and it is recalled again in section 5.5.

The table below sets this chapter beside the books that hold the neighboring ground, so that a reader knows what is treated here and what is cited and left alone. Read it as a boundary rather than as a summary: each row names a neighbor, what that neighbor holds, and what this chapter holds instead.

Table 5.1 — Where this chapter stops and its neighbors begin

The table below sets the five rungs between the market and ownership in order, with what each rung is in this book's own words and what it protects and what it costs. Read it as a ladder rather than as a ranking: a business can stand on more than one rung at once, and the case of section 5.10 does.

Table 5.2 — Five rungs between the market and ownership

DEFINITION The words this book uses for a framework supply agreement, flexible quantities, dedicated tooling, an exclusive agreement and a second source

These are this book's working words, and no source read defines them in this sense. A framework supply agreement is an agreement that sets prices, quality, forecasts and flexible quantities for a term. Flexible quantities are bands within which the buyer may change what it takes and the supplier must deliver. Dedicated tooling is a tool made for one buyer's product. An exclusive agreement is an agreement by which a business buys only from one supplier, or a supplier sells only to one buyer, for a thing or a territory. A second source is a supplier able to make the same thing. The negotiated alliance and the joint venture are Book L4's (Strategic Partnerships, Alliances and Joint Ventures, chapters 2 and 4) and are not redefined here. No figure belongs to any of these words, and the case's use of them is posed by this book.

5.2 Long-Term Supply Agreements and Flexible Quantities

A long-term supply agreement is the first rung above the spot purchase. It holds four things: the price or the way it is set, the quality and the acceptance, the forecast and the quantities that the business commits to and that the supplier holds, and the term and the exit. An agreement that holds all four can carry a relationship for years; an agreement that holds only a price is a spot purchase with a longer paper trail.

The quantities raise a difficulty that a model describes. The abstract of the model read for this book reports that a customer's planning forecast without commitment gives the customer an incentive to overforecast, and the supplier anticipates it, which makes the chain inefficient; the quantity flexibility contract couples the customer's commitment to buy no less than a percentage below the forecast with the supplier's guarantee to deliver up to a percentage above, and under certain conditions allocates the costs of demand uncertainty so that both parties reach the system-wide optimum (Tsay, 1999, abstract only). A model shows what follows from its assumptions; the abstract gives no percentage for a reader, and a real agreement sets its own bands. What the model gives a business is the shape of the clause: a floor the buyer commits to and a ceiling the supplier holds.

The law that governs what such an agreement is has to be read at its text and called to the book that holds it. The Civil Code of Québec, CQLR c CCQ-1991, up to date to August 12, 2026, read on October 15, 2026, provides in article 1708 that sale transfers ownership of property for a price in money, in article 2098 that a contract of enterprise or for services is one by which a contractor or provider undertakes to carry out physical or intellectual work or to supply a service for a price, and in article 2103 that the contract is a sale, not a contract of enterprise, where the work is merely accessory to the value of the property supplied; the client's unilateral resiliation in articles 2125, 2126 and 2129 is recalled from section 4.4. All of that is called to Book H1 (Contracts and Commercial Agreements) and to a lawyer, and none of it is applied here to any real agreement.

What a business writes into the agreement is its own work. The price clause can fix a price or set a formula that moves with a named input, and either way it should say what happens when the input moves. The quality clause should say what is accepted, how it is measured and what happens to a lot that fails. The forecast clause should say who sends the forecast, when, and how far ahead. The quantity clause should say the floor and the ceiling, and what happens when the buyer falls below the floor or the supplier cannot reach the ceiling. The term clause should say how long the agreement runs and how it is renewed. The exit clause should say how either side leaves, with what notice, and what is owed at the end.

The case reaches this rung on day 57, posed by this book. The supply and quality agreement with the contract manufacturer is signed on day 57, after the outside lawyer has read it, and the weekly build release of Karim Haddad is its working rhythm, as Chapter 8 sets it out. The agreement covers the final assembly and the end-of-line test of the kits on the company's duplicate test bench, with the firmware image supplied by the company. The agreement does not make the company exclusive to the contract manufacturer, and the company keeps a second source of the sensor component, which section 5.6 tells.

THE RULE A model shows what follows from its assumptions, never a percentage for a reader (Tsay, 1999)

The rule is the house rule of this book applied to a model. The abstract of the model read for this book reports a quantity flexibility contract in which the customer commits to buy no less than a percentage below the forecast and the supplier guarantees to deliver up to a percentage above, and in which, under certain conditions, the costs of demand uncertainty are allocated so that both parties reach the system-wide optimum (Tsay, 1999, abstract only). A model shows what follows from its assumptions, and the abstract gives no percentage for a reader. A real agreement sets its own bands, and those bands are read by Book H1 (Contracts and Commercial Agreements) and by a lawyer, never by this book. The model is not evidence about any market, and no business should copy a band from it. What a business can take from the model is the shape of the clause and the reason for it: a floor that the buyer commits to and a ceiling that the supplier holds, so that neither side carries all of the uncertainty of demand.

5.3 Dedicated Capacity and Tooling Held at a Supplier

A tool made for one product is a specific asset. It fits one enclosure, one board or one assembly, and it is worth little to anyone else. That is why the question of who owns it and where it sits matters more than its price.

The theory of the specific asset was set out in a law-and-economics article read for this book at its opening only. The opening text, deposited with its reprint, explores one particular cost of using the market system, the possibility of postcontractual opportunistic behavior, and identifies its likely source in appropriable specialized quasi rents: once a specific investment is made and such quasi rents exist, the possibility of opportunistic behavior is very real (Klein et al., 1978). That is a theory, read at its opening only, and it is recalled here from section 3.6. What it says to a business is that a tool made for one product changes the bargaining position of both sides.

Ownership of the tool is a separate question from its location, and the theory of residual rights explains why. The abstract of the theory read for this book reports that contractual rights are specific or residual, that when it is costly to list all specific rights over assets in a contract it may be optimal to let one party purchase all residual rights, that ownership is the purchase of these residual rights, that residual rights bought by one party are lost by the other, which creates distortions, and that one firm should buy another when its control raises the productivity of its management more than the loss of control lowers the other's (Grossman & Hart, 1986, abstract only). The theory holds; it measures no effect, and it is recalled here from section 1.4. What it gives a business is the reason to ask who holds the residual right over the tool, and not only who paid for it.

Supplier networks have been compared on this ground. The abstract of the study read for this book reports that, comparing two Japanese and three United States automakers, supplier-automaker asset specialization is positively related to performance, that human cospecialization is positively related to quality and new-model cycle time, and that site specialization is associated with lower inventory costs (Dyer, 1996, abstract only). The abstract reports associations in one industry and one period, and no cause and no finding outside that industry follows from it. It is recalled again in section 5.8.

The case's molds are the example, and the facts around them are printed by Book A5. Book A5 prints that the enclosure's molds would need to be replaced or redesigned within about five years, and that few buyers exist for a mold made for someone else's product. Posed by this book: the second-generation molds cost $72,000, are owned by the company, and are held at the molder under a tooling agreement signed on day 58, the day the molds are ordered. The financing of the tooling and its loan is Book A5's (Financing a Business, chapters 1, 3 and 6), cited here and not retold. The decision to own the molds and to hold them at the molder is this chapter's, and the numbers behind it are read in section 5.10.

The table below sets the three ways a tool can be held, with who can move it and what the agreement must say in each case. Read it as a checklist for a tooling agreement rather than as a description of any real one: the row that matters most is the first, because it is the arrangement that keeps a business able to leave.

Table 5.3 — Three ways to hold a tool and what each agreement must say

KEY CONCEPT A tool made for one product is a specific asset, and who owns it and where it sits decide who can leave

This is the idea of the section, and it is this book's own reading and never a finding of a source. Ownership of the tool and its location are two separate questions, and a business that answers only one of them has answered half of the problem. A tool owned by the business and held at the supplier under a written agreement keeps the business able to leave: the agreement names the ownership, the location, the upkeep and the release on request, and the business can move the tool when it chooses. A tool owned by the supplier leaves the business tied to that supplier, because leaving means leaving the tool behind. The molds of the case are the example: the company owns the second-generation molds, posed at $72,000, and holds them at the molder under a tooling agreement signed on day 58, so that the asset on which the enclosure's price depends stays the company's. No figure of a source appears here, and the case's figures are posed by this book.

TOOLS OF THE TRADE What a business needs to follow the assets it holds at a supplier: a register, stored agreements, a calendar and a drawing file

Five functions carry this work, and a business should look for them rather than for a product. The first is a register of the assets held at a supplier, with the location and the owner of each. The second is a stored copy of each tooling agreement with its date, so that the terms can be read when a question arises. The third is a calendar of the dates when a tool is checked, so that a tool that has not been seen for a year is seen. The fourth is a place for the drawings and the dimensions that the company owns, because a tool is only as useful as the drawings that go with it. The fifth is one register of decisions, so that the reason a tool was bought and the trigger that would move it are written beside it. Chapter 10 gives these functions in full, with the tool tables of that chapter. No brand name belongs on this list and no price; the requirement is the function.

5.4 Exclusive Agreements: What They Secure and What They Cost

None of the texts read for this book obliges a business to sign an exclusive agreement, and none forbids it. This book does not say that no other rule exists, and it does not say whether any real exclusivity is lawful or unlawful; only the Tribunal or a court decides that.

What an exclusive agreement secures is worth naming first. It can reserve a supply, so that the buyer is not one of several customers competing for the same capacity. It can hold a capacity, so that the supplier keeps room for the buyer's volumes. It can protect a know-how, so that a method or a design shared with the supplier does not travel to a competitor. And it can give both sides a reason to invest in the relationship, because neither can walk away without cost. What it costs is the freedom to buy elsewhere, a dependence on one party, and the question whether the practice falls under the Competition Act, which is called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements, chapter 5) and never applied here.

That question is called to Book H4 and is never answered here. The Competition Act, R.S.C. 1985, c. C-34, s. 77, current to September 30, 2026 (last amended March 26, 2026), read on October 16, 2026, provides in subsection 77(1) the definitions of exclusive dealing, market restriction and tied selling; in subsection 77(2) that the Tribunal may act on exclusive dealing or tied selling, and in subsection 77(3) on market restriction, only where the practice of a major supplier or a widespread practice is likely to lessen competition substantially; in paragraph 77(4)(a) that exclusive dealing or market restriction engaged in only for a reasonable period to facilitate entry is excluded, and that exclusion does not reach tied selling; in the last sentence of subsection 77(4) that practices between affiliated entities are excluded; and in subsection 77(6) that, for market restriction, a supplier of ingredients and the person who processes them into a food or drink sold under the first person's trademark are deemed affiliated. All of it is called to Book H4, and no conclusion about any real agreement is drawn.

The regulator's own approach is a separate matter from the Act. The Competitor collaboration guidelines, dated May 6, 2021, with the page modified June 27, 2024, read on October 15, 2026, carry a notice saying that the Bureau is reviewing the guidelines after recent amendments to the Act, so they are the Bureau's approach, under review, and never the law (Competition Bureau, 2021). The guidelines say that purely vertical agreements between a supplier and its customer will be assessed under the civil provisions of Part VIII rather than under section 45. That is the Bureau's stated approach on the day it was read, and no conclusion about any real agreement follows from it.

The table below sets the three practices named in section 77 of the Act beside the provision that governs each, what the text says in this book's words, and the book that holds it. Read it as a map to Book H4 rather than as a summary of the law: each row points to a provision and to the book that reads it, and none of them decides anything about a real agreement.

Table 5.4 — Exclusive dealing, tied selling and market restriction in section 77 of the Act

The case does not sign an exclusive agreement, posed by this book. The supply and quality agreement of day 57 does not make the company exclusive to the contract manufacturer, and the company keeps a second source of the sensor component, which section 5.6 tells. That is a choice about the rung the company stands on rather than a statement about the Act, and it is posed by this book.

QUÉBEC & CANADA NOTE Exclusive dealing, tied selling and market restriction are held by Book H4, and this book explains the law as published and is not legal advice

This book explains the law as published and is not legal advice. The Canadian point of this section is that exclusive dealing, tied selling and market restriction are named in section 77 of the Competition Act, R.S.C. 1985, c. C-34, current to September 30, 2026 (last amended March 26, 2026), read on October 16, 2026: subsection 77(2) reaches exclusive dealing and tied selling, subsection 77(3) reaches market restriction, and the reasonable period of paragraph 77(4)(a) reaches exclusive dealing and market restriction and never tied selling. Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements, chapter 5) holds the Act, and a lawyer reads a given agreement. The Competitor collaboration guidelines of 2021 are the Bureau's approach, under review, and never the law (Competition Bureau, 2021). This book recommends no vendor, no supplier and no adviser, and it says nothing about whether any real exclusivity complies with any rule.

5.5 Joint Investment Without Merging

A joint investment is two firms funding one asset or one capacity together without merging. It is the fourth rung of the ladder, and this book names it and goes no further, because the negotiated alliance and the joint venture are Book L4's (Strategic Partnerships, Alliances and Joint Ventures, chapters 2 and 4), cited here and not retold.

Why a business might take this rung is easy to state. The capital may be too large for one firm to carry alone, and a shared asset spreads the outlay across two balance sheets. The know-how may be split, with one firm holding the process and the other the market, and neither able to build the asset alone. The risk may be too concentrated, and a partner may be willing to hold half of it. And the volume may be too small for one firm to justify the asset and large enough for two firms together.

What it costs is the independence of both. Two firms that fund one asset have to agree on how the asset is used, who operates it, how the output is divided, what happens when one wants out and what happens when one fails. That governance is the work of the arrangement, and it is written before the money moves or it is written in a dispute. The exit is the hardest clause, because an asset built for two firms is worth less to either one alone, and the agreement has to say what happens then.

The review recalled from section 5.1 gives the ground on which such arrangements sit. Its abstract reports that alliances are shaped by the social networks in which firms are embedded, and that the choice of governance structure is one of five issues, each with research and an agenda (Gulati, 1998, abstract only). A review measures no effect. What it suggests to a business is that the choice of governance structure is a question in its own right, separate from the question of whether to enter the arrangement at all.

The case makes no joint investment, posed by this book. The company owns its molds alone and holds them at the molder under a tooling agreement. No second firm is a party to any asset of the company, and no joint venture is formed. The section is short for that reason: the rung exists on the ladder, the case does not stand on it, and the book that holds the negotiated alliance and the joint venture is Book L4.

5.6 Safeguards Against Hold-Up: Second Sources, Pledges and Ownership of Tools

Once a specific investment is made, the party that did not make it can behave opportunistically, as the theory of appropriable quasi rents holds. That article was read at its opening only and is recalled here from section 3.6: it identifies the likely source of the behavior in appropriable specialized quasi rents (Klein et al., 1978). The practical question is what a business does about it.

Three safeguards answer the problem, and a business can hold more than one at a time. The first is a second source, which is a supplier able to make the same thing. The second is a pledge, which is something the other party puts at stake so that leaving costs it too: a deposit, a dedicated capacity, a stake in the relationship. The third is the ownership of the tool, which section 5.3 covers and which keeps the specific asset on the buyer's side of the line.

A second source is discussed in a business-review article read for this book at its opening only, published in 2004. The article holds that supply chain risks fall into delays, disruptions, forecast inaccuracies, systems breakdowns, intellectual property breaches, procurement failures, inventory problems and capacity issues, which can hurt sales, raise costs or both; that a manufacturer with several suppliers kept producing after a plant fire while a single-sourced competitor was disrupted for months; and that no single strategy protects every supply chain, so the mitigation must match the risk (Chopra & Sodhi, 2004). That is the position of two authors read at their opening only, and the exhibits of the article are not visible on the page read and are not printed here. It is recalled from section 3.5.

The pledge has been studied in one kind of firm. The abstract of the study read for this book reports that offsetting investments safeguard the specific assets of small firms where vertical integration is not feasible and long-term contracts are absent, and that in 199 manufacturers' agencies, agencies with more specific assets bonded more closely to their own accounts, lowered their dependence on the principal, and performed better financially when dependence was reduced at relatively high levels of specific investment (Heide & John, 1988, abstract only). The abstract reports associations in one kind of firm, and no cause and no finding outside agencies follows from it. What it gives a business is the name for the safeguard: an offsetting investment, something the other side puts at stake.

The case holds two of the three safeguards, and the exposure is printed by Book G8. Book G8 prints that a supplier's component shortage that stops kit assembly is rated at $40,000 and 30 days, reduced by a second source and a safety stock. Posed by this book: samples from the second source of the sensor component are received on day 68, the second source is qualified on day 120, and it holds 30 percent of the component by day 365. The company also owns the second-generation molds, which is the third safeguard, and holds the molds at the molder under the tooling agreement of day 58.

WATCH OUT A second source that exists only on a list is not a safeguard

The trap of this section is a second source that exists only on a list. A name is written into a document, a column is filled in a register, and nothing is ever bought from the second supplier. The signs are easy to read once a business looks for them: a second source that has never sent a sample, a qualification that nobody owns, a share of the volume that has never moved, and a review that reads the name and not the share. The remedy is to qualify the second source, to buy a share from it, and to read that share at each review, so that the second source is a supplier rather than a line in a document. The reasoning is this book's own, and no source read measures how often a second source stays on a list. In the case, posed by this book, the second source of the sensor component sends samples on day 68, is qualified on day 120, and holds 30 percent of the component by day 365, so that the safeguard is a real one rather than a name.

5.7 Working With a Supplier That Also Competes With You

A supplier can also be a competitor. The same firm may make a component for a business and sell a product that competes with the business's own, and the same agreement then deals with a seller and a rival at once. That changes what the agreement has to say and what the business has to keep.

The subjects a lawyer reads first are prices, customers and volumes. An agreement that fixes a price with a competitor, that allocates customers or territories with a competitor, or that limits what either side produces or supplies can fall under the criminal provision of the Competition Act, R.S.C. 1985, c. C-34, s. 45, current to September 30, 2026 (last amended March 26, 2026), read on October 16, 2026: subsection 45(1) makes it an offense for a person, with a competitor in respect of a product, to agree to fix the price, to allocate sales, territories, customers or markets, or to fix, control, prevent, lessen or eliminate production or supply, with the ancillary-restraint defense of subsection 45(4) and with "competitor" including a person likely to compete absent the agreement under subsection 45(8). Section 90.1(1) lets the Tribunal make a civil order where an agreement between persons of whom two or more are competitors prevents or lessens competition substantially, and subsection 90.1(1.01) reaches even non-competitors where a significant purpose is to prevent or lessen competition. All of it is called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements, chapter 5), and no conclusion about any real agreement is drawn here.

The regulator's approach adds a layer that a business should know exists without treating it as the law. The Competitor collaboration guidelines of 2021, under review, say that the ancillary restraints defense of subsection 45(4) requires a restraint directly related to and reasonably necessary for a broader lawful agreement among the same parties (Competition Bureau, 2021). Section 3.7.4 of the guidelines treats joint production agreements, which include specialization agreements, shared facilities and subcontracting where one firm has another make products for it, and it names as concerns, in the page's terms, a substantial lessening of competition, a significant reduction of output, an input price lowered below the competitive level, and a substantial degree of commonality of costs that restricts price competition when the common process is a substantial share of the costs of production. The guidelines also say that a vertical arrangement that contains an agreement between competitors to fix prices, allocate markets or limit supply will generally be assessed under section 45. The page predates subsections 90.1(1.01) and (1.3) of the Act read today, so the guidelines are read as the Bureau's approach and never as the law. The guidelines give no rule on toll manufacturing by name.

What a business does with all of that is practical. It keeps its talk to the work: the specifications, the delivery dates, the quality and the price of the thing being bought. It keeps a record of what was agreed, so that the agreement can be read later. It keeps the subjects that a lawyer reads first — prices, customers and volumes — out of any conversation that is not with its own lawyer. And it reads the agreement again when the supplier's own products begin to compete with the business's products.

The case has no supplier that competes with it, posed by this book. The contract manufacturer, the molder and the second source of the sensor component make parts and assemblies for the company and sell no competing product, and no question of a supplier that is also a rival arises for the company.

WHAT CHANGED The Bureau's collaboration guidelines of 2021 are under review after recent amendments to the Act (Competition Bureau, 2021)

The dated change comes from the guidelines read for this book. The Competitor collaboration guidelines are dated May 6, 2021, and the page was modified June 27, 2024; a notice on the page says that the Bureau is reviewing the guidelines after recent amendments to the Act (Competition Bureau, 2021). The guidelines are therefore the Bureau's approach, under review, and never the law, and the page predates subsections 90.1(1.01) and (1.3) of the Act read today. Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements, chapter 5) holds the Act, and a lawyer reads a given agreement. The guidelines say that purely vertical agreements between a supplier and its customer will be assessed under the civil provisions of Part VIII rather than under section 45, and that a vertical arrangement containing an agreement between competitors to fix prices, allocate markets or limit supply will generally be assessed under section 45. No other date belongs to this callout, and no conclusion about any real agreement is drawn from it.

5.8 Relational Advantage: Shared Knowledge and Trust

A relationship that lasts produces something that a spot purchase does not: shared knowledge and a reason to use it. The study recalled from section 5.1 gives the name. Its abstract reports that closer buyer-supplier ties are distinguished by joint action, with roles integrated by undertaking activities together (Heide & John, 1990, abstract only). It reports an association and no effect size.

The comparison of supplier networks recalled from section 5.3 gives a second reading. Comparing two Japanese and three United States automakers, the abstract reports that human cospecialization is positively related to quality and new-model cycle time (Dyer, 1996, abstract only). It reports associations in one industry and one period, and no cause and no finding outside that industry follows from it.

The relational view is a theory, and it is written as one. The abstract of the theory read for this book reports that a firm's critical resources may span its boundaries, and that four potential sources of interorganizational advantage are relation-specific assets, knowledge-sharing routines, complementary resources and capabilities, and effective governance, with isolating mechanisms that preserve relational rents (Dyer & Singh, 1998, abstract only). The theory holds; it measures no effect, and no business should read it as a promise of a return.

What a business shares with a good supplier is worth naming. It shares the forecast, so that the supplier can plan capacity. It shares the drawings and the dimensions that the supplier needs to make the part, and it keeps the drawings that are its own. It shares the lessons of a defect, so that the same defect does not return. It shares the dates of a change, so that the supplier is not surprised by a new revision. And it shares the readings of the monthly scorecard, so that both sides see the same numbers.

What a business keeps is equally worth naming. The firmware image and the alert logic stay in-house, and Book H5 (Intellectual Property, chapters 1 and 5) holds the firmware as an asset of the company and the trade secrets around it. The customer relationship stays with the business, and Book H2 (Privacy and Personal Information) holds the personal information that a provider may see. The method of making the product stays with the business, and the supplier receives what it needs to make the part and no more.

Trust is built by routines that are kept rather than by a declaration. A weekly release that arrives on the same day, a monthly scorecard that is read and answered, a yearly audit that is held at the supplier's plant: those are the things that make a relationship work, and they are the same routines that Chapter 8 sets out. Trust never replaces the written agreement, and the agreement never replaces the routine.

WHO DOES THIS One person keeps each supplier relationship, and the owner reads the monthly record

Who keeps a supplier relationship depends on the size of the business, and the roles below are this book's suggestion except where the case is named. A solo owner keeps the relationship alone and writes down what was agreed, because a memory is not a record. In a small team one person is the single contact of the supplier, so that the supplier hears one voice, and the owner reads the monthly record. In an established business a supplier manager keeps the relationship and an operations lead owns the quality side, as Chapter 9 sets out. In the case, posed by this book, Karim Haddad keeps the weekly release and the monthly scorecard, and Laurence Bergeron keeps the yearly audit at the contract manufacturer and the yearly check of the molds at the molder. No hire is made for this work and no dollar is spent on it beyond the hours the two founders already carry.

5.9 When a Partnership Should Become Ownership, or End

A partnership is not a permanent arrangement, and the triggers that reopen it are written at the start rather than found later. The triggers of the case are written at Gate 3 of the plan of Chapter 11: a defect rate, a price and a number of late releases.

The triggers that reopen a partnership are ordinary events. The volume grows, and the stage that was too small to own becomes large enough. The supplier's quality falls, and the readings of the monthly scorecard show it. The price rises, and the comparison between making and buying changes. The supplier changes owner, and the new owner has different terms. The tool needs replacing, and the replacement is a moment to decide who owns it. Each is a reason to read the decision again, not to act the same day.

The theory of how vertical scope changes over time is a theory. The abstract of the theory read for this book reports that transaction costs and capabilities are intertwined in determining vertical scope, that capability differences are a necessary condition for vertical specialization, that lower transaction costs lead to specialization only when capabilities along the chain are heterogeneous, and that four evolutionary mechanisms reshape vertical scope over time, illustrated by mortgage banking in the United States, which disintegrated, and Swiss watch manufacturing, which integrated (Jacobides & Winter, 2005, abstract only). The theory holds; it measures no effect, and it gives no forecast for any industry. It is recalled here from section 1.4.

A partnership ends cleanly when its exit was written. The exit clause says how much notice either side gives, what is owed at the end, what happens to the tool, what happens to the drawings, and what happens to the stock in the supplier's hands. A partnership that ends without an exit clause ends in a negotiation, and a negotiation at the end of a relationship is a negotiation in which the side that needs the tool has less room than the side that holds it.

Ownership comes when the stage has become core and specific and the business has the volume, the skills and the capital to own it, which is the test of section 1.6. The grid of Chapter 3 is the instrument: the stage is scored on the same six criteria as every other activity, and the verdict follows. A stage that scores high on core and specificity and that the business can carry moves to ownership; a stage that scores low stays on the rung it is on, or moves down to a second source and a changed agreement.

METHOD Deciding whether a partnership should become ownership or end in five steps

First, read the trigger that fired and the readings behind it, so that the decision rests on the monthly scorecard and the weekly releases rather than on an impression. Second, ask whether the stage has become core and specific, using the criteria of the grid of Chapter 3. Third, ask whether the business has the volume, the skills and the capital to own it, which is the test of section 1.6. Fourth, compare ownership, a second source and a changed agreement on the grid of Chapter 3, so that the three options are read on the same basis. Fifth, write the decision, the date and the exit, and tell the supplier on the terms of the agreement. The sequence is this book's own, no source read tests it, and it is a method rather than a finding. The case writes its triggers at Gate 3 and reaches none of them by day 365, posed by this book.

5.10 The Case's Supply and Tooling Agreements

Hydrosentinelle inc. is a fictional business created for this encyclopedia, and its supply relationships are posed by this book. The company designs and sells connected water-leak detectors for commercial and rental buildings, buys its electronic boards from the contract manufacturer, holds the tooling that Book A5 financed, and installs its kits through the staff installer and the certified installers of Book F9 (Building and Running a Partner Program, chapter 4).

The company stands on four of the five rungs of the ladder. The sensor component is bought as a spot purchase from two sources. The electronic boards and the kit assembly sit under a framework supply agreement with the contract manufacturer, signed on day 57. The second-generation molds are dedicated tooling at a supplier, owned by the company and held at the molder under a tooling agreement signed on day 58. The firmware and the alert logic are owned outright and made in-house, as Book H5 (Intellectual Property, chapters 1 and 5) holds them. No joint investment is made, and the company stands on no rung of that kind.

Figure 5.1 sets five forms between the market and ownership on a ladder and places each of the case's supply relationships on its rung. Read it from left to right, rung by rung: the spot purchase, the framework supply agreement, the dedicated tooling at a supplier, the joint investment, and ownership; under the ladder, the case's relationships sit on four of the five rungs, with no joint investment. Above the ladder, more market is flexible and cheap to leave, and more ownership brings control and fixed cost; the note adds that a specific asset is written down before it is built.

Figure 5.1 — From market to ownership. Two sources for a component, a supply and quality agreement for boards and assembly, molds owned at the molder for $72,000, firmware made in-house; no joint investment (posed).

The table below places each of the case's supply relationships on its rung, with what is posed and the day it happens. Read it as the case's ladder in one view: every figure in it is posed by this book, and the relationships are illustrative.

Table 5.5 — The case's supply relationships placed on the ladder, posed by this book

The supply and quality agreement of day 57 is the company's framework supply agreement. It sets the price of the assembled and tested kit, the quality and the acceptance, the forecast and the quantities, the term and the exit, and it is signed after the outside lawyer has read it. The weekly build release of Karim Haddad is its working rhythm, as Chapter 8 sets out. The agreement does not make the company exclusive to the contract manufacturer.

The tooling agreement of day 58 is the company's dedicated tooling arrangement. The second-generation molds cost $72,000, are owned by the company, and are held at the molder; the agreement names the ownership, the location, the upkeep, the release on request and what happens if the molder fails. The financing of the molds is Book A5's (Financing a Business, chapters 1, 3 and 6), cited here and not retold. The numbers behind the decision are read in the callout below.

In La Machine, as of the application copy of September 7, 2026, the pages that touch this chapter are CRM → Companies (business/marketing/companies), where a supplier or a molder can be recorded as a company with its sector, size, site, status and notes, and CRM → Contacts (business/marketing/contacts), which holds the people of the companies recorded (La Machine, 2026). They have no field for a contract, a price, a tooling or a performance, and La Machine holds no purchase order, no bill of materials, no supplier scorecard, no inventory, no production order and no register of tooling; check it in your installation. The case does not use La Machine, this book is valid with any tool, and La Machine is described and never sold.

CASE IN POINT Hydrosentinelle inc. signs a supply and quality agreement on day 57 and a tooling agreement on day 58, and qualifies a second source

Hydrosentinelle inc. is a fictional business created for this encyclopedia; its founders, its employees, its customers, its partners, its suppliers, its figures and the events described are invented, and any resemblance to a real business is coincidental. The company signs a supply and quality agreement with the contract manufacturer on day 57, after the outside lawyer has read it, and the weekly build release of Karim Haddad is its working rhythm. The company signs a tooling agreement with the molder on day 58, the day the second-generation molds are ordered, and the molds are owned by the company and held at the molder. Samples from the second source of the sensor component are received on day 68, and the second source is qualified on day 120. Laurence Bergeron keeps the yearly audit at the contract manufacturer and the yearly check of the molds at the molder. No joint investment is made, no exclusive agreement is signed, and no result of the company is credited to any of these agreements. Every element is posed by this book and illustrative.

DO THE MATH Molds at $72,000 and $70 saved a kit: why the tooling agreement protects the payback

Illustrative figures; the case is fictional. Step 1 — The molder's price is $134 an enclosure set and the company's own price with the new molds is $64, so owning saves $70 a kit. Step 2 — The molds cost $72,000, posed. Step 3 — $72,000 divided by $70 is 1,028.6 kits, the volume at which the molds are paid back. Step 4 — The five-year life of the molds is 1,400 kits, so the payback comes inside the life, in year 4, as Chapter 7 reads. Step 5 — The tooling agreement protects the asset on which these savings depend, because a payback counted in kits is worth counting only if the company can still use and move the molds when the volume arrives, and every figure is posed by this book and illustrative.

IN LA MACHINE La Machine can record a supplier and its people as a company and contacts, with no field for a contract, a price or a tooling

In the application copy of September 7, 2026, CRM → Companies (business/marketing/companies) holds one record for each company, with its name, sector, size, site, status and notes, and it has no field for a supplier, a contract, a price, a tooling or a performance (La Machine, 2026). CRM → Contacts (business/marketing/contacts) holds the people of the companies recorded, under the rules that Book H2 (Privacy and Personal Information) holds for personal information. Neither page is supplier management, and neither holds a tooling register, a contract repository or a scorecard. They are an example of a record of whom a business deals with, never of what it agreed with them. A business should check it in its own installation. The case does not use La Machine, and La Machine is described and never sold.

Costly Mistakes

• Treating every supply relationship as either a purchase or an acquisition. The middle is a real choice, and a business that sees only two choices writes neither the agreement nor the safeguard that the middle needs.
• Writing a long-term agreement with a price and nothing else. A price without a quality, a forecast, a quantity band, a term and an exit is a spot purchase with a longer paper trail.
• Leaving a tool made for one product in the supplier's ownership. A tool owned by the supplier ties the buyer to that supplier, because leaving means leaving the tool behind.
• Signing an exclusive agreement without reading what it costs. An exclusive agreement reserves a supply and gives up the freedom to buy elsewhere, and the question whether the practice falls under the Competition Act is called to Book H4 and to a lawyer.
• Keeping a second source that exists only on a list. A second source that has never sent a sample and whose qualification nobody owns is a name in a document rather than a safeguard.
• Sharing the wrong things with a supplier. The forecast, the drawings the supplier needs and the lessons of a defect are worth sharing; the firmware image, the alert logic and the customer relationship stay with the business.
• Letting a partnership end without an exit clause. A relationship that ends without a written exit ends in a negotiation, and the side that needs the tool has less room than the side that holds it.

Key Takeaways

• Between a purchase priced order by order and the ownership of a stage there are agreements that last, tooling made for one buyer, and investments made together, and each of them has a price, which is the freedom that it gives up.
• The vocabulary of the middle comes from a conceptual analysis that distinguishes market, hybrid and hierarchy and holds that the cost-effective form varies with the attributes of transactions (Williamson, 1991, abstract only), and the hybrid mode was introduced later (Williamson, 2009).
• A long-term supply agreement holds a price or the way it is set, a quality and an acceptance, a forecast and the quantities each side commits to, a term and an exit; the quantity flexibility contract is a model that shows the shape of the quantity clause and gives no percentage for a reader (Tsay, 1999, abstract only).
• A tool made for one product is a specific asset, and who owns it and where it sits decide who can leave; the case owns its second-generation molds, posed at $72,000, and holds them at the molder under a tooling agreement signed on day 58.
• The three safeguards against hold-up are a second source, a pledge and the ownership of the tool, and the case holds two of them: the second source of the sensor component, qualified on day 120 and holding 30 percent of the component by day 365, and the molds it owns.
• A partnership is read again when a written trigger fires, and it becomes ownership when the stage has become core and specific and the business has the volume, the skills and the capital to own it; the case writes its triggers at Gate 3 and reaches none of them by day 365, posed by this book.

Your Action Plan

1. Place every supply relationship on the ladder. Write each relationship of the business on one of the five rungs — a spot purchase, a framework supply agreement, dedicated tooling at a supplier, a joint investment or ownership — and say in one sentence why it sits there. Deliverable: a written ladder of the business's supply relationships. Verification: every relationship has a rung, and no relationship is left without one.
2. Write the four clauses of each long-term agreement. For each framework supply agreement, write the price or the way it is set, the quality and the acceptance, the forecast and the quantity bands, and the term and the exit, and have the agreement read by a lawyer. Deliverable: a written clause list for each agreement. Verification: every agreement names a price, a quality, a quantity band, a term and an exit, and the outside lawyer has read it.
3. Decide who owns each tool and where it sits. For every tool made for one product, write the owner, the location, the upkeep, the release on request and what happens if the holder fails, and put the tooling agreement in a place where it can be found. Deliverable: a tooling register with an owner and a location for each tool. Verification: every tool has an owner, a location and a stored agreement with its date.
4. Qualify a second source and buy a share from it. Choose the component that would stop the work if it ran short, qualify a second supplier, buy a share of the volume from it, and read that share at each monthly review. Deliverable: a qualified second source with a share of the volume. Verification: the second source has sent samples, is qualified, and its share is written in the monthly record.
5. Write the triggers that would move a relationship up or down the ladder. For each partnership, write the trigger that would reopen the decision — a defect rate, a price, a number of late releases, a change of owner, a tool that needs replacing — and the date on which it is read. Deliverable: a written trigger beside each partnership. Verification: every partnership has a trigger with a level and a period, and the trigger is read at the quarterly review.

In La Machine

In the application copy of September 7, 2026, the pages that touch this chapter are CRM → Companies (business/marketing/companies), which holds one record for each company with its name, sector, size, site, status and notes, and CRM → Contacts (business/marketing/contacts), which holds the people of the companies recorded (La Machine, 2026). A supplier, a molder or a second source can be recorded there as a company, and the people met at each can be recorded as contacts, under the rules that Book H2 (Privacy and Personal Information) holds for personal information. Neither page is supplier management, and neither holds a tooling register. A business should check it in its own installation.

What a business needs for this chapter lives outside the platform. La Machine holds no field for a contract, a price, a tooling or a performance, no purchase order, no bill of materials, no supplier scorecard, no inventory, no production order and no register of tooling; it holds no place for the clauses of a framework supply agreement, no place for the ownership and the location of a tool, no place for the second source's share of the volume and no place for the triggers that would move a relationship up or down the ladder. Those live in a written register, a stored agreement file and a calendar that the business keeps, and this book is valid with any tool. The case does not use La Machine, and La Machine is described and never sold.

Self-Check

1. What are the five rungs between the market and ownership, and what does each one protect and cost?
2. What four things does a long-term supply agreement hold, and what does the quantity flexibility contract show and not show?
3. Why is the ownership of a tool a separate question from its location, and what must a tooling agreement say when the business owns the tool and the supplier holds it?
4. What does an exclusive agreement secure and cost, and where are exclusive dealing, tied selling and market restriction held?
5. What are the three safeguards against hold-up, and when should a partnership become ownership or end?

Answers. 1. The five rungs are a spot purchase, which protects freedom and costs security of supply; a framework supply agreement, which protects supply and costs some freedom; dedicated tooling at a supplier, which protects the asset and costs a dependence on the place where it sits; a joint investment, which shares the risk and costs the independence of both; and ownership, which protects control and costs capital, skills and volume. 2. A long-term supply agreement holds the price or the way it is set, the quality and the acceptance, the forecast and the quantities each side commits to, and the term and the exit; the quantity flexibility contract is a model in which the customer commits to buy no less than a percentage below the forecast and the supplier guarantees to deliver up to a percentage above, and it shows the shape of the quantity clause while giving no percentage for a reader (Tsay, 1999, abstract only). 3. Ownership of the tool and its location are two separate questions, because a tool owned by the business and held at the supplier keeps the business able to leave while a tool owned by the supplier ties the business to that supplier; when the business owns the tool and the supplier holds it, the agreement must say the ownership, the location, the upkeep, the release on request and what happens if the supplier fails. 4. An exclusive agreement can reserve a supply, hold a capacity, protect a know-how and give both sides a reason to invest, and it costs the freedom to buy elsewhere, a dependence on one party and the question whether the practice falls under the Competition Act; exclusive dealing, tied selling and market restriction are named in section 77 of the Competition Act, current to September 30, 2026, and are called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements, chapter 5). 5. The three safeguards are a second source, a pledge and the ownership of the tool; a partnership should become ownership when the stage has become core and specific and the business has the volume, the skills and the capital to own it, and it should end when a written trigger fires and the exit clause is read, with the decision, the date and the exit written down.

Chapter References

Chopra, S., & Sodhi, M. S. (2004, Fall). Managing risk to avoid supply-chain breakdown. MIT Sloan Management Review. https://sloanreview.mit.edu/article/managing-risk-to-avoid-supplychain-breakdown/

Civil Code of Québec, CQLR c CCQ-1991, arts. 1708, 1726, 1729, 2098, 2099, 2100, 2103, 2125, 2126 and 2129 (Québec). LégisQuébec, up to date to August 12, 2026 — read October 15, 2026. https://www.legisquebec.gouv.qc.ca/en/document/cs/CCQ-1991

Competition Act, R.S.C. 1985, c. C-34, ss. 45, 77, 90.1, 91, 92, 109 and 110 (Canada). Justice Laws Website, current to September 30, 2026 (last amended March 26, 2026) — read October 16, 2026. https://laws-lois.justice.gc.ca/eng/acts/C-34/

Competition Bureau. (2021). Competitor collaboration guidelines. Government of Canada. (Enforcement guidelines dated May 6, 2021; page modified June 27, 2024.) Retrieved October 15, 2026, from https://competition-bureau.canada.ca/en/how-we-foster-competition/education-and-outreach/competitor-collaboration-guidelines

Dyer, J. H. (1996). Specialized supplier networks as a source of competitive advantage: Evidence from the auto industry. Strategic Management Journal, 17(4), 271–291. https://doi.org/10.1002/(SICI)1097-0266(199604)17:4<271::AID-SMJ807>3.0.CO;2-Y

Dyer, J. H., & Singh, H. (1998). The relational view: Cooperative strategy and sources of interorganizational competitive advantage. The Academy of Management Review, 23(4), 660–679. https://doi.org/10.5465/amr.1998.1255632

Grossman, S. J., & Hart, O. D. (1986). The costs and benefits of ownership: A theory of vertical and lateral integration. Journal of Political Economy, 94(4), 691–719. https://doi.org/10.1086/261404

Gulati, R. (1998). Alliances and networks. Strategic Management Journal, 19(4), 293–317. https://doi.org/10.1002/(SICI)1097-0266(199804)19:4<293::AID-SMJ982>3.0.CO;2-M

Heide, J. B., & John, G. (1988). The role of dependence balancing in safeguarding transaction-specific assets in conventional channels. Journal of Marketing, 52(1), 20–35. https://doi.org/10.1177/002224298805200103

Heide, J. B., & John, G. (1990). Alliances in industrial purchasing: The determinants of joint action in buyer-supplier relationships. Journal of Marketing Research, 27(1), 24–36. https://doi.org/10.1177/002224379002700103

Jacobides, M. G., & Winter, S. G. (2005). The co-evolution of capabilities and transaction costs: Explaining the institutional structure of production. Strategic Management Journal, 26(5), 395–413. https://doi.org/10.1002/smj.460

Klein, B., Crawford, R. G., & Alchian, A. A. (1978). Vertical integration, appropriable rents, and the competitive contracting process. The Journal of Law and Economics, 21(2), 297–326. https://doi.org/10.1086/466922

La Machine. (2026). Documentation — route reference inventory. Internal product documentation, read October 16, 2026, together with the code of the pages this book describes: business/marketing/companies and business/marketing/contacts.

Mahoney, J. T. (1992). The choice of organizational form: Vertical financial ownership versus other methods of vertical integration. Strategic Management Journal, 13(8), 559–584. https://doi.org/10.1002/smj.4250130802

Stuckey, J., & White, D. (1993, Spring). When and when not to vertically integrate. Sloan Management Review. https://sloanreview.mit.edu/article/when-and-when-not-to-vertically-integrate/

Tsay, A. A. (1999). The quantity flexibility contract and supplier-customer incentives. Management Science, 45(10), 1339–1358. https://doi.org/10.1287/mnsc.45.10.1339

Williamson, O. E. (1991). Comparative economic organization: The analysis of discrete structural alternatives. Administrative Science Quarterly, 36(2), 269–296. https://doi.org/10.2307/2393356

Williamson, O. E. (2009, December 8). Transaction cost economics: The natural progression [Prize lecture]. The Nobel Foundation. https://www.nobelprize.org/uploads/2018/06/williamson_lecture.pdf

Chapter 6 — Integration in Digital Businesses

When the asset is data, owning a stage means holding the records, the code and the customer relationship and renting the rest with a way out, so this chapter says what owning means for data, how a business owns its customer relationship, when a direct channel stands beside partners and resellers, who holds the buyer in a marketplace and in a reseller, how a business owns its firmware, its method and its product records, what to do with data that lives at its suppliers, when to build software and when to rent it, how lock-in builds and how to leave with one's records, and how personal information moves through an integrated chain, and it applies all of it to Hydrosentinelle inc., whose device records, alert service and rented hosting are posed by this book.

What You'll Be Able to Do

• Define owning data as a practical test — being able to read a record, to move it and to keep it without anyone's permission — and say where the boundary of a digital business falls.
• Distinguish the customer relationship from the channel that carries the product, and say what a business holds when it owns the line to the people who pay.
• Place a direct channel beside partners and resellers, and read a study of one industry at its own level rather than as a rule for another.
• Ask the five questions that decide who holds the buyer in a marketplace or a reseller, and write the answers in the register of decisions.
• Choose between software the business builds and software it rents, and write the export into every rental agreement.
• Call the Québec privacy rules to Book H2 (Privacy and Personal Information) and keep personal information out of the decisions this chapter makes.

Why It Matters

In a workshop business, the boundary of a firm is visible from the street. A workshop, a machine, a stock of parts and a payroll are the firm, and what stands outside them is bought. In a business whose product carries firmware, whose service runs on a hosted platform and whose customer relationship lives in a billing record, the boundary is drawn around records instead, and records are harder to see. A business can own every machine in its workshop and still rent the only thing that makes the machines worth running, because the record of what each unit is, what it was tested for and who bought it sits inside somebody else's system.

The practical consequence is that the question of what to own has a second half that the earlier chapters of this book did not need. Chapters 3 to 5 asked which activity to keep, what moves when an activity leaves and what a hybrid form holds. This chapter asks a narrower and more modern question: when the thing that crosses the boundary is not a part but a record, who holds it, in what form does it come back, and what does the business keep if it decides to leave.

The cost of getting this wrong is quiet. Nothing breaks on the day a record is kept only inside a supplier's system; the arrangement is convenient, and convenience is the reason it was chosen. What breaks is the day the business wants to move, and the export does not exist, or exists in a form no other system can read. The case of this book, Hydrosentinelle inc., a fictional company in Sherbrooke that designs and sells connected water-leak detectors for commercial and rental buildings, holds its firmware, its alert logic and its device records, rents its hosting with a written export, and has the test records of each supplier-built kit sent back every day under an agreement. Every figure of that arrangement is posed by this book and illustrative, and no result of the company is credited to it.

In This Chapter

Section 6.1 defines what owning means when the asset is data, sets the boundary of a digital business where the records can leave with it, and opens the chapter with the table of frontier that says where this chapter stops and its neighbors begin. Section 6.2 covers owning the customer relationship, the direct line between a business and the people who pay for its product and use it. Section 6.3 places a direct channel beside partners and resellers and reads two sources at their own level. Section 6.4 asks who holds the buyer in a marketplace or a reseller. Section 6.5 covers the firmware, the test method and the record of each unit made. Section 6.6 covers the records a supplier creates for the business and the return named in the agreement. Section 6.7 covers software the business builds against software it rents. Section 6.8 covers lock-in and the right to leave with one's records. Section 6.9 covers personal information in an integrated chain and calls the Québec privacy rules to Book H2 (Privacy and Personal Information). Section 6.10 applies all of it to the device records and the alert service of Hydrosentinelle inc.

6.1 What Owning Means When the Asset Is Data

In a business whose product carries firmware, whose service runs on a hosted platform and whose customer relationship lives in a billing record, the boundary is drawn around records, and a record is a different kind of asset from a machine or a stock of parts. It can be copied at no cost, it can be moved across a border in a second, and it can sit inside a system that the business does not control while looking, from the outside, exactly like a record the business owns.

This book uses a practical test rather than a legal definition, and the test is its own. Owning data, in this book, means being able to read a record, to move it and to keep it without anyone's permission. The test has three parts and all three must pass. A business that can read a record but cannot move it holds a view, not an asset. A business that can move a record but cannot keep it after the agreement ends holds a loan. A business that can do all three holds something it can build on, sell, insure and hand to a successor. The test is deliberately plain, because the alternative — a definition built on the law of property, on the law of contracts or on the law of intellectual property — belongs to other books, and this chapter needs a test an owner can apply in an afternoon.

The boundary of a digital business, then, falls where the records can leave with it. That sentence is this book's own reading and not a finding of any source. It says that the question of vertical integration in a digital business is not only which activities to keep and which to buy, but which records to hold and which to rent, and that the two questions are answered together. A business that keeps an activity in-house and rents the record of that activity has kept the work and rented the memory of it. A business that buys an activity and holds the record of it has bought the work and kept the memory, which can be the better trade, because the memory is what the next decision is made from.

Two assets of the case illustrate the test. The firmware of the sensors and of the gateway, and the mobile application, are assets of the company, printed by Book H5 (Intellectual Property, chapter 1). The testing of the gateway firmware and of modules of the mobile application is rented under control, printed by Book I8 (Freelancers, Agencies and Outsourcing, chapter 2). The company therefore owns the code and rents a test of the code, and the record of what was tested and what the result was stays with the company. That arrangement passes the three-part test: the company can read the test result, move it into its own records and keep it after the testing agreement ends.

The table below sets out where this chapter stops and its neighbors begin, one row per neighbor, with what the neighbor holds and what this chapter holds. Read it as a map of the shelf rather than as a list of priorities: each row names a subject that another book owns and states, in the third column, the narrower thing this chapter does with it. Retain the shape of the division — this chapter holds the ownership of the code, the product records and the customer relationship in a chain of making, and it cites the rest rather than rewriting it.

Table 6.1 — Where this chapter stops and its neighbors begin

The table is a division of labor and not a ranking: the other books hold the platform, the marketplace, the purchase of software, the core systems, the governance of data, personal information and the firmware as an asset, and this chapter holds the ownership of the records that a chain of making produces.

DEFINITION The words this book uses for owning data, a device record, the customer relationship, lock-in and an export

These are this book's working words, and no source read defines them in this sense. Owning data is being able to read a record, to move it and to keep it without anyone's permission. A device record is the record of one unit made, with its serial number, its firmware version and its test result. The customer relationship is the direct line between a business and the people who pay for its product and use it. Lock-in is the cost of leaving a supplier of software or data. An export is a copy of one's records in a form that another system can read. The test of owning data is practical and not a legal definition, and a record kept only inside a supplier's system is rented rather than owned. Personal information is Book H2's (Privacy and Personal Information, chapter 8) and platforms are Book A9's (Platform Strategy: Integrating with Dominant Offers and Partner Ecosystems), and neither is redefined here.

6.2 Owning the Customer Relationship

The customer relationship is the direct line between a business and the people who pay for its product and use it. A channel can carry the product and keep the line, and a channel can carry the product and take the line away. The difference is not visible in a sales report, because both arrangements produce revenue; it is visible in a list of names, and in who holds it. Owning the line means holding the contact, the billing and the history of the customer: who the customer is, what the customer pays, what the customer has bought before, and what the customer has asked for. A business that holds those three things can decide what to offer next, can price it, and can keep the customer through a change of channel. A business that holds none of them has a supplier of customers rather than a customer relationship.

The case holds its line. A building pays $89, $129 or $179 a month on Essential, Plus or Portfolio, printed by Book L5 (Subscription and Recurring Revenue Models, chapter 2), and the subscription is billed by the company. The billing is the part that matters for this chapter. A subscription billed by the company produces a record of who pays, how much and since when, and that record sits with the company rather than with a channel. The company can therefore read its own revenue by building, by tier and by month without asking anyone's permission, which is the three-part test of section 6.1 applied to the customer relationship rather than to a device record.

The research read for this chapter is a conceptual article and is used at its own level. The abstract of the article reports that four types of digital intermediation and disintermediation affect branding: transaction intermediation tied to e-commerce retailers, transaction disintermediation tied to direct-to-consumer models, marketing intermediation powered by crowdsourcing, and marketing disintermediation tied to consumer-to-consumer models, with issues for future research (Gielens & Steenkamp, 2019, abstract only). The article is read at abstract only, it is a conceptual article, and the abstract reports no measured effect. What it offers this chapter is a vocabulary for the four positions a business can occupy between itself and its customer, and nothing more. It does not say which position is better, it does not measure the effect of any of them on a brand, and it does not tell a reader which channel to use.

The vocabulary is still useful, because it separates two things that can be confused. Transaction intermediation is about who completes the sale; marketing intermediation is about who carries the message. A business can complete its own sales and still have its message carried by others, and it can have its sales completed by a retailer while carrying its own message. The four types name the combinations, and the practical question for an owner is which combination the business is in today and which one it wants. The answer is written down, because a position that is not written down drifts: a business adds a channel for a season, the channel keeps the buyer's name, and later the business cannot say who its customers are.

The customer relationship is also a part of a digital business that can be costly to rebuild, in this book's reading: a firmware can be rewritten, a hosting provider can be replaced, and a device record can be exported and re-imported, but a relationship handed to a channel has to be won back. This is why the case keeps its billing, and why the chapter returns to the question in section 6.4, where the buyer's identity is the thing at stake.

KEY CONCEPT Own the code, the records and the customer, and rent the rest with a way out

This is the central idea of the chapter and it is this book's own reading, never a finding of a source. A digital business keeps what makes the product and the line to the people who pay for it: the code that runs on the product, the method that tests it, the record of each unit made, and the contact, the billing and the history of the customer. It rents everything else — hosting, messages, testing, storage — and it rents with a written export, so that each rental can end without ending the business. The three columns of Figure 6.1 in section 6.10 show the arrangement for the case: what the company owns, what it rents, and what lives at its suppliers and comes back. The idea is a working principle and not a law: a business can rent a record and still be sound, if the rental is priced well and the export is real. What the idea rules out is the arrangement in which the business cannot say who its customers are, cannot read the record of its own product, and cannot leave.

WHAT CHANGED A 2019 article names four types of digital intermediation and disintermediation, and reports no measured effect (Gielens & Steenkamp, 2019)

The dated change comes from the article read at abstract only on October 15, 2026: the abstract names four types of digital intermediation and disintermediation — transaction intermediation tied to e-commerce retailers, transaction disintermediation tied to direct-to-consumer models, marketing intermediation powered by crowdsourcing, and marketing disintermediation tied to consumer-to-consumer models — and sets out issues for future research (Gielens & Steenkamp, 2019, abstract only). The four types are the authors' conceptual frame, and the abstract reports no measured effect of any of them. The article does not tell a reader which channel to use, and it does not measure what a channel does to a brand or to a margin. What it changes for this chapter is the vocabulary: a business can now say whether it is intermediated on the transaction, on the message, on both or on neither, and it can write that position down. The date is the article's own, and it is a date of a source, never a date of the case.

6.3 Direct Channels Beside Partners and Resellers

A direct channel stands beside partners and resellers when a business wants to keep the customer relationship for part of its sales while letting others carry the rest. The arrangement is not a compromise; it is a choice about which sales the business wants to make itself. A partner can reach a market the business cannot reach, can carry the product into a catalog, can install it and can invoice for it, and in exchange the partner holds some part of the relationship. A direct channel costs the business the selling work that a partner would do: the prospecting, the demonstration, the quote, the negotiation and the follow-up. The trade is between reach and the line, and it is made sale by sale rather than once for the whole business.

The two sources read for this section are used at their own level and no further. The first is the conceptual article recalled from section 6.2, whose abstract names four types of digital intermediation and disintermediation and reports no measured effect (Gielens & Steenkamp, 2019, abstract only). The second is a study of one industry, read at abstract only: for United States semiconductor companies entering foreign markets, choosing between captive agents and independent intermediaries is a question of downstream vertical integration; integration is associated with the transaction specificity of assets in the distribution function and with a highly differentiated product, and a product tends to be sold through the channel already in place (Anderson & Coughlan, 1987, abstract only). The study is one industry and one period, the abstract reports an association and not a cause, and no finding outside that industry is drawn from it here.

The table below sets the two readings side by side, with the setting of each, what its abstract reports and what it does not support. Read it as a guard against taking a finding from one industry and applying it to another. Retain the last column: it is the part of each reading that a business must not use.

Table 6.2 — Two readings on intermediation and channels, at the level read

What the two readings leave a business with is a way of asking rather than an answer. The first says that a business can be intermediated on the transaction, on the message, on both or on neither, and that the position is a choice. The second says that in one industry, at one time, the choice between a captive agent and an independent intermediary was associated with how specific the assets in distribution were and with how differentiated the product was, and that a product tended to be sold through the channel already in place. Neither says what a maker of detectors should do. The business decides on its own channel and its own numbers, and it writes the decision down.

Two subjects are cited here and not retold. Distributors and wholesale channels are Book F3's (Distributors and Wholesale Channels, chapter 2). Marketplaces are Book F5's (Marketplaces: Selling Through Distributor, Retailer and App Marketplaces, chapter 2). What this chapter holds is the question that comes before either of them: whether the business keeps the buyer, and if so for which sales. The case draws no distributor and no marketplace for its kits; the subscription is billed by the company, printed by Book L5 (Subscription and Recurring Revenue Models, chapter 2), and the certified installers of Book F9 (Building and Running a Partner Program, chapter 4) install beside the staff installer without holding the buyer. The installers' program is Book F9's and is not retold here.

THE RULE A study is printed in its setting, and semiconductor companies abroad are not a rule for a maker of detectors (Anderson & Coughlan, 1987)

The rule is the house rule of this book: a study is printed with its industry, its period and its sample, and it is never turned into a rule for a reader. The abstract of the study read on October 15, 2026 reports that for United States semiconductor companies entering foreign markets, integration of distribution is associated with the transaction specificity of assets in the distribution function and with a highly differentiated product, and that a product tends to be sold through the channel already in place (Anderson & Coughlan, 1987, abstract only). An association in one industry is not a rule for a maker of detectors, and the abstract reports an association rather than a cause. The business decides on its own channel and its own numbers, and it writes the decision in the register of decisions of section 6.4. No figure of that study is printed here, and no other finding is drawn from it.

6.4 Marketplace or Reseller: Who Holds the Buyer

The choice between selling through a marketplace and selling through a reseller is a choice about who holds the buyer, and the choice is made before the channel is chosen. The reasoning of this section is this book's own and rests on no source. The choice itself belongs to Book F5 (Marketplaces: Selling Through Distributor, Retailer and App Marketplaces, chapter 2), and the dependence on a dominant offer belongs to Book A9 (Platform Strategy: Integrating with Dominant Offers and Partner Ecosystems). What this chapter holds is the question that decides the answer, and the question has five parts.

The first question is who sees the buyer's name and contact. A channel that shows the business the name, the address and the contact of the buyer leaves the line intact, even if the channel completes the sale. A channel that keeps the name and shows the business only an order number has taken the line, and the business has a supplier of customers rather than a customer relationship. The second question is who sets the price. A channel that sets the price sets the margin, and a business that cannot set its own price cannot decide what to offer next. The third question is who handles returns and support. A channel that handles support holds the conversation with the customer, and the conversation is where the next sale is made. The fourth question is who owns the data of the sale: the order, the address, the history and the record of what was bought. The fifth question is what the business keeps if it leaves. A channel that lets a business walk away with the names, the orders and the history has rented a storefront; a channel that keeps them has rented the business.

The five questions are asked before the channel is chosen, and the answers are written in the register of decisions. The register is the same one that Chapter 3 uses for make-or-buy decisions and Chapter 5 uses for hybrid forms, and it holds one line per decision with the reason, the date and the trigger that would reopen it. Writing the answers down matters because the answers change quietly. A channel that showed the buyer's name in the first year can stop showing it in the third, and a business that never wrote down what it was promised has no way to notice. The register is also what makes the decision reviewable at the yearly review of Chapter 8, where every activity is read again.

A channel that keeps the buyer's identity is a rented relationship, and renting a relationship is not wrong. A business that reaches a market it could not reach alone, at a price that covers the rent, has made a sound trade. What is wrong is making the trade without knowing it was made. The case draws no marketplace and no reseller for its kits, and the subscription is billed by the company itself, printed by Book L5 (Subscription and Recurring Revenue Models, chapter 2). The company therefore holds the buyer for every kit it sells, and the question of this section is answered by default rather than by a decision. A business that does sell through a marketplace or a reseller should write the five answers down, and should read them again at each review of the agreement.

6.5 Owning the Firmware, the Method and the Product Records

Three things stay with a maker even when others build for it: the code that runs on the product, the method that tests it, and the record of each unit made. The reasoning of this section is this book's own. The firmware as an asset and trade secrets are Book H5's (Intellectual Property, chapters 1 and 5), and the buying of work is Book I8's (Freelancers, Agencies and Outsourcing, chapter 2), and both are cited here rather than retold.

The code that runs on the product is the part of a modern product that carries its behavior. A detector that reports a leak within minutes, printed by Book A7 (Pricing Strategy and Revenue Models, chapter 2), does so because of the code that reads the sensor and decides what to send, and that code is the product's behavior rather than its housing. A business that buys its boards and has its kits assembled can still own the code, and the case does: the firmware of the sensors and of the gateway, and the mobile application, are assets of the company, printed by Book H5 (Intellectual Property, chapter 1). The code is supplied to the contract manufacturer as an image, and the contract manufacturer flashes the image that the company supplies. The company therefore owns the behavior of its product while buying the assembly of it, which is the arrangement this book calls owning the code and renting the hands.

The method that tests the product is the second thing that stays. A test method is the set of steps, limits and instruments that decide whether a unit is good, and it is the part of quality that a business cannot buy back once it has been handed over. The case keeps its method: the contract manufacturer runs the company's end-of-line test on the duplicate test bench, and the duplicate bench is the company's, posed by this book. The test protocol and the duplicate bench are what let the company read a first-pass yield of 98.6 percent at day 365, posed by this book, from records the company holds rather than from a report the contract manufacturer writes. Quality management is Book J4's (Quality Management and Continuous Improvement), and this chapter cites it rather than restating it.

The record of each unit made is the third thing that stays. A device record is the record of one unit made, with its serial number, its firmware version and its test result, and it is what makes a defect traceable, a warranty arguable and a recall bounded. The case keeps its device records, posed by this book, and the end-of-line test records of each supplier-built kit are sent back every day under the supply agreement, posed by this book. The return is written into the agreements of days 57 and 58, which Chapter 5 covers, and the form and the day of the return are the subject of section 6.6. A business that buys assembly and does not get the test records back has bought the units and rented the knowledge of what they are.

WHO DOES THIS The owner owns the records, and one person holds the export and the firmware as the business grows

The roles below are this book's suggestion for a business of each size, and the case's arrangement is posed by this book. A solo owner owns the records and keeps the copies, in a form that can be read without the tool that made them. In a small team one person holds the export and the owner reads it, so that the export is tested by someone other than the person who wrote it. In an established business a data lead owns the register of records and a product lead owns the firmware, and Chapter 9 covers who helps with that work and when to get help. In the case, posed by this book, Laurence Bergeron, the president for product and engineering, holds the firmware image, the test protocol and the structure of the device records. No hire is made and no amount is printed for that work. No person of the case takes a pronoun in this book.

6.6 Data That Lives at Your Suppliers

A supplier creates records for the business, and the records are easy to forget because they are made somewhere else. A contract manufacturer creates the end-of-line test record of each unit it builds. A hosting provider creates the logs of the service it runs. A molder creates the dimensions and the drawings of the molds it holds. A testing provider creates the results of the tests it runs. Each of those records is a record of the business's own product or service, made by somebody else, and each one is worth having back. The reasoning of this section is this book's own, and the law it touches is called rather than rewritten.

The law is the Act respecting the protection of personal information in the private sector, CQLR c P-39.1, ss. 17, 18.3 and 27, LégisQuébec, up to date to August 12, 2026, read on October 15, 2026, and it is called to Book H2 (Privacy and Personal Information). Section 17 provides that before communicating personal information outside Québec, or entrusting a person outside Québec with collecting, using, communicating or keeping it, a privacy impact assessment and a written agreement are required. Section 18.3 provides that personal information may be communicated without consent to a mandatary or a person performing a contract of enterprise or for services, if the contract is in writing and specifies the measures that protect confidentiality, limit the use to the contract and prevent the information from being kept after it ends, and that person must notify the person in charge of any violation without delay. Whether a given flow of personal information complies with either provision is a question for Book H2 and a lawyer, and this book decides nothing about it.

A record that carries personal information is Book H2's, and this chapter sends it there. A telephone number or an email address used to send an alert can be personal information, and so can the contact of a building owner; whether it is, in a given case, is Book H2's question. A record that carries no personal information — a test result, a firmware version, a drawing — is a record of the business's own product, and this chapter holds the question of getting it back. The distinction matters because the two kinds of record travel under different rules and are written into an agreement in different clauses. A business that mixes them writes one clause where two are needed, and the clause that protects the product record is the one that gets lost.

Every record the business needs must be named in the agreement with the day and the form in which it comes back. Naming the record is not enough; the form matters, because a test result delivered as a picture of a screen cannot be read by another system, and the day matters, because a record delivered at the end of the agreement is a record delivered too late to act on. The case writes the return into the agreements of days 57 and 58, which Chapter 5 covers, and the end-of-line test records of each supplier-built kit are sent back every day, posed by this book. The form is a file the company can read, and the day is the day of the build.

METHOD Getting back the records a supplier creates for you in five steps

First, list the records that the supplier creates for the business, one line per record, with the supplier that makes it. Second, mark the ones that carry personal information and send them to Book H2 (Privacy and Personal Information, chapter 8), because those travel under different rules. Third, write for each remaining record the owner inside the business, the form in which it comes back and the day it comes back. Fourth, test the return once with a real record, on an ordinary day, and read the file in the system that will use it. Fifth, read the list again at each review of the agreement, because a supplier adds a record and drops one without telling anyone. The sequence is this book's own, no source tests it, and the case's return is posed by this book. The method produces a list, not a contract; the contract is Book H1's (Contracts and Commercial Agreements) and a lawyer reads it.

6.7 Software You Build Versus Software You Rent

Software the business builds is an asset the business must maintain, and software the business rents is a service the business must be able to leave. The reasoning of this section is this book's own. The buying of software is Book K1's (The SME Technology Stack, chapter 4), and core systems and their modules are Book K2's (CRM and ERP Systems, chapter 2), and both are cited here rather than retold. What this chapter holds is the same choice read as a question about ownership rather than as a purchase.

The choice follows the grid of Chapter 3, which scores an activity on core, capability, specificity, cost, risk and control and gives a verdict. Software that carries the product's behavior scores high on core and on specificity, and it is built. Software that runs a common function — hosting, messaging, accounting, a help desk — scores low on core and on specificity, and it is rented. The grid is the same grid the case uses for its six activities, and the hosting of the monitoring service scores as a buy on it, posed by this book. The verdict of the grid is not a preference for building; it is a reading of how much the activity matters to the product and how hard it would be to replace.

What is rented is rented with a written export. An export is a copy of one's records in a form that another system can read, and it is the clause that turns a rental into a rental rather than a trap. The export clause names what leaves, in what form and how often, and it is tested before it is needed. A business that rents a service without an export clause has rented the service and the records together, and the day it wants to leave it will find that the records are not its to take. The case writes the export into the hosting agreement, posed by this book, and the hosting of the monitoring service is rented with an export of the company's data written into the agreement.

The two arrangements are not exclusive, and a digital business can run both at once. The case builds its firmware, its alert logic and its mobile application, printed by Book H5 (Intellectual Property, chapter 1), and rents the hosting of the monitoring service and the testing of the gateway firmware and of modules of the mobile application, printed by Book I8 (Freelancers, Agencies and Outsourcing, chapter 2). The company therefore owns the behavior of its product and rents the machines that run it and the hands that test it. That arrangement is sound only because the export is written and the test records come back.

TOOLS OF THE TRADE What a business needs to own its software and its records: a list, stored agreements, an export calendar and one register

Four functions carry this work, and a business should look for them rather than for a product. The first is a list of the software the business builds and the software it rents, with an owner and a date for each line, so that the boundary of the digital business is written down in one place. The second is a stored copy of each rental agreement with its export clause, kept where the person who needs it can find it. The third is a calendar of the dates when an export is tested, because an export that has never been tested is a promise rather than a capability. The fourth is one register of decisions, holding the choice to build or to rent, the reason, the date and the trigger that would reopen it. Chapter 10 gives the functions in full. No brand name belongs on this list and no price; the requirement is the function.

6.8 Lock-In and the Right to Leave With Your Records

Lock-in is the cost of leaving a supplier of software or data, and it builds one convenience at a time. The reasoning of this section is this book's own. A record is kept only inside a supplier's system because it is easy to keep it there, and the ease is real: the record is written once, it is visible to everyone who needs it, and nobody has to copy it anywhere. The second convenience is a report that reads the record in place. The third is a workflow that depends on the report. By the time the business wants to leave, the record, the report and the workflow are one thing, and the export does not exist. Nothing in that sequence is a mistake at the moment it is made, and the sum of it is a business that cannot leave.

The law is called here and not rewritten. The Act respecting the protection of personal information in the private sector, CQLR c P-39.1, s. 27, LégisQuébec, up to date to August 12, 2026, read on October 15, 2026, provides for access on request, and for computerized personal information collected from the applicant, and not created or inferred from it, to be communicated at the applicant's request in a structured, commonly used technological format, unless doing so raises serious practical difficulties. The provision is called to Book H2 (Privacy and Personal Information), and this book never writes the format without its two limits.

Section 27 speaks of a person's information, and the format it provides covers computerized personal information collected from the applicant and not created or inferred from it, unless doing so raises serious practical difficulties; the right of a business to leave with its own records comes from its agreement rather than from that provision. An agreement that names the export, the form and the day is the instrument that makes leaving possible, and it is Book H1's (Contracts and Commercial Agreements) and a lawyer reads it. This chapter does not interpret a clause and does not say whether a given agreement gives a business the right to leave; it says only that the right is written into the agreement or it is not there.

The way out is an export in a form another system can read, tested before it is needed. Testing matters because an export that has never been run is an assumption. A business that runs its export once a year, on an ordinary day, into a system that is not the supplier's, knows three things it would otherwise guess: that the export exists, that it contains what the agreement says it contains, and that the business can read it. The test is cheap and it is the only thing that turns a clause into a capability.

The platform described in this book offers one example of leaving on the user's side. In the application copy of September 7, 2026, Account → Profile → Privacy (account/privacy) exports the account's own data as a file and deletes the account after a typed confirmation phrase (La Machine, 2026). It exports none of the data of a contact held in the customer relationship system, and Book H2 (Privacy and Personal Information) holds personal information. The page is an example of a way out and it is not the reader's export of a supplier's records; the case does not use the platform.

WATCH OUT A convenience becomes a condition when a record is kept only in a supplier's system

The trap is quiet. A record is kept only in a supplier's system because it is easy to keep it there, and the day the business wants to leave, the export does not exist. The signs are two: a record that nobody has ever exported, and an agreement that is silent on the export. Both signs are visible from inside the business, and neither requires a lawyer to see. The remedy is to write the export into the agreement, with the form and the day, and to test it once a year on an ordinary day. The reasoning is this book's own and no source measures how often the trap is sprung. The trap is not a reason to avoid renting; renting can be the right choice, and a rental with a tested export is a rental a business can leave. What the trap punishes is the rental whose export was assumed rather than written.

6.9 Personal Information in an Integrated Chain

Personal information moves through an integrated chain in several places. A telephone number or an email address used to send an alert is personal information. The contact of a building owner is personal information. A record that a supplier holds on the business's behalf, and that names a person, is personal information held by a supplier. The reasoning of this section is this book's own, and the rules it touches are called rather than rewritten. Personal information and its transfer to third parties are Book H2's (Privacy and Personal Information, chapter 8), and this section never teaches them.

The three provisions read for this chapter are set out in the table below, each with its jurisdiction and its date, what it says in this book's words and the book it is called to. Read the table as a map of the provisions this chapter touches and not as a summary of the law: it names three provisions of one Québec statute, it does not cover the federal statute or the statutes of other jurisdictions, and it does not say whether any given flow complies. Retain the last column, because each provision is called to Book H2 rather than applied here.

Table 6.3 — The privacy provisions called to Book H2, with their date

The three provisions are read in the Act respecting the protection of personal information in the private sector, CQLR c P-39.1, ss. 17, 18.3 and 27, LégisQuébec, up to date to August 12, 2026, read on October 15, 2026, and each is called to Book H2 (Privacy and Personal Information). Section 17 and section 18.3 govern what happens before personal information is entrusted to a provider.

Whether a given flow of personal information complies with any of the three provisions is a question for Book H2 and a lawyer, and this book decides nothing about it. What this chapter does is place the question where it belongs in a chain. A business that buys assembly, rents hosting, rents testing and holds molds at a molder has four places where personal information can sit, and each of them is written into an agreement or it is not. The agreement is Book H1's (Contracts and Commercial Agreements) and a lawyer reads it; the privacy rules are Book H2's and this chapter calls them.

The practical consequence for the register of decisions is that a line about a supplier carries a second question beside the cost and the control: does this supplier hold personal information for the business, and if so, under what agreement. A business that answers that question when the supplier is chosen has one clause to write. A business that answers it later has a flow to unwind, and unwinding a flow is harder than writing a clause.

QUÉBEC & CANADA NOTE Personal information in a chain is held by Book H2, and this book explains the law as published and is not legal advice

This book explains the law as published and is not legal advice. The Québec point of this section is personal information entrusted to a provider. The Act respecting the protection of personal information in the private sector, CQLR c P-39.1, ss. 17, 18.3 and 27, LégisQuébec, up to date to August 12, 2026, read on October 15, 2026, requires a privacy impact assessment and a written agreement before personal information goes outside Québec or a person outside Québec is entrusted with it, sets the conditions under which personal information may be communicated without consent to a person performing a contract of enterprise or for services, and provides for access on request and for a structured, commonly used technological format for computerized personal information collected from the applicant and not created or inferred, unless serious practical difficulties arise. Both limits of that format are part of the provision. Book H2 (Privacy and Personal Information) holds these rules and a lawyer reads a given agreement. This book recommends no vendor and no adviser, and it decides nothing about the reader's own case.

6.10 The Case's Device Records and Alert Service

The digital layer of Hydrosentinelle inc. is printed in part by published books and posed in part by this book, and the section says which is which. The company designs and sells connected water-leak detectors for commercial and rental buildings, and its digital layer is what makes a detector a service: the code that reads the sensor, the rule that decides what to send, the record of each unit made, the customer relationship and the subscription, the hosting that runs the service, and the messages that carry an alert. Some of those the company owns, some it rents, and some live at its suppliers and come back.

Part of this layer is printed by published books: the firmware and the alert logic are Book H5's (Intellectual Property, chapter 1), and the customer relationship and the subscription are Book L5's (Subscription and Recurring Revenue Models, chapter 2), a building paying $89, $129 or $179 a month on Essential, Plus or Portfolio, billed by the company. The rest is posed by this book: the device records, the rule that every alert leaves directly by text message and email, the rented hosting and message providers, and the records that live at the suppliers and come back. The firmware and the alert logic stay in-house, and the return of the records is written into the agreements of days 57 and 58, which Chapter 5 covers.

Figure 6.1 sorts the digital layer of the case's chain into what the company owns, what it rents and what lives at its suppliers and comes back.

Figure 6.1 — Own the code, the records and the customer; rent the hosting. The return of each record held at a supplier is written into the agreement (posed).

The figure is read column by column. The first column holds what the company owns: the firmware and the alert logic, printed by Book H5; the customer relationship and the subscription billed by the company, printed by Book L5; the device records, with the serial number, the firmware version and the test result of each kit, posed by this book; and the rule that every alert leaves directly by text message and email, posed by this book. The second column holds what the company rents: the hosting of the monitoring service, with an export of the company's data written into the agreement, and two message providers, so that one can fail without silencing an alert, both posed by this book. The third column holds what lives at the company's suppliers and comes back: the end-of-line test records of each supplier-built kit, sent back every day and owned by the company under the supply agreement, and the dimensions and drawings of the molds, owned with the molds under the tooling agreement, both posed by this book. The note under the figure says what the chapter has said twice: the return of the records is written into the agreement.

The table below sets the seven items of the digital layer beside whether each is owned, rented or returned, where it sits and where it comes from. Read it as the register of the digital boundary of the case, one line per item, with the source of each line in the last column. Retain the last column: it is what separates the facts printed by published books from the facts posed by this book, and no line of the table is a measured result of the company.

Table 6.4 — The case's digital layer: owned, rented and returned, with where each comes from

The platform described in this book is one worked example of a business platform, and the case does not use it. In La Machine, as of the application copy of September 7, 2026, the pages that touch this chapter are Control center → Tools (inventory) (/outils), an inventory of the installation's software tools with a measured version and date; Business → Store → Products (business/shop), whose product page records a product's cost beside its price, with Offers → Store (client/store) opening the external link where the customer is sent; and Account → Profile → Privacy (account/privacy), which exports the account's own data as a file (La Machine, 2026). It holds no register of tooling and no traceability of lots, and its export covers none of the data of a contact held in the customer relationship system; check it in your installation. Every fact holds in the application copy of September 7, 2026, read in its code on October 15, 2026. This book is valid with any tool, and La Machine is described and never sold.

CASE IN POINT Hydrosentinelle inc. owns its firmware, its device records and its customers, and rents its hosting with a written export

Hydrosentinelle inc. is a fictional business created for this encyclopedia; its founders, its employees, its customers, its partners, its suppliers, its figures and the events described are invented, and any resemblance to a real business is coincidental. The company owns its firmware and its alert logic, printed by Book H5 (Intellectual Property, chapter 1), its customer relationship and its subscription, printed by Book L5 (Subscription and Recurring Revenue Models, chapter 2), its device records and the rule that every alert leaves directly by text message and email, posed by this book. It rents the hosting of the monitoring service with an export of the company's data written into the agreement, and two message providers, posed by this book. The end-of-line test records of each supplier-built kit and the dimensions and drawings of the molds live at its suppliers and come back, posed by this book, and the return is written into the agreements of days 57 and 58, which Chapter 5 covers. Laurence Bergeron, the president for product and engineering, holds the firmware image, the test protocol and the structure of the device records, posed by this book. No result of the company is credited to any of it.

DO THE MATH Two hundred thirty kits, one record each: what the company holds when assembly is bought

Illustrative figures; the case is fictional. Step 1 — The forecast of the year is 230 kits, of which 207 are built by the contract manufacturer and 23 in-house, and 207 plus 23 are 230. Step 2 — Each kit has one device record with three items, a serial number, a firmware version and a test result. Step 3 — The 207 supplier-built kits are 90 percent of the 230, and their end-of-line test records are sent back every day under the supply agreement. Step 4 — The 23 in-house kits are 10 percent, and their records are written by the staff installer. Step 5 — The trigger of a defect rate above 2 percent for 2 months running can be read only from records that the company holds, and every count is posed by this book and illustrative. The sums are exact, and no figure here is a result of the company.

IN LA MACHINE La Machine exports the account's own data and lists software tools, and it holds no register of tooling and no traceability of lots

In the application copy of September 7, 2026, Control center → Tools (inventory) (/outils) lists the installation's tools with an identity measured by a script — a version, a location and a date — and it can serve as a model for an inventory of what a business depends on, for software only. Business → Store → Products (business/shop) records a product's cost beside its price, and Offers → Store (client/store) opens the external link where the customer is sent, with no cart and no payment in the platform. Account → Profile → Privacy (account/privacy) exports the account's own data as a file and none of the data of a contact held in the customer relationship system. The platform holds no register of tooling and no traceability of lots, and its export is not an export of a supplier's records; check it in your installation. The case does not use La Machine, and La Machine is described and never sold.

Costly Mistakes

• Keeping a record only inside a supplier's system. The arrangement is convenient and it works, until the day the business wants to leave and the export does not exist; the remedy is to write the export into the agreement and to test it once a year.
• Confusing the channel with the customer relationship. A channel can carry the product and keep the line, and a channel can carry the product and take the line away, and the difference is visible only in a list of names and in who holds it.
• Choosing a marketplace or a reseller before asking who holds the buyer. The five questions of section 6.4 are asked before the channel is chosen, and a channel that keeps the buyer's identity is a rented relationship whether or not the business noticed.
• Reading a study of one industry as a rule. The abstract of the study of United States semiconductor companies entering foreign markets reports an association in that industry and that period, and it is not a rule for a maker of detectors (Anderson & Coughlan, 1987, abstract only).
• Letting the test method leave with the work. A business that buys assembly and does not keep its own test protocol and its own bench has bought the units and rented the knowledge of what they are.
• Mixing personal information with product records in one clause. The two kinds of record travel under different rules, and a single clause written for both protects neither; personal information is Book H2's (Privacy and Personal Information, chapter 8).
• Writing the export into the agreement and never testing it. An export that has never been run is an assumption rather than a capability, and the only way to know that the business can leave is to leave once, on an ordinary day, into a system that is not the supplier's.

Key Takeaways

• Owning data is a practical test and not a legal definition: a business owns a record when it can read it, move it and keep it without anyone's permission, and the boundary of a digital business falls where the records can leave with it.
• The customer relationship is the direct line between a business and the people who pay for its product and use it, and owning the line means holding the contact, the billing and the history of the customer; the case bills its own subscriptions, printed by Book L5 (Subscription and Recurring Revenue Models, chapter 2).
• A direct channel stands beside partners and resellers when a business wants to keep the customer relationship for part of its sales, and the choice costs the selling work a partner would do; the study read for this chapter is of United States semiconductor companies entering foreign markets, one industry and one period, and its abstract reports an association rather than a cause (Anderson & Coughlan, 1987, abstract only).
• The question of who holds the buyer in a marketplace or a reseller has five parts — who sees the name and contact, who sets the price, who handles returns and support, who owns the data of the sale, and what the business keeps if it leaves — and the answers are written in the register of decisions before the channel is chosen.
• Three things stay with a maker even when others build for it: the code that runs on the product, the method that tests it and the record of each unit made; the case keeps its firmware, its test protocol and its device records, and the end-of-line test records of each supplier-built kit come back every day, posed by this book.
• Software the business builds is an asset it must maintain and software it rents is a service it must be able to leave, and the way out is an export in a form another system can read, written into the agreement and tested once a year; personal information in a chain is Book H2's (Privacy and Personal Information, chapter 8), and this book explains the law as published and is not legal advice.

Your Action Plan

1. List the records your chain produces and mark who holds each one. Write one line per record — the device record, the test record, the customer record, the drawings, the logs — with the owner inside the business, the system it sits in and whether the business can read it, move it and keep it without anyone's permission. Deliverable: a written list of the records of the chain with an owner for each. Verification: every record passes the three-part test of section 6.1, or it is marked as rented with a note of what would have to change.
2. Write the export clause into every rental agreement and put the test on a calendar. For each service the business rents, name what leaves, in what form and how often, and set a date once a year on which the export is run into a system that is not the supplier's. Deliverable: an export clause in each rental agreement and a dated export test on the calendar. Verification: the first export has been run and read, and the file contains what the clause says it contains.
3. Ask the five questions of every channel and write the answers down. For each marketplace, reseller, distributor or installer that carries the product, answer the five questions of section 6.4 and record the answers in the register of decisions with the date and the trigger that would reopen the line. Deliverable: one register line per channel, with the five answers. Verification: every channel has a written answer to all five questions, and the register is read at the yearly review of Chapter 8.
4. Keep the code, the method and the device records in-house and say so in the agreements. Confirm that the firmware image, the test protocol and the structure of the device record are held by the business, and write into each supply agreement the form and the day in which the supplier's records come back. Deliverable: a written statement of what the business owns and a return clause in each supply agreement. Verification: the return has been tested once with a real record, and the file has been read in the system that will use it.
5. Separate personal information from product records and send the first to Book H2. Mark every record that carries a name, a telephone number or an email address, and check that each flow of personal information to a provider sits under a written agreement; take the question to Book H2 (Privacy and Personal Information, chapter 8) and to a lawyer rather than deciding it in-house. Deliverable: a marked list of the records that carry personal information, with the agreement that covers each flow. Verification: every flow has a written agreement, and no flow is left to a clause written for a product record.

In La Machine

In the application copy of September 7, 2026, the pages that touch this chapter are Control center → Tools (inventory) (/outils), an inventory of the installation's software tools with a measured version and date, which can serve as a model for an inventory of what a business depends on, for software only; Business → Store → Products (business/shop), whose product page records a product's cost beside its price, with Offers → Store (client/store) opening the external link where the customer is sent; and Account → Profile → Privacy (account/privacy), which exports the account's own data as a file (La Machine, 2026). The people of the companies a business deals with can be recorded in CRM → Contacts (business/marketing/contacts), under the rules that Book H2 (Privacy and Personal Information) holds. The register of workflows, Control center → Workflows (/registre), shows one line for each workflow or agent route with its code, type, trigger, entry point and state, and only the state can be changed on the screen. Check it in your installation.

What La Machine does not do for the subject of this chapter is the larger part. It holds no register of tooling and no traceability of lots or serial numbers, so the device record of a unit made is kept elsewhere; it holds no supplier record, no supplier scorecard and no contract, so the return clause of a supply agreement lives in a contract repository the business keeps; its export covers the account's own data and none of the data of a contact held in the customer relationship system, so the export of a supplier's records is a clause in an agreement rather than a button; and it computes no total cost, no break-even volume and no make-or-buy verdict, so the choice to build or to rent is made on the grid of Chapter 3 and written in the register of decisions. The case does not use La Machine, this book is valid with any tool, and La Machine is described and never sold.

Self-Check

1. What does it mean to own data in this book, and where does the boundary of a digital business fall?
2. What is the customer relationship, and what does a business hold when it owns the line to the people who pay?
3. What does the study of United States semiconductor companies entering foreign markets report, and what does it not support?
4. What are the five questions that decide who holds the buyer in a marketplace or a reseller?
5. What are the two limits of the structured, commonly used technological format of section 27 — the format that covers computerized personal information collected from the applicant and not created or inferred from it, unless serious practical difficulties arise — and why are they always stated with it?

Answers. 1. Owning data, in this book, means being able to read a record, to move it and to keep it without anyone's permission, and the test is practical rather than a legal definition; the boundary of a digital business falls where the records can leave with it, so the question of what to own is answered together with the question of which records to hold and which to rent. 2. The customer relationship is the direct line between a business and the people who pay for its product and use it, and owning the line means holding the contact, the billing and the history of the customer; a channel can carry the product and keep the line, or carry the product and take the line away, and the difference is visible in a list of names and in who holds it. 3. The abstract reports that for United States semiconductor companies entering foreign markets, integration of distribution is associated with the transaction specificity of assets in the distribution function and with a highly differentiated product, and that a product tends to be sold through the channel already in place (Anderson & Coughlan, 1987, abstract only); it supports no finding outside that industry, it reports an association rather than a cause, and it is not a rule for a maker of detectors. 4. The five questions are who sees the buyer's name and contact, who sets the price, who handles returns and support, who owns the data of the sale, and what the business keeps if it leaves; they are asked before the channel is chosen and written in the register of decisions. 5. The format covers computerized personal information collected from the applicant and not created or inferred from it, and it does not apply where doing so raises serious practical difficulties; both limits are part of the provision and are always stated with it, and the provision is called to Book H2 (Privacy and Personal Information, chapter 8).

Chapter References

Act respecting the protection of personal information in the private sector, CQLR c P-39.1, ss. 17, 18.3 and 27 (Québec). LégisQuébec, up to date to August 12, 2026 — read October 15, 2026. https://www.legisquebec.gouv.qc.ca/en/document/cs/P-39.1

Anderson, E., & Coughlan, A. T. (1987). International market entry and expansion via independent or integrated channels of distribution. Journal of Marketing, 51(1), 71–82. https://doi.org/10.1177/002224298705100106

Gielens, K., & Steenkamp, J.-B. E. M. (2019). Branding in the era of digital (dis)intermediation. International Journal of Research in Marketing, 36(3), 367–384. https://doi.org/10.1016/j.ijresmar.2019.01.005

La Machine. (2026). Documentation — route reference inventory. Internal product documentation, read October 16, 2026, together with the code of the pages this book describes: /outils, /registre, business/marketing/contacts, business/shop, client/store and account/privacy.

Chapter 7 — Financial Analysis of Integration Decisions

A make-or-buy decision is a financial comparison before it is anything else, and it goes wrong when the comparison is not made on one basis, so this chapter says which costs belong in the decision, how a made unit and a bought unit are put on the same basis, what the capacity a business does not have costs, how an investment, its fixed cost and its break-even volume are read, how present value over the life of the asset is read, which margins an integration captures and which it gives up, how cash and the timing of an outlay differ from cost, how sensitivity to price, volume and lead time is read and what the purchase of a supplier or a customer adds, then applies it to three numbers of Hydrosentinelle inc. — a kit at $175.50 made and $202 bought, a bench that lacks 113 hours, and molds at $72,000 with a net present value of $3,068 at a posed rate of 9 percent — saying plainly that every figure of the case is posed by this book, that nothing is measured, that no source gives a typical saving or premium and that no saving or sale is credited to any decision.

What You'll Be Able to Do

• List the cost lines that belong in a make-or-buy comparison and the ones that stay the same whichever way the business decides.
• Write the cost of a made unit and the cost of a bought unit on one basis, with the same unit, the same period, the same volume and the hours valued at stated rates.
• Read the hours an activity needs against the hours that are free, and put a price on the waiting that a missing hour causes.
• Compute a break-even volume in words, as an investment divided by the saving on each unit, and read it beside the life of the asset.
• Bring the savings of an asset back to today at a stated rate, and read the result beside the break-even rather than alone.
• Test a decision under a different price, a different volume and a longer lead time, and say what the purchase of a supplier or a customer adds to the comparison.

Why It Matters

A comparison that is not made on one basis is not a comparison. The price a supplier quotes and the cost a business computes for itself are two different objects: the quote is a price for a unit delivered, and the computed cost is a set of hours, consumables and overheads that the business pays whether or not the unit is built. Put side by side without adjustment, the two numbers look comparable and are not, and the decision that follows is made on a difference that does not exist. The first work of this chapter is to put both sides on the same basis, and that work is arithmetic rather than judgment.

The second work is to read the comparison over the right time. An investment is paid once and its saving arrives unit by unit, so a difference per unit says nothing until it is set against the amount paid up front and the number of units that must pass before the saving covers it. A rate brings the later savings back to today, and the result is read beside the break-even rather than alone, because a thin result at one volume can turn negative at another. None of this is a forecast. It is a reading of what follows from the assumptions a business writes down, and the assumptions are the business's own.

The third work is to keep the comparison honest about capacity. A made unit assumes hours that exist. When the hours are missing, the cost of making is not the cost of the materials and the labor; it is the cost of waiting, and the waiting is paid in the fees that a late installation delays. That is why a business can correctly buy a unit that costs more than the unit it could make, and why the price alone never settles the question. This chapter prints no typical saving, no premium of a contract manufacturer and no hidden cost as a share of a price, because no source read measures any of them; the figures of the case are posed by this book, are illustrative, and are never a norm or a benchmark for any reader.

In This Chapter

Section 7.1 sets out which costs belong in the decision and which do not, with the boundary table of the chapter and the cost lines of a comparison. Section 7.2 puts a made unit and a bought unit on the same basis. Section 7.3 reads the cost of capacity a business does not have. Section 7.4 covers investment, fixed cost and the break-even volume. Section 7.5 covers present value over the life of the asset. Section 7.6 covers the margins an integration captures and the margins it gives up. Section 7.7 covers cash, financing and the timing of the outlay. Section 7.8 covers sensitivity to price, volume and lead time. Section 7.9 reads the purchase of a supplier or a customer. Section 7.10 puts the case's three numbers side by side and says what each one decides.

7.1 Which Costs Belong in the Decision

A make-or-buy comparison counts the costs that differ between the alternatives and leaves out the costs that stay the same whichever way the business decides. That single rule is what keeps a comparison honest, and it is the book's reasoning rather than a finding of any source read. The rent of a workshop, for a business that rents one, is paid whether the kits are assembled there or bought from a contract manufacturer, so it does not belong in the difference; the hours of the staff installer are paid only if the kits are assembled in-house, so they do. A cost that does not move with the decision is a cost that makes both sides look larger and does not change which side is larger. Writing it down anyway is a way of hiding the difference.

The wider idea has a name. Total cost of ownership looks beyond the price of a purchase to many other purchase-related costs, and the abstract of the article read for this book reports case studies of 11 firms, with dollar-based and value-based models used for supplier selection or supplier evaluation, a strong relationship between model type and use, and barriers and benefits of the approach (Ellram, 1995, abstract only). The article was read at its abstract only, on October 15, 2026, and the abstract reports no typical share of hidden costs; this chapter prints none. The practice page of the Crown corporation read for this book, read on October 15, 2026 and undated, is practice guidance and never an obligation, and it lists, for a piece of equipment, the total cost of ownership as the principal, floor space, financing interest, installation, training and the downtime associated with maintenance (Business Development Bank of Canada, n.d.). That list is the page's own guidance for equipment and is not a standard list of drivers for every purchase.

Two neighbors hold parts of this ground and are cited rather than retold. The full cost of bought work is Book I8 (Freelancers, Agencies and Outsourcing, chapter 2 of Book I8), and the hidden costs of moving an activity to a supplier are section 4.6 of this book. The purchasing of goods, the selection of a supplier and the holding of inventory lie outside this book, in Book J5 (Supply Chain, Procurement and Inventory).

In the case, the kit made in-house carries 3.1 hours of the staff installer with rework included, half an hour of the president and the consumables, and the kit bought carries the contract manufacturer's quote, the freight and the company's oversight, all posed by this book. The two sides are written as a cost per kit in section 7.2, and the difference is $26.50 in favor of making. Hydrosentinelle inc. buys 207 kits anyway, and the reason is in section 7.3.

The table below sets seven neighbors of this chapter beside what each holds and what this chapter holds, so that a reader can see where the arithmetic of the decision stops and where the neighboring books begin. Nothing of those books is rewritten here, and this chapter is the arithmetic of the decision and not the strategy, the financing, the valuation or the quality behind it.

Table 7.1 — The neighbors of this chapter and what each holds

The table below sets eight cost lines beside how each appears when the activity is made and when it is bought, and what to write down for each. It is the book's reasoning, illustrative and not a finding, and no source read gives a typical share of any line; it comes before the case's numbers of section 7.10 and is the skeleton those numbers fill.

Table 7.2 — The cost lines of a make-or-buy comparison

DEFINITION Total cost of ownership, relevant cost, break-even volume and net present value, as this book uses the words

These are this book's working words, and no source read defines them in this sense. Total cost of ownership is the price of a purchase plus the other costs that come with it over its life, such as installation, training, financing and downtime. A relevant cost is a cost that differs between the alternatives being compared. A break-even volume is the number of units at which the savings of an investment equal its cost. Net present value is the value today of the savings of an asset over its life at a stated rate, minus its cost. The article read for this book is read at its abstract only, and it reports that total cost of ownership looks beyond the price of a purchase to many other purchase-related costs (Ellram, 1995, abstract only). No figure is printed here, and no source read gives a typical share of any cost line.

KEY CONCEPT A make-or-buy decision is a comparison on one basis over the life of the asset, and capacity can outweigh price

The price of a bought unit and the cost of a made unit are compared on one basis, with the hours valued at stated rates and every cost line that differs written down. The comparison is read over the life of the asset, with the outlay, the saving on each unit and a stated rate. Capacity can outweigh price, because a made unit that needs hours the business does not have is a unit that waits, and the waiting is paid in the fees a late installation delays. This is the book's reasoning and never a finding. No source read gives a typical saving or premium. The case's figures are the case's own, posed by this book: a kit at $175.50 made and $202 bought, a gap of 113 hours and molds at $72,000. No figure other than the case's is printed here, and nothing is credited to any decision.

7.2 Cost per Unit, Made and Bought, on the Same Basis

The same unit, the same period, the same volume and the same quality on both sides: that is what makes two numbers comparable. A made cost that counts labor and a bought price that counts freight and oversight are two different things until both are written as a cost per unit of the same kit. The unit is the kit, the period is the year, the volume is the forecast, and the quality is the same specification on both sides, because a cheaper unit that fails more often is not the same unit.

The abstract of the article read for this book reports that a generic total cost of ownership model is not appropriate, and that a model built on a core set of cost drivers present in most computations, with auxiliary drivers chosen for the purchase at hand, is suggested (Ferrin & Plank, 2002, abstract only). The article was read at its abstract only, on October 15, 2026, and the abstract gives no list of drivers beyond that; this section prints no list of its own beyond the core set the book reads into it. The core set, as this book reads it, is the price or the direct cost, the hours, the consumables, the freight and the oversight, and the auxiliary drivers are the activity's own: equipment, floor space, the one-time transfer, and the cost of capacity the business does not have. That reading is the book's own and is not a finding of the article.

Unit economics are Book G3's (Budgeting, Cost Control and Profitability, chapter 3 of Book G3), cited here in one sentence and not retold. The purchasing of goods lies outside this book, in Book J5 (Supply Chain, Procurement and Inventory).

In the case, the kit made in-house costs $175.50 and the kit bought costs $202, so buying costs $26.50 more per kit, and the company says so before it says why it buys 207 kits, all posed by this book. The premium on the year is $5,485.50, and the hours that the purchase frees return to installations, posed by this book. Nothing is credited to the purchase beyond the hours it frees and the costs it compares.

The comparison is written in five steps, and the steps are the book's reasoning drawn from the article read at its abstract only, not a tested procedure. First, write the unit, the period and the volume, as the case writes one kit and 230 kits for the year. Second, list the core cost drivers every comparison carries, which are the price or the direct cost, the hours, the consumables, the freight and the oversight, and add the auxiliary drivers the activity needs, which for the case are the equipment, the floor space and the one-time transfer. Third, value the hours at stated rates, as the case values them at the printed illustrative rates. Fourth, write the cost per unit made and the cost per unit bought and read the difference. Fifth, read the hours the made option needs against the hours that are free, and write both readings beside the difference, because a difference per unit that ignores the hours is half a comparison.

THE RULE Compare made and bought with a core set of cost drivers and the auxiliary drivers the activity needs

The abstract of the article read for this book, read on October 15, 2026, reports that a generic model is not appropriate and that a model built on a core set of cost drivers present in most computations, with auxiliary drivers chosen for the purchase at hand, is suggested (Ferrin & Plank, 2002, abstract only). Adapted by this book for a small firm as its own rule, neither a standard imposed on the reader nor a legal duty: every comparison carries the same core lines, which are the price or the direct cost, the hours, the consumables, the freight and the oversight, and adds the lines that the activity alone needs. The case's comparison carries the lines of Table 7.2 and writes the gap of hours beside the difference of $26.50 a kit, posed by this book. No figure beyond the case's is printed here, and the rule promises no saving to any reader.

METHOD Putting a made unit and a bought unit on one basis in five steps

First, write the unit, the period and the volume, as the case writes one kit and 230 kits for the year. Second, list the core cost drivers every comparison carries, which are the price or the direct cost, the hours, the consumables, the freight and the oversight, and add the auxiliary drivers the activity needs, such as equipment, floor space and the one-time transfer. Third, value the hours at stated rates, as the case values them at the printed illustrative rates. Fourth, write the cost per unit made and the cost per unit bought and read the difference. Fifth, read the hours the made option needs against the hours that are free, and write both readings beside the difference. This is the book's reasoning drawn from the article read at its abstract only, and it is not a tested procedure. The case follows these steps with posed figures: $175.50 made, $202 bought, a premium of $26.50 and a gap of 113 hours.

7.3 The Cost of Capacity You Do Not Have

A made cost assumes hours that exist. When the hours are missing, the cost of making is the cost of waiting, and the waiting has a price that the unit cost does not show. Capacity, bottlenecks and throughput are Book J1's (Operations Management Fundamentals, chapter 3 of Book J1), cited here in one sentence and not retold; what this section adds is the arithmetic of the gap.

The gap is read in three numbers. The first is the hours the activity needs in the year. The second is the hours that are free after the work the business has already committed. The third is the difference, and the difference is what the business would have to find somewhere else: overtime, a second person, a second bench, or a supplier. The cost of the gap is then read in what the missing hours delay, and the delay is priced in the fees that a late installation postpones. That price is not a cost the business pays to anyone; it is a cost it does not collect, and it belongs in the comparison because it differs between the alternatives.

In the case, the bench needs 713 staff installer hours in the year and 600 are free after installations, a gap of 113 hours, which would hold back 37 kits by about three months each, or $14,319 of monthly fees at the Plus tier that Book L5 (Subscription and Recurring Revenue Models) prints at $129 a month, all posed by this book. The fees are Book L5's and are cited as theirs, never recounted. The company therefore buys 207 kits, which is 90 percent of the forecast, and keeps 23 in-house for repairs, pilot builds and urgent kits, which is making and buying the same thing at once, a choice section 3.7 of this book treats in one sentence.

The table below sets six readings of the bench beside their value and what each shows, so that the arithmetic of the gap can be read in one place. It is posed by this book, the fee tier is Book L5's, and nothing in it is measured.

Table 7.3 — The case's bench against its free hours (posed)

WATCH OUT Comparing a made cost with a bought price and ignoring the hours you do not have are two traps

Two traps catch a make-or-buy comparison, and each has a question that catches it. The first is comparing a made cost with a bought price, where the two are not on the same basis and the bought price leaves out freight and oversight while the made cost leaves out rework. The question is whether both are written as a cost per unit of the same kit, with the same cost lines. The second is ignoring the hours you do not have, where the made cost is the lowest and the bench cannot build the kits in time. The question is how many hours the activity needs in the year, how many are free, and what the gap costs. The book's reasoning is illustrative, and no frequency of either trap in any business is printed. In the case, the made kit is $26.50 cheaper than the bought kit and the company still buys 207 kits, because the bench lacks 113 hours, posed by this book.

7.4 Investment, Fixed Cost and the Break-Even Volume

An investment is paid up front and a fixed cost does not move with the count of units, so the question a break-even answers is how many units must pass before the saving on each unit pays the investment back. Told in words, the break-even volume is the investment divided by the saving on each unit. It is a count of units and never a promise of volume, and it is read beside the life of the asset, because an investment that pays back after the asset is worn out has not paid back.

Break-even analysis is Book G3's (Budgeting, Cost Control and Profitability, chapter 4 of Book G3), cited here in one sentence and not retold. The practice page of the Crown corporation read for this book, read on October 15, 2026 and undated, is practice guidance and never an obligation, and it says that outsourcing may be preferable when the equipment would serve only a short period or a small part of the work, its example being a three-year contract with no other profitable use (Business Development Bank of Canada, n.d.). The same page lists the total cost of ownership of equipment as more than the principal, which is the point of section 7.1. Book A5 (Financing a Business) prints that the enclosure's molds would need to be replaced or redesigned within about five years and that few buyers exist for a mold made for someone else's product (chapter 3 of Book A5), cited as theirs and not recounted, so that a mold is a specific asset: it is worth much less to anyone else than it is to the business that ordered it.

In the case, owning the molds costs $72,000 and gives an enclosure set at $64 a kit, and leaving them to the molder costs nothing up front and $134 a kit, so the difference is $70 a kit in favor of owning, all posed by this book. The difference of $70 a kit pays the molds back after 1,028.6 kits, rounded to one decimal, which falls in year 4 of a five-year life of 1,400 kits, posed by this book. The break-even is a count of kits that the saving must reach, and it is not a promise that the kits will come.

The page Productivity → Calculator & Tools (productivity/tools) of La Machine offers a general calculator, timers and a timesheet by project, and it has no template of unit cost, break-even or present value (La Machine, 2026). The timesheet can time a sample of assemblies, and it is not a time study. A business should check it in its own installation.

The table below sets five readings of the two options side by side, so that the cost up front, the cost per kit, the difference, the break-even and the life can be read together. It is posed by this book, the five-year life follows the facts of Book A5, and the break-even is a count of kits and not a promise of volume.

Table 7.4 — Owning the molds against leaving them to the molder (posed)

TOOLS OF THE TRADE What any cost model or ownership calculator must offer, whatever it is called

Nine functions carry a make-or-buy comparison, and a business should look for them rather than for a product. The first is a place to write the unit, the period and the volume. The second is the core cost lines and a place for auxiliary lines. The third is hours valued at a rate that can be changed. The fourth is the cost per unit made and the cost per unit bought, side by side. The fifth is an investment, a saving per unit and a break-even count of units. The sixth is a rate and a life, with the net present value shown beside the break-even. The seventh is a way to change the volume, the price and the rate and see the result move. The eighth is a way to keep the assumptions with the result, so that a reader can see what the numbers rest on. The ninth is an export in a form another tool can read. No brand name belongs on this list and no price; the requirement is the function.

7.5 Present Value Over the Life of the Asset

A dollar received later is worth less than a dollar today, so the savings of an asset over its life are brought back to today at a stated rate and compared with the outlay. The result is the net present value, told in words as the present value of the savings minus the cost. The rate is an assumption and not a market figure, and no source read gives a rate or a norm; the rate of the case is posed at 9 percent by this book, and it is the case's assumption and nothing else.

The reading has three parts, and each of them is a sentence rather than a formula. The first part is the saving on each unit, which is the difference between the cost of making and the cost of buying, or between the cost of owning and the price of leaving the work to someone else. The second part is the number of units over the life of the asset, which is a volume assumption and not a forecast. The third part is the rate, which brings each year's savings back to today. The present value of the savings is then compared with the outlay, and the difference is the net present value. A positive result says that, at that rate and that volume, the savings exceed the cost; it says nothing about whether the volume will come.

Valuation and the discounted cash flow are Book M2 (Business Valuation, chapter 4 of Book M2), cited here in one sentence and not retold. What this section adds to that book is not the method but the use: the same arithmetic that values a business is used here to value one asset against the alternative of not owning it.

In the case, at a posed rate of 9 percent the present value of the savings of the molds is $75,068 and the net present value of owning is $3,068, which is thin, all posed by this book. A net present value is read with the break-even of section 7.4 and never alone, because the two answer different halves of the same question: the break-even says how many kits the saving must reach, and the net present value says what the savings are worth today at a stated rate and a stated volume. In the case's posed figures, a result of $3,068 on an outlay of $72,000 is a result that a small change in the volume can erase, and section 7.8 shows what happens when the volume falls.

No source other than the book's own method is read for this section, and no figure of any source is printed in it. The rate, the life and the volume are the case's assumptions, posed by this book, and they are never a norm for a reader. A business that writes its own rate, its own life and its own volume can say what its reading rests on.

7.6 Margins Captured and Margins Given Up

Integrating a stage captures the margin of that stage and takes on its costs, its volume and its risks; contracting the stage gives the margin up and keeps the capital free. That is the book's reasoning and a way of reading a comparison, not a finding of any source. The margin of a stage is the difference between what the stage's output is worth to the business and what the stage costs to run, and it is captured only if the business can run the stage at a cost below the price it would otherwise pay. When it cannot, the margin is not captured; it is bought at a loss, and the loss is the difference between the made cost and the bought price.

The margin of a kit is Book A7's (Pricing Strategy and Revenue Models), which prints that the kit sells for $4,800 and costs $2,100, cited as theirs and not recomputed here. The margin by product, by customer and by channel is Book G3's (Budgeting, Cost Control and Profitability, chapter 7 of Book G3), and ratios are Book G4's (Financial Analysis, Ratios and KPIs), each cited here in one sentence and not retold.

In the case, buying the final assembly gives up the saving of $26.50 a kit that making would carry, and the company pays it for 207 kits, a premium of $5,485.50 on the year, in return for hours that return to installations, all posed by this book and credited with no sale. Owning the molds takes a saving of $70 a kit on the enclosure set and the cost of waiting for the first production enclosures on day 150, posed by this book. The two decisions point in opposite directions for the same reason: the assembly is bought because the hours are missing, and the molds are owned because the asset is specific and the saving is large enough to pay for it over the life.

The page Business → Store → Products (business/shop) of La Machine holds a store product with a name, a brand, a price, the cost of the product, the shipping cost, the shipping fees charged and a profit panel on the screen (La Machine, 2026). It is an example of a reading of margin by product, and it is never a make-or-buy model. A business should check it in its own installation.

IN LA MACHINE A store product records a cost beside its price, a calculator exists, and no make-or-buy model does

In the application copy of September 7, 2026, Business → Store → Products (business/shop) records a product's cost and shipping cost beside its price and shows a profit panel on the screen, which is an example of a reading of margin by product and never a make-or-buy model (La Machine, 2026). Productivity → Calculator & Tools (productivity/tools) offers a general calculator, timers and a timesheet by project, and it has no template of unit cost, break-even or present value. La Machine holds no total cost, break-even, present-value or make-or-buy model, no supplier record and no purchase order, and a business should check it in its own installation. La Machine is described and never sold, and the case does not use it.

7.7 Cash, Financing and the Timing of the Outlay

Cost and cash are two readings of the same decision. The molds cost $72,000, and that amount is paid when they are ordered; the saving of $70 a kit arrives over the years, one kit at a time. A comparison that reads only the cost sees a decision that pays back in year 4; a comparison that reads the cash sees an outlay that leaves the business long before the first saving arrives, and the two readings are both true. The question a business asks of cash is not whether the decision is profitable but whether the business can carry the outlay until the saving starts.

The practice page of the Crown corporation read for this book, read on October 15, 2026 and undated, is practice guidance and never an obligation, and it quotes one of its vice-presidents of financing saying that a large piece of equipment can take six, eight or 12 months to be made and that, with progress payments to a foreign manufacturer, 90 percent of the money can be out the door before the equipment is on site (Business Development Bank of Canada, n.d.). Those are the quoted words of one vice-president about one kind of payment, and they are never a norm and never a figure of the corporation. The page gives no cost benchmark.

The financing of the tooling and its loan are Book A5's (Financing a Business, chapters 1, 3 and 6 of Book A5), and cash flow and its forecast are Book G2's (Cash Flow Management and Forecasting), each cited here in one sentence and not retold.

In the case, the molds are ordered on day 58, the first samples come on day 128 and the first production enclosures on day 150, so the outlay of $72,000 comes long before the first saving, all posed by this book. The plan commits $89,540 in ninety days, which is 121 hours at the printed illustrative rates, or $8,340, beside the duplicate test bench at $9,200 and the molds at $72,000, posed by this book. The case does not say how the molds are paid, and this section invents no financing; a business reading this chapter should read its own payment terms before it reads its own break-even.

WHO DOES THIS Karim Haddad holds the quotes, Laurence Bergeron holds the bench and the molds, and the accountant reads the cost comparison

The roles below are the case's, posed by this book. Karim Haddad, the head of sales, holds the kit forecast, the quotes and the commercial terms. Laurence Bergeron, the president, holds the duplicate test bench, the molds and the check of the molds at the molder. The staff installer keeps the in-house build cell. The accountant reads the cost comparison at the gate of day 30 and spends 2 hours on it at Book G7's illustrative rate of $120 an hour. The outside lawyer reads the supply and quality agreement and the tooling agreement. No one is hired. In general, one person writes the cost lines and a second person reads them, which is the book's suggestion and not part of the case. In a solo business the owner writes the lines and an accountant reads them. In a larger business a finance lead owns the comparison. No hour and no amount is promised to a reader.

7.8 Sensitivity: Price, Volume and Lead Time

A decision is read under the assumption that moves it most. The three assumptions that move a make-or-buy comparison are the price of the bought unit, the volume over the life of the asset, and the lead time of the hours the business does not have. Each is tested by changing it alone and reading the result, and the test is written down beside the base reading so that a later reader can see what the decision rests on.

The abstract of the study read for this book reports that hidden costs are the costs of implementation neglected in strategic decisions, and that, in services offshoring, decision makers are more likely to make cost-estimation errors as configuration and task complexity rise, while experience and a strong orientation toward organizational design reduce those errors (Larsen et al., 2013, abstract only). The study was read at its abstract only, on October 15, 2026; it is a study of services offshoring and never of manufacturing, it gives no size of the error, and it is applied to nothing else here. What it supports is a habit rather than a figure: write the assumptions down, and test the one that moves the result most.

In the case, the price of the bought kit has a trigger at $215, written at the gate of day 90 and read in Chapter 8, posed by this book. The volume of the molds is tested at 15 percent less, and the lead time of a missing hour is the three months by which 37 kits would be held back, all posed by this book. The three tests answer three different questions: the price test asks how much more the bought kit can cost before buying stops being worth the hours it frees; the volume test asks how few kits the molds can serve before the saving no longer covers the outlay; and the lead-time test asks what the missing hours cost in fees not collected.

The table below sets two readings of the same molds beside the assumption of each, so that the effect of the volume on the result can be read in one place. Both are posed by this book, the rate and the volume are the case's assumptions and not market figures, and the sign of the result changes with the volume.

Table 7.5 — Two readings of the molds' net present value (posed)

7.9 Reading the Purchase of a Supplier or a Customer

The purchase of a supplier or a customer is a vertical integration bought rather than built, and it is read as one more make-or-buy comparison. What the price buys is a stage that already runs: its margin, its volume, its skills and its fixed cost. Against that price stands the cost of integrating the same stage by building it, which is the arithmetic of sections 7.4 to 7.7 applied to a whole activity rather than to a unit. The comparison is the same comparison; what changes is that the volume and the skills come with the purchase, and so do the fixed costs and the risks that the seller carried.

The valuation and the structure of a transfer are Books M2 (Business Valuation) and M4 (Deal Structures, Tax and Financing the Transfer), cited here in one sentence and not retold, and buying a business lies outside this book, and Book M6 (Buying a Business) is only named here.

One rule of law belongs in this reading, and it is called rather than rewritten. The Competition Act, R.S.C. 1985, c. C-34, ss. 91, 92, 109 and 110 (Canada), Justice Laws Website, current to September 30, 2026 (last amended March 26, 2026), read on October 16, 2026, is called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements), which holds competition law. Section 91 defines a merger as the acquisition of control over or a significant interest in the whole or a part of a business of a competitor, a supplier, a customer or another person, so that the purchase of a supplier or of a customer is a merger. Section 92(1) lets the Tribunal act on a merger that prevents or lessens competition substantially, including among the sources from which a trade obtains a product and among its outlets, and it applies to a merger of any size. Sections 109(1) and 110 make pre-merger notification apply only above the party-size threshold of $400 million and the transaction-size threshold, which the release of the Bureau gives as $93 million for 2026 (Competition Bureau, 2026). A threshold decides only who must notify and is never an exemption. No merger is said here to be lawful or unlawful, and only the Tribunal or a court decides.

In the case, the two integrations examined and set aside in section 2.10 are not valued, and Hydrosentinelle inc. acquires nothing, posed by this book. A lawyer reads a real purchase, and this book explains published rules and recommends no vendor.

QUÉBEC & CANADA NOTE A supplier or a customer bought is a merger under the Competition Act, and this book is not legal advice

The Canadian point of this section is that the purchase of a supplier or of a customer is a merger under the Competition Act, R.S.C. 1985, c. C-34, ss. 91, 92, 109 and 110 (Canada), current to September 30, 2026, read on October 16, 2026, and the Act is called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements), which holds competition law. Section 91 makes the purchase of a supplier or of a customer a merger, section 92 lets the Tribunal act on a merger of any size, and the notification thresholds of sections 109 and 110, which the release of the Bureau gives as $400 million and $93 million as of 2026 (Competition Bureau, 2026), decide only who must notify and are never an exemption. This book explains the law as published and is not legal advice. This book explains published rules and recommends no vendor. A reader with a real purchase should read Book H4 and ask a lawyer; nothing here says whether the reader's own purchase must be notified or reviewed. The case acquires nothing, posed by this book.

WHAT CHANGED The notification threshold stays at $93 million for 2026, and the Act is read again on the day of writing

The dated change comes from the release read on October 15, 2026: as of 2026 the transaction-size threshold for pre-merger notification stays at $93 million, measured on the assets in Canada of the business acquired or the revenues they bring from sales in, from or into Canada, and notice is also generally required when the parties and their affiliates exceed $400 million in combined assets or revenues (Competition Bureau, 2026). The Competition Act was read on October 16, 2026, current to September 30, 2026, the day this chapter was written; its rules are called to Book H4 and never rewritten here. A threshold can change and a law can change, so a reader reads each on the reader's own date, and the state of either after the date it was read is not stated here. The application code of La Machine was read at its copy of September 7, 2026, on October 16, 2026 (La Machine, 2026). The dates of sources are never dates of the case, which is counted in days.

7.10 The Case's Numbers, Made Against Bought

The three numbers of the case are now read side by side, and each of them is posed by this book and none of them is measured. The first is the kit: $175.50 made and $202 bought, so that buying costs $26.50 more per kit. The second is the bench: 713 hours needed against 600 free, a gap of 113 hours and 37 kits held back by about three months each. The third is the molds: $72,000 that save $70 a kit and pay back after 1,028.6 kits, in year 4, with a net present value of $3,068 at a posed rate of 9 percent and minus $8,304 at 15 percent less volume.

The reading of each is the book's reasoning. The price says make, because the made kit costs $26.50 less than the bought kit and the difference is real. The capacity says buy, because the bench lacks 113 hours and the waiting would delay $14,319 of monthly fees at the Plus tier that Book L5 prints. The molds are paid late and own a specific asset, so the comparison is read over five years and 1,400 kits, and the result is thin enough that the control of a specific asset decides rather than the arithmetic alone. The verdicts are those of the grid of Chapter 3, cited and not recomputed: the final assembly scores 57 and is a hybrid, and the enclosure tooling scores 68 and is a hybrid.

Figure 7.1 sets three of the case's numbers side by side: the cost of a kit made and bought, the hours the bench lacks, and the payback of the molds.

Figure 7.1 — Made against bought. A kit costs $175.50 made and $202 bought; the bench lacks 113 hours; the molds pay back in year 4, with a net present value of $3,068 at 9 percent (posed).

The table below sets the three decisions of the case beside the made option, the bought option and what decides, so that the arithmetic of the chapter can be read in one place. Every figure is posed by this book, nothing is measured, and nothing is credited to a decision.

Table 7.6 — The case's three decisions in numbers (posed)

Nothing is credited to any of the three decisions beyond the costs compared and the hours freed. No sale, no installation and no growth is attributed to the purchase of the assembly, to the ownership of the molds or to any sourcing decision, and the hours that the purchase frees return to installations, which is the only effect the case records. Chapter 8 puts the reviews that keep these numbers current on the calendar, and Chapter 11 writes the plan that puts the first of them in place.

DO THE MATH The case's three numbers, step by step

Illustrative figures; the case is fictional. Step 1 — The kit made in-house costs $175.50 and the kit bought costs $202, so buying costs $26.50 more per kit, and 207 kits at $26.50 make a premium of $5,485.50 on the year. Step 2 — The bench needs 713 staff installer hours and 600 are free, which leaves a gap of 113 hours, and 37 kits held back by about three months each would delay $14,319 of monthly fees at the Plus tier of Book L5. Step 3 — The molds cost $72,000 and save $70 a kit, and 72,000 divided by 70 is 1,028.6 kits, rounded to one decimal, which falls in year 4 of a five-year life of 1,400 kits. Step 4 — At a posed rate of 9 percent, the present value of the savings is $75,068 and the net present value of owning is $3,068. Step 5 — In the case's illustrative figures, at 15 percent less volume the net present value is minus $8,304. The figures are posed by this book, are the case's and not a norm, nothing is measured, nothing is credited to any decision, and the sums are exact except where the rounding is declared.

CASE IN POINT Hydrosentinelle inc. buys 207 kits at a higher price for want of hours and owns its molds on thin numbers

Hydrosentinelle inc. is a fictional business created for this encyclopedia; its founders, its employees, its customers, its partners, its suppliers, its figures and the events described are invented, and any resemblance to a real business is coincidental. The final assembly is bought at $202 a kit against $175.50 made, because the bench lacks 113 hours, and 23 kits stay in-house. The molds are owned at $72,000 and held at the molder, with a net present value of $3,068 at a posed rate of 9 percent and minus $8,304 at 15 percent less volume. Karim Haddad holds the quotes and the commercial terms, and Laurence Bergeron holds the bench and the molds. All of it is posed by this book, illustrative and not a forecast, and no verdict is given beyond the grid's.

Costly Mistakes

• Comparing a made cost with a bought price. The two are not on the same basis until both are written as a cost per unit of the same kit, with the same cost lines and the hours valued at the same rates.
• Leaving the hours you do not have out of the comparison. A made cost assumes hours that exist, and when the hours are missing the cost of making is the cost of waiting, read in the fees a late installation delays.
• Reading a break-even as a promise of volume. A break-even is a count of units that the saving must reach, and it says nothing about whether those units will come.
• Reading a net present value alone. A thin result at one volume can turn negative at another, so the net present value is read beside the break-even and beside the assumptions it rests on.
• Forgetting that cost and cash are two readings. An outlay leaves the business long before the first saving arrives, and a decision that pays back in year 4 can still be one the business cannot carry.
• Testing no assumption at all. A decision read at one price, one volume and one lead time is a decision that has not been read under the assumption that moves it most.
• Treating a notification threshold as an exemption. The thresholds decide only who must notify, and the Tribunal may act on a merger of any size.

Key Takeaways

• A make-or-buy comparison counts the costs that differ between the alternatives and leaves out the costs that stay the same whichever way the business decides, and the rule is the book's reasoning rather than a finding of any source read.
• A made unit and a bought unit are comparable only when they are written as a cost per unit of the same unit, in the same period, at the same volume and at the same quality, with the hours valued at stated rates.
• Capacity can outweigh price: a made unit that needs hours the business does not have is a unit that waits, and the waiting is read in the fees a late installation delays.
• An investment is read as a break-even, which is the investment divided by the saving on each unit, and as a net present value, which brings the savings back to today at a stated rate and is read beside the break-even rather than alone.
• Cost and cash are two readings of the same decision, and the purchase of a supplier or a customer is one more make-or-buy comparison, read with the Competition Act called to Book H4 and with a lawyer for a real purchase.
• No source read gives a typical saving, a premium of a contract manufacturer or a hidden cost as a share of a price, and every figure of the case in this chapter is posed by this book, illustrative, and credited with no sale and no result.

Your Action Plan

1. Write the cost lines of one activity on one basis. Take one activity, write the unit, the period and the volume, list the core cost lines and the auxiliary lines the activity needs, and value the hours at stated rates. Deliverable: a written cost per unit made and a written cost per unit bought for the same unit. Verification: both sides carry the same cost lines, and every line that does not differ between the alternatives is marked as excluded.
2. Read the hours the made option needs against the hours that are free. Count the hours the activity needs in the year, count the hours that are free after the work already committed, and write the gap beside the difference per unit. Deliverable: a written gap of hours with the cost of the waiting it causes. Verification: the gap is written beside the difference per unit, and the cost of the waiting is read in a fee or a price the business can name.
3. Compute the break-even and the net present value of the investment. Divide the investment by the saving on each unit to get the break-even count, then bring the savings back to today at a rate the business writes down, and read the result beside the break-even. Deliverable: a break-even count and a net present value, each with its assumptions written beside it. Verification: the rate, the life and the volume are written down, and the result is read beside the break-even and never alone.
4. Test the assumption that moves the result most. Change one assumption at a time, the price of the bought unit, the volume over the life, or the lead time of the missing hours, and write the result beside the base reading. Deliverable: at least two readings of the same decision under different assumptions. Verification: each reading names the assumption it changes and the sign of the result is read, not only its size.
5. Write the decision and its triggers in one place. Put the comparison, its assumptions, its break-even, its net present value and the triggers that would reopen it in a register the business keeps, and name the person who reads it. Deliverable: a written decision with its triggers and its reader. Verification: a second person has read the comparison and can say what would reopen the decision.

In La Machine

In the application copy of September 7, 2026, the pages that touch this chapter are few. Business → Store → Products (business/shop) holds a store product with a name, a brand, a price, the cost of the product, the shipping cost, the fees charged, a commission and a profit panel on the screen, which is an example of a reading of margin by product and never a make-or-buy model (La Machine, 2026). Productivity → Calculator & Tools (productivity/tools) offers a general calculator, timers and a timesheet by project, and the timesheet can time a sample of assemblies, though it is not a time study. A business should check both in its own installation.

What a business needs elsewhere is the arithmetic of this chapter. La Machine holds no total cost, break-even, present-value or make-or-buy model, no supplier record, no request for quotation, no comparison of quotes and no purchase order, and its calculator has no template for any of them. A cost model that sets a made unit beside a bought unit, a break-even and a net present value is chosen by function, and Chapter 10 sets out the functions to look for. La Machine is one worked example of a business platform, described and never sold, and the case does not use it.

Self-Check

1. Which costs belong in a make-or-buy comparison, and which ones stay out of it?
2. What makes a made unit and a bought unit comparable, and what are the core cost lines every comparison carries?
3. Why can a business correctly buy a unit that costs more than the unit it could make?
4. How is a break-even volume read, and why is a net present value read beside it rather than alone?
5. What does the purchase of a supplier or a customer add to the comparison, and which book holds the law that applies to it?

Answers. 1. The costs that differ between the alternatives belong in the comparison, and the costs that stay the same whichever way the business decides stay out of it, because a cost that does not move with the decision makes both sides look larger without changing which side is larger; the rule is the book's reasoning, and the article read for this book reports that total cost of ownership looks beyond the price of a purchase to many other purchase-related costs (Ellram, 1995, abstract only). 2. The same unit, the same period, the same volume and the same quality on both sides, with the hours valued at stated rates, make the two comparable; the core cost lines the book reads into the comparison are the price or the direct cost, the hours, the consumables, the freight and the oversight, and the auxiliary lines are the activity's own, which is the book's own reading of the article that reports a generic model is not appropriate (Ferrin & Plank, 2002, abstract only). 3. Because a made cost assumes hours that exist, and when the hours are missing the cost of making is the cost of waiting; in the case the made kit is $26.50 cheaper and the company still buys 207 kits, because the bench needs 713 hours and 600 are free, a gap of 113 hours that would hold back 37 kits, all posed by this book. 4. A break-even is the investment divided by the saving on each unit, and it is a count of units and never a promise of volume; a net present value brings the savings back to today at a stated rate and is read beside the break-even, because a thin result at one volume can turn negative at another, as the case's $3,068 at a posed rate of 9 percent and minus $8,304 at 15 percent less volume show, both posed by this book. 5. It adds a stage that already runs, with its margin, its volume, its skills and its fixed cost, read against the cost of building the same stage, and it is a merger under section 91 of the Competition Act, R.S.C. 1985, c. C-34, current to September 30, 2026, read on October 16, 2026, which is called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements), which holds competition law, with the notification thresholds of sections 109 and 110 deciding only who must notify (Competition Bureau, 2026).

Chapter References

Business Development Bank of Canada. (n.d.). Manufacturing equipment: 6 steps to plan your purchase. Retrieved October 15, 2026, from https://www.bdc.ca/en/articles-tools/money-finance/get-financing/manufacturing-equipment-steps-plan-purchase

Competition Act, R.S.C. 1985, c. C-34, ss. 45, 77, 90.1, 91, 92, 109 and 110 (Canada). Justice Laws Website, current to September 30, 2026 (last amended March 26, 2026) — read October 16, 2026. https://laws-lois.justice.gc.ca/eng/acts/C-34/

Competition Bureau. (2026, March 2). Pre-merger Competition Bureau notification threshold to remain at $93M in 2026 [News release]. Government of Canada. Retrieved October 15, 2026, from https://www.canada.ca/en/competition-bureau/news/2026/03/pre-merger-competition-bureau-notification-threshold-to-remain-at-93m-in-2026.html

Ellram, L. M. (1995). Total cost of ownership: An analysis approach for purchasing. International Journal of Physical Distribution & Logistics Management, 25(8), 4–23. https://doi.org/10.1108/09600039510099928

Ferrin, B. G., & Plank, R. E. (2002). Total cost of ownership models: An exploratory study. Journal of Supply Chain Management, 38(2), 18–29. https://doi.org/10.1111/j.1745-493X.2002.tb00132.x

La Machine. (2026). Documentation — route reference inventory. Internal product documentation, read October 16, 2026, together with the code of the pages this book describes: business/shop and productivity/tools.

Larsen, M. M., Manning, S., & Pedersen, T. (2013). Uncovering the hidden costs of offshoring: The interplay of complexity, organizational design, and experience. Strategic Management Journal, 34(5), 533–552. https://doi.org/10.1002/smj.2023

Chapter 8 — Typical Tasks, Cadence and Workflows

Sourcing work is a routine of short reviews and not a rescue, so this chapter puts the decisions of the earlier chapters on the calendar — a weekly build release to the contract manufacturer, a monthly supplier scorecard, a quarterly capability and trigger review, a yearly make-or-buy review of every activity with the audit at the contract manufacturer and the check of the molds, and the monthly upkeep of the in-house build cell — sets each task in the format of the application's task table in two tables of six columns at most, shows the routine in three sizes of business, reads what to automate and what to leave to a person, names what breaks a sourcing routine, and applies all of it to Hydrosentinelle inc., where the routine takes 89 hours a year, posed by this book.

What You'll Be Able to Do

• Describe the five rhythms of a sourcing routine — each week, each month, each quarter, once a year and the monthly upkeep of the in-house build cell — and say what each one keeps alive.
• Write each task of the routine in the ten-field task format that Book N2 (Piloting Tasks, Executions and Projects, chapter 2) treats for the plan task, with a person, a trigger, an input, an output and an observable end.
• Cost the routine in hours a year, line by line, and read the total against the hours the same people have for the work.
• Hold a weekly build release that carries the versions of the firmware image and of the end-of-line test protocol, and a monthly scorecard that reads counts and dates and never a grade.
• Read the three written triggers each quarter whether or not one has been reached, and rescore every activity on the grid once a year.
• Decide what to automate and what to leave to a person, and name the ordinary ways a sourcing routine breaks and the remedy for each.

Why It Matters

Sourcing work fails quietly, by being postponed. The release to the contract manufacturer is written when somebody remembers, the scorecard is read when a lot comes back held, the triggers are looked at when a price rises, and the grid is rescored when a new person asks why the boards are bought and the firmware is kept. Every one of those decisions is reasonable on the day it is taken, and together they produce a business that reacts to its own supply instead of running it. A rescue is a recall of attention because something has already gone wrong: a lot that arrived short, a price that moved past what the company expected, a release that arrived late enough to hold an installation. A routine is the opposite arrangement, and it is smaller than it sounds.

The reason a routine matters more than a decision is that a decision without a day has no owner. The earlier chapters of this book decided the direction of integration, the grid and the written decision, the agreement and the road back, the hybrid forms and their safeguards, the records and the data, and the numbers. None of those decisions survives on its own. What converts a decision into a habit is a small number of short reviews with fixed days, and the reviews are unglamorous: half an hour a week with the release, an hour a month with the scorecard, two hours a quarter with the triggers, four hours a year with the grid, and an hour a month at the in-house bench.

The cost of the routine is worth stating plainly. The routine this chapter adds to Hydrosentinelle inc., a fictional business created for this encyclopedia, takes 89 hours a year, and every hour of that figure is posed by this book and illustrative. It sits beside routines that other books already own and that this chapter never recounts: the routine of buying work and its bench of providers, which belongs to Book I8 (Freelancers, Agencies and Outsourcing, chapter 9 of Book I8); the routine of operations, which belongs to Book J1 (Operations Management Fundamentals, chapter 8 of Book J1); and the routine of quality management, which belongs to Book J4 (Quality Management and Continuous Improvement, chapter 9 of Book J4). None of their hours is added to the 89 hours of this chapter, and none of their tasks is described again here.

In This Chapter

Section 8.1 explains why sourcing work is a routine of short reviews rather than a rescue, sets out the five rhythms and the ten lines of the routine with their hours, and gives the working words this chapter uses. Section 8.2 covers the weekly build release with the contract manufacturer. Section 8.3 covers the monthly supplier scorecard. Section 8.4 covers the quarterly capability and trigger review. Section 8.5 covers the yearly make-or-buy review of every activity, with the audit at the contract manufacturer and the check of the molds at the molder. Section 8.6 sets the routine in the ten-field task format that Book N2 (Piloting Tasks, Executions and Projects, chapter 2) treats for the plan task, in two tables of six columns at most. Section 8.7 distributes the same tasks across three sizes of business. Section 8.8 separates what can be automated from what stays with a person. Section 8.9 names what breaks a sourcing routine and what to do when a task is missed.

8.1 Sourcing Work as a Routine

A rescue is a recall of attention because something has already gone wrong; a routine is the handful of short reviews that runs all year whether or not anything has gone wrong, so that fewer readings come late.

The routine of this chapter has five rhythms, and each of them keeps something different. Each week, the build release tells the contract manufacturer what to build and carries the versions of the firmware image and of the end-of-line test protocol, so that the week's kits are the kits the company asked for. Each month, the supplier scorecard reads a few counts and dates about the contract manufacturer and the second source, so that a drift is seen while it is still small. Each quarter, the capability and trigger review reads the capacity of the three suppliers against the scorecards of the quarter and reads the three written triggers, so that a condition the company wrote down is read even in a quarter when nothing has happened. Once a year, the make-or-buy review rescores every activity on the grid and rewrites each verdict with its reason and its triggers, so that a decision taken in one year is not carried into another unread. And each month, the upkeep of the in-house build cell keeps the means to bring an activity back, so that the road back is a real option and not a sentence in a document.

The rhythms are deliberately unequal in size. The weekly release takes half an hour; the monthly scorecard takes an hour and its quality lines half an hour; the quarterly review takes two hours for each founder; the yearly review takes four hours for each founder, the audit six hours and the check of the molds three hours; the upkeep of the in-house cell takes an hour a month. What they have in common is that each one has a person, a day or a trigger, an input, an output and a way to see that it was done.

The cadences below are this book's own method, and no source read measures how often a supplier fails a release, how long a scorecard takes, what a review returns or how many reviews a small firm should hold. The hours of the case are posed by this book, are illustrative, and are never a norm, a benchmark or a target for any reader. This chapter decides nothing that the earlier chapters decided. Chapter 2 owns the direction of integration, Chapter 3 the grid and the written decision, Chapter 4 the agreement, the pilot builds, the transition and the road back, Chapter 5 the hybrid forms and the safeguards, Chapter 6 the records and the data, and Chapter 7 the numbers. This chapter answers one question: what does a small firm do each week, each month, each quarter and each year to keep its make-or-buy decisions and its suppliers in order, who does it, how long does it take, and what can be automated without the decision being automated?

Two routines of other books keep their own cadence and are cited here rather than counted again. The routine of buying work and its bench of providers belongs to Book I8 (Freelancers, Agencies and Outsourcing, chapter 9 of Book I8). The routine of operations belongs to Book J1 (Operations Management Fundamentals, chapter 8 of Book J1). Neither is retold here, and neither is added to the hours below.

Figure 8.1 lays out the case's sourcing routine by rhythm, with its hours a year. It reads as four columns, one for each rhythm, and each column lists the tasks of that rhythm with who holds each task and the hours a year it takes, from the weekly build release of 26 hours to the monthly upkeep of the in-house build cell of 12 hours; a note gives the total of 89 hours a year, Karim Haddad 50, Laurence Bergeron 27 and the staff installer 12, and a second note says that the records can be gathered automatically and that the decision to keep, move or bring back an activity stays with the founders.

Figure 8.1 — The routine by rhythm. A weekly release, monthly scorecards, quarterly trigger reviews and yearly reviews, audits and checks: 89 hours a year (posed).

The table below sets ten lines of the routine beside who does each, the hours each time, the times a year and the hours a year. It is posed by this book, it is never a norm, and it does not count the routine of buying work of Book I8 (Freelancers, Agencies and Outsourcing, chapter 9 of Book I8) or the routine of operations of Book J1 (Operations Management Fundamentals, chapter 8 of Book J1) again. Read it as the skeleton of a cadence rather than as a description of any real business: the columns are the same for a solo owner and for a business of forty people, and what changes with size is who fills each cell, which is the subject of section 8.7.

Table 8.1 — The routine of the case, with its hours (posed)

The ten lines add up to 89 hours a year, Karim Haddad 50, Laurence Bergeron 27 and the staff installer 12, and the hours are posed by this book.

DEFINITION Release, scorecard, trigger review and make-or-buy review, as this book uses the words

These are this book's working words, and no source read defines them in this sense. A release is the written instruction of the week: which kits to build, with the version of the firmware image that the company supplies and the version of the end-of-line test protocol. A scorecard is the monthly reading of counts and dates about a supplier, and it is never a grade. A trigger review is the quarterly reading of the written conditions that would bring an activity back in-house or move it, read whether or not a condition has been reached. A make-or-buy review is the yearly rescoring of every activity on the grid, with each verdict rewritten with its reason and its triggers. A release is not a purchase order, a scorecard is not an evaluation of a supplier's worth, a trigger review is not a meeting about a problem, and a make-or-buy review is not a plan. The words are this book's own, and the case's hours are posed by this book.

DO THE MATH The 89 hours a year of the routine, line by line

Illustrative figures; the case is fictional. Step 1 — The weekly build release takes 0.5 hour and is sent 52 times, which gives 26 hours a year. Step 2 — The monthly supplier scorecard takes 1 hour and is read 12 times, which gives 12 hours a year for Karim Haddad, and its quality lines take 0.5 hour and are read 12 times, which gives 6 hours for Laurence Bergeron. Step 3 — The quarterly review takes 2 hours and is held 4 times, which gives 8 hours for each of the two founders. Step 4 — The yearly review takes 4 hours for each founder, the audit takes 6 hours, the check of the molds takes 3 hours, and the upkeep of the in-house cell takes 1 hour and is done 12 times, which gives 12 hours for the staff installer. Step 5 — The lines add up to 89 hours a year, Karim Haddad 50, Laurence Bergeron 27 and the staff installer 12, which is $5,160 at the printed illustrative rates of $60 an hour for each founder and $45 an hour for the staff installer. The hours are posed by this book, are the case's and not a norm, are valued at illustrative rates and never at a pay, and the sums are exact.

8.2 The Weekly Build Release With the Contract Manufacturer

Once a week one person sends the contract manufacturer the written instruction of the week, and reads the previous week's answer. The release is the shortest line of the routine and the one that keeps the week's kits aligned with what the company asked for.

The release is written because the contract manufacturer performs two steps that belong to the company. The contract manufacturer flashes the firmware image that the company supplies, and runs the company's end-of-line test on the company's duplicate test bench. Both steps carry a version, and a version that is not written down is a version that will drift. So the release carries the count of kits for the week, the version of the firmware image, the version of the end-of-line test protocol, the due date, and a line for the changes since last week, or the words "no change". The last line matters as much as the others: a release that says "no change" is a release that has been read, and a week with no release is a week with no record.

What happens when the contract manufacturer cannot meet the release is part of the routine rather than an exception to it. The contract manufacturer calls the same day, the lot is written into the weekly note with its reason, and the reading of the trigger on late releases is taken at the next quarterly review of section 8.4. Nothing is decided on the call beyond what the week's build needs; the decision about whether a pattern of late releases changes the sourcing decision belongs to the quarterly review, where the counts of the quarter are in front of the two founders.

The supply and quality agreement that governs the release is read in Chapter 4 and is never retold here. What the agreement lets the company require of the contract manufacturer, and what it lets the company audit, is a question for the agreement, for Book H1 (Contracts and Commercial Agreements) and for a lawyer, and this book decides nothing about any real agreement.

In the case, Karim Haddad, the head of sales, holds the release. It takes 0.5 hour each time, is sent 52 times a year, and is 26 hours a year, all posed by this book. The first release of 25 supplier-built kits is delivered on day 84, which is inside the ninety days of the plan of Chapter 11, and the routine starts when the gate of day 90 has set it as plan tasks with a verification criterion.

THE RULE A weekly release is written, numbered with its versions and confirmed in writing before the week's build begins

The rule is this book's own for the case, drawn from its reasoning and not from a source. Each week the release names the kits, the version of the firmware image and the version of the end-of-line test protocol, and the contract manufacturer confirms it in writing before the build begins, so that a record exists for each week of the year. A week whose release was never written is a week whose kits cannot be checked against anything. What an agreement requires of either side is a question for Book H1 (Contracts and Commercial Agreements) and a lawyer, and this book decides nothing about any real agreement. The rule is neither a standard nor a legal duty, and it promises no result. The case's weekly release takes 0.5 hour each time, is sent 52 times a year, and is 26 hours a year, posed by this book. No figure beyond the case's is printed here, and no saving is promised to any reader.

METHOD Setting up the weekly build release in five steps

First, write the release template, with the count of kits, the version of the firmware image, the version of the end-of-line test protocol, the due date and a line for the changes since last week. Second, name who sends it, who is the backup, and on which day of the week it goes. Third, send it and ask for the written confirmation of the contract manufacturer. Fourth, read the previous week's returned test records against the release and write what differs, or the words "no change". Fifth, after a quarter, read the releases sent against the releases delivered on time and change the template if a line is never used. This is the book's reasoning and not a tested procedure, and no source read measures what a release returns. The case follows these steps with a weekly release of 0.5 hour, sent 52 times a year, posed by this book.

8.3 The Monthly Supplier Scorecard

Once a month one person reads a few counts and dates about the contract manufacturer and about the second source of the sensor component, and a second person reads the quality lines. The scorecard is the line of the routine that turns a month of records into a reading.

What is read is short and fixed. It reads the first-pass yield at the end-of-line test, because that is the count that says how many kits passed the company's own test the first time. It reads the lots held and for how many days, because a lot held is a lot that did not ship. It reads the releases on time, because a release that arrives late moves the week's build. It reads the price per kit against the trigger, because the trigger is a written number and a written number is read rather than remembered. And it reads the share of the sensor component bought from the second source, because a second source that is qualified and never used is not a second source. Five readings, in the same order, every month.

The counts come from records the company owns. The end-of-line test records of each supplier-built kit come back to the company every day under the supply agreement, so the counts of the scorecard are drawn from records the company holds rather than from a report the supplier writes. That distinction is the whole point of the line: a supplier's own report is a supplier's own reading, and the company's reading is drawn from the records the agreement returns.

The scorecard is made of counts and dates and never of grades. A first-pass yield is a count of kits, not a mark for a supplier; a lot held for four days is a date and a count, not a judgment about a firm. Each reading that calls for action leaves the scorecard with an owner and a day, and no reading is a target. The routine of quality management is Book J4's (Quality Management and Continuous Improvement, chapter 9 of Book J4), and the purchasing of goods lies outside this book, in Book J5 (Supply Chain, Procurement and Inventory); neither is retold here.

In the case, Karim Haddad, the head of sales, holds the scorecard, 1 hour each time, 12 times a year, 12 hours a year, and Laurence Bergeron, the president, holds its quality lines, 0.5 hour each time, 12 times a year, 6 hours a year, all posed by this book. The scorecard is live on day 88, near the end of the ninety days of the plan of Chapter 11, so that a month of returned test records sits behind it before it is first read. The readings of day 365 are posed by this book: a first-pass yield of 98.6%, one lot of 25 kits held for 4 days and released, 21 of 23 releases on time, and the second source at 30% of the sensor component. The 23 releases of the year are the delivered lots the company received, read as on time or late, and they are not the 52 weekly build releases of section 8.2.

WATCH OUT Reading a count as a grade and skipping a quiet month are two traps

Two traps catch a monthly scorecard, and each has a question that catches it. The first is reading a count as a grade: a first-pass yield becomes a mark for a supplier, and the work bends toward the number rather than toward the flow. The question is what action this reading calls for, and who owns it. The second is skipping a quiet month: nothing seemed to change, so no page exists for the month, and a drift is read late. The question is whether a page exists for every month, even a page whose only words are "no change". The book's reasoning is illustrative, and no frequency of either trap in any business is printed, because no source read measures one. In the case, the scorecard is made of counts and dates and shows 21 of 23 releases on time on day 365, posed by this book, and each reading that calls for action leaves with an owner and a day.

8.4 The Quarterly Capability and Trigger Review

Once a quarter the two founders sit down and read the capacity of the contract manufacturer, of the molder and of the second source against the scorecards of the quarter, and read the three written triggers. The quarterly review is the line of the routine that looks up from the month and asks whether the arrangement still fits.

The triggers of the case are posed by this book and are written at the gate of day 90. Assembly comes back in-house if the end-of-line defect rate exceeds 2% for 2 months running, if the price per kit exceeds $215, or if 3 releases in a year arrive late. Each trigger is a condition the company wrote down in advance, and each is read each quarter whether or not it has been reached. That is the part of the line that is easy to lose: a trigger that has not been reached is still read, because a trigger read only when it fires is a trigger that was never a control.

A trigger reached is a reading that leads to a decision and never to an automatic move. The review reads the condition, reads the counts behind it, and decides; the decision may be to bring the activity back, to renegotiate, to qualify another source, or to keep the arrangement and watch. What the review never does is move work because a number crossed a line. The written trigger is what makes the decision reviewable, and the decision is what makes the trigger useful.

The format of a meeting is Book I7's (Communication, Meetings and Decision-Making), and this chapter asks only two things of a quarterly review: the same time each quarter, and a written end. Each review ends with an owner and a day for each decision, written into the register of decisions the company keeps.

In the case, Karim Haddad, the head of sales, and Laurence Bergeron, the president, each hold the review, 2 hours each time, 4 times a year, 8 hours a year each, all posed by this book. On day 365 no trigger is reached and nothing is brought back in-house, posed by this book, and nothing says that the review caused that: a quarter in which nothing fires is a quarter in which the routine did its work quietly, and no result is credited to it.

KEY CONCEPT A routine keeps control only when each line has an owner, a day and a written output

The idea is this book's reasoning and never a finding. A release, a scorecard and a review keep a business in control of its suppliers only if each one has a person who holds it, a day on which it is held, and an output that can be read by somebody else. A reminder only prompts the person who set it, and a task marked done proves nothing until a second person reads the criterion. The automation collects the records, and the founders decide. The routine of the case takes 89 hours a year, Karim Haddad 50, Laurence Bergeron 27 and the staff installer 12, all posed by this book and never a norm, and nothing is credited to it. No source read measures a cadence of sourcing work, and no source read measures what a review returns. The figures of the case are these: 89 hours a year, and the three shares of it, and they are illustrative.

8.5 The Yearly Make-or-Buy Review of Every Activity

Once a year the two founders rescore the six activities on the grid of Chapter 3 (section 3.8), with the rule make at 75 or more, hybrid from 55 to 75, buy below 55, and rewrite each verdict with its reason and its triggers. The yearly review is the line of the routine that keeps a decision from becoming an assumption.

The grid of the case is cited from section 3.10 and posed by this book. The six activities scored 92 for the firmware and the alert logic, 68 for the enclosure tooling of the second generation, 35 for the electronic boards, 57 for the final kit assembly and the end-of-line test, 71 for the installation in buildings, and 35 for the cloud hosting of the monitoring service. The yearly review reads each score again rather than reading last year's verdict: a score that has moved is a verdict that may have moved with it, and a score that has not moved is a verdict that can be kept with its reason written beside it. The firmware and the alert logic stay in-house, and the review never reopens that.

The yearly audit at the contract manufacturer and the yearly check of the molds at the molder belong to the same rhythm, because both are the company looking at its own assets and its own records first-hand rather than through a report. What the supply and quality agreement lets the company audit or require of the contract manufacturer, and what the tooling agreement lets it check at the molder, are questions for the agreement, for Book H1 (Contracts and Commercial Agreements) and for a lawyer, and this book decides nothing about any real agreement. The audit is not an inspection of a supplier's worth; it is the company reading the line, the records and the molds that carry its own name.

In the case, Karim Haddad, the head of sales, and Laurence Bergeron, the president, each hold the review, 4 hours once a year, 4 hours a year each, all posed by this book. Laurence Bergeron holds the audit at the contract manufacturer, 6 hours once a year, and the check of the molds at the molder, 3 hours once a year, also posed by this book. The yearly review is the line that produces the register of decisions the other lines feed.

QUÉBEC & CANADA NOTE An audit clause lives in the agreement, and a record that names a person is a question for Book H2

The Canadian point of this section is where an audit right lives. What the supply and quality agreement lets the company audit or require of the contract manufacturer, and what the tooling agreement lets it check at the molder, are questions for the agreement, for Book H1 (Contracts and Commercial Agreements) and for a lawyer, and this book decides nothing about any real agreement. A scorecard or a minute that names a person can hold personal information, and that is Book H2's (Privacy and Personal Information). This book explains published rules and recommends no vendor, no supplier, no contract manufacturer, no molder and no advisor. The case's scorecard reads counts and dates of kits and names no customer, and the case has no union and makes no acquisition. Nothing here says whether the reader's own agreement, audit clause or record complies, and a business with such a question should ask an adviser.

8.6 The Task Table

The routine of this chapter is written in the ten-field task table of the application (task, objective, trigger, frequency, automation level, input, output, verification criterion, role and tools), which Book N2 (Piloting Tasks, Executions and Projects, chapter 2) treats for the plan task, and the ten lines of Table 8.1 can be written in that format without inventing a new one. The table is split in two, because a table of ten columns does not read aloud, and each of the two opens with the Task column.

The table below sets seven tasks beside their objective, their trigger, their frequency, their automation level and their input. It is the first of two tables, and the second continues the same seven tasks in the same order. Read the automation column as a statement about what a tool may prompt and never as a statement about what a tool may decide: the highest level any line reaches is a count drawn from returned test records and a release sent by a person.

Table 8.2 — The tasks of the routine: objective, trigger, frequency, automation and input

The table below continues the same seven tasks in the same order, with their output, their verification criterion, their role and their tools. The verification criteria are observable, and the tools are functions and not products: no tool is chosen for the case, and no brand name belongs in a cell. Read the criterion column as the part of the row that matters most, because a task whose criterion cannot be checked by a second person is a task that will be marked done without being done.

Table 8.3 — The tasks of the routine: output, verification, role and tools

WHAT CHANGED The application copy was read on October 16, 2026, and a recurring review can be written as a plan task

The dated reading is this one: the application code was read at its copy of September 7, 2026, on October 16, 2026 (La Machine, 2026). In that copy the weekly release, the monthly scorecard, the quarterly review and the yearly review can be written as recurring plan tasks with a required "Done when" criterion, and the Mark as done button does not verify that criterion; the platform computes no scorecard, sends no release to a supplier and reads no test record. None of these dates is a date of the case, which is counted in days and never in calendar years or months, and the readings of day 365 belong to section 8.9 and never to this box. A reader checks every page on the reader's own date, because an installation can differ from the copy read here. No tool is chosen for the case, and no result is promised to any reader.

8.7 Sharing Sourcing Work in Three Sizes of Business

The routine does not change with the size of the business, but the number of people who play its roles does. The three sizes below are a way of distributing the same tasks and no claim about how businesses of that size actually work. The first and the third columns are this book's suggestion.

The table below sets the five lines of the routine beside who does the work in a solo business, in the case's business and in an established business. The case's business is Hydrosentinelle inc., and every cell in that column is posed by this book. Read the table as a distribution of responsibility: the tasks are identical in every row, and what changes is how many people carry them. No hour is promised to a reader, and no cell is a norm.

Table 8.4 — The routine in three sizes of business

Three rules hold in every size, and they are the part of this section worth keeping when the table is forgotten. The first is that whoever holds each line has a named backup, because a line whose holder is away is a line that does not happen. The second is that a reading leaves with an owner and a day, because a reading that leaves with nothing will be taken again next month on the same item. The third is that a task marked done is always read by a second person, because the box being ticked and the criterion being met are two different facts.

WHO DOES THIS Karim Haddad holds the release and the scorecard, Laurence Bergeron holds the quality lines, the audit and the molds, and the staff installer keeps the build cell

The roles below are the case's, posed by this book. Karim Haddad, the head of sales, holds the weekly build release, the monthly scorecard and, with Laurence Bergeron, the quarterly and the yearly reviews. Laurence Bergeron, the president, holds the quality lines of the scorecard, the yearly audit at the contract manufacturer and the yearly check of the molds at the molder. The staff installer keeps the in-house build cell, one hour a month. The contract manufacturer and the molder perform under agreements and hold no line of the routine. In general, one person holds each line with a named backup, and that is this book's suggestion. In a solo business the owner holds every line and a trusted advisor reads the register each quarter. In a larger business a purchasing lead holds the release and the scorecard. No hour and no amount is promised to a reader.

8.8 Automating the Records Without Automating the Decision

The routine has parts that a tool can carry and parts that only a person can carry, and the line between them is worth drawing carefully because it is easy to cross by accident.

What can be automated is the prompting and the collecting. A reminder can reach the person who set it at a set day or on a recurrence. A recurring task can carry a required observable end, a short procedure and a next due date. A count can be drawn from the returned test records rather than counted by hand, and a first list of the open releases can be assembled from the forecast. A first summary of the month's counts can be assembled automatically from the returned test records, and the head of sales reads it, corrects it and approves it; a reading is never made by a tool. None of those is a decision, and each of them saves minutes that the routine would otherwise spend on gathering.

What stays with a person is everything that decides. Reading the count stays with a person, because a count shows that a number moved and never why it moved. Choosing what the count means stays with a person, because a first-pass yield is a count of kits and not a verdict about a supplier. Deciding to keep, move or bring back an activity stays with the founders, because that decision rests on the grid, the scorecards and the triggers together. And every decision that moves a promise to a customer stays with a person, because a promise is not a count. Automation as a method is Book J7's (Business Process Automation), and workflows with human approval are Book N3's (Workflows, Automation and Human Approval); neither is retold here.

The pages of the platform are worth naming one by one, with their limits, because a business that builds a routine on a page should know what the page does. In the application copy of September 7, 2026, Control center → Plan tasks (/taches) holds recurring or one-off tasks with a "Done when" criterion that is required when the task is created, six statuses and a next due date, and the Mark as done button does not check that criterion (La Machine, 2026). Control center → Tools (inventory) (/outils) lists the installation's software tools with a name, a role, a category, a state, a cost and an identity measured by a script, a version, a location and a date, and it measures no dependence on a supplier of goods, no total cost and no volume (La Machine, 2026). Control center → Today (/aujourdhui) counts what falls due today in the plan and is never a production schedule (La Machine, 2026). Control center → Audit log (/journal) traces what happened to a record, never why, and is never a register of lots or a traceability of products (La Machine, 2026). La Machine computes no scorecard, sends no release to a supplier, reads no test record and keeps no register of sourcing decisions, and the reading stays with a person; check it in your installation. The case does not use La Machine.

TOOLS OF THE TRADE What any release, scorecard and reminder function must offer, whatever it is called

Eight functions carry a sourcing routine, and a business should look for them rather than for a product. The first is a reminder that reaches the person who set it, at a set day or on a recurrence. The second is a recurring task with a required observable end, a short procedure and a next due date. The third is a place to write a release and the confirmation received. The fourth is a way to record a count and a date for each reading, with the month it belongs to. The fifth is a register with a column for the owner, the trigger and the next review of each activity. The sixth is a view of what falls due this week. The seventh is a way to name a backup for each line. The eighth is an export in a form another tool can read, because the counts of the routine will be read somewhere else. No brand name belongs on this list and no price; the requirement is the function, and no tool is chosen for the case.

IN LA MACHINE A recurring review can be a plan task, an inventory can model what a business depends on, and nothing computes a scorecard

In the application copy of September 7, 2026, Control center → Plan tasks (/taches) holds a recurring review as a task with a required "Done when" criterion, and the Mark as done button does not verify that criterion (La Machine, 2026). Control center → Tools (inventory) (/outils) lists the installation's software tools with a measured version and date, and it measures no dependence on a supplier of goods. Control center → Today (/aujourdhui) counts what falls due today in the plan and is never a production schedule, and Control center → Audit log (/journal) traces what happened to a record, never why. La Machine computes no scorecard, sends no release to a supplier, reads no test record and keeps no register of sourcing decisions, so the reading stays with a person. Check it in your installation. La Machine is described and never sold, and the case does not use it.

8.9 What Breaks a Sourcing Routine

A sourcing routine breaks in ordinary ways, and none of them is dramatic. The causes below are this book's reasoning and are illustrative, no study read measures what breaks a routine, and no frequency of any cause in any business is printed. The first is that the person who holds the release goes away and no backup is named. The release is not sent, the week's kits are built to the version the contract manufacturer already had, and the drift is found at the next audit. The remedy is a named backup and a written template, so that the line can be held by somebody else on the day its holder is away.

The second break is a week skipped and the release sent late. The week that most needs the release is the week with the least time for it, and a release sent after the build has begun is a record of what happened rather than an instruction. The remedy is to keep the release through the busy weeks and to shorten something else instead. The third break is a month skipped and the scorecard with no page. A month with no page is a month whose counts are read late, and a drift read late is a drift that has already moved a price or a yield. The remedy is a page for every month, even a page whose only words are "no change".

The fourth break is a trigger read and not written. The condition is discussed, everybody agrees that it has not been reached, and nothing is recorded, so that the next quarter reads it from memory. The remedy is a reading of each trigger written into the minutes, reached or not. The fifth break is a review held and its decisions not written. A decision that lives in the memory of the two people who attended will be taken again next quarter on the same item. The remedy is one decision per open item, with an owner and a day.

The sixth break is a task marked done with nobody checking the criterion. The box is ticked, the page does not exist, and the routine appears to run while nothing happens. The remedy is the second reader: a task marked done is read by a person other than the one who marked it. The seventh break is the yearly audit forgotten until the agreement is renewed. The audit is the one line with no monthly rhythm behind it, so it is the line most easily postponed. The remedy is a date set once a year and a note filed when it is done. The eighth break is a supplier's own report replacing the company's reading. A report written by the supplier is the supplier's reading, and the company's reading is drawn from the records the agreement returns.

When a task is missed, the remedy is small and it is written down. The task is done on the next working day rather than folded into the next cycle. The miss is written down with its reason, because a miss that is not recorded is a miss that will be repeated. The cause of the miss is found, and it is read as a fact about the system rather than a person's fault. And where the miss belongs to the plan of Chapter 11, the gate is read again 7 days later, and nothing new starts until it passes.

On day 365 the scorecard shows 21 of 23 releases on time, one lot of 25 kits held for 4 days and released, and no trigger reached, all posed by this book. Nothing is measured and no result is credited to the routine. Chapter 9 reads who helps with this work and when to get help.

CASE IN POINT Hydrosentinelle inc. runs ten lines of routine for 89 hours a year and reads 21 of 23 releases on time on day 365

Hydrosentinelle inc. is a fictional business created for this encyclopedia; its founders, its employees, its customers, its partners, its suppliers, its figures and the events described are invented, and any resemblance to a real business is coincidental. The weekly build release, the monthly scorecard, the quarterly capability and trigger review, the yearly make-or-buy review, the audit at the contract manufacturer, the check of the molds at the molder and the upkeep of the in-house build cell take 89 hours a year, Karim Haddad 50, Laurence Bergeron 27 and the staff installer 12. On day 365 the scorecard shows 21 of 23 releases on time, one lot of 25 kits held for 4 days and released, a first-pass yield of 98.6%, and no trigger reached. All of it is posed by this book, illustrative and not a forecast, and no verdict and no result is credited to the routine.

Costly Mistakes

• Treating sourcing work as a rescue rather than a routine. A release written when somebody remembers and a scorecard read when a lot comes back held are a business reacting to its own supply instead of running it.
• Sending a release without its versions. A release that names the kits but not the version of the firmware image or of the end-of-line test protocol is a release that lets the week's build drift from what the company asked for.
• Reading a count as a grade. A first-pass yield is a count of kits and not a mark for a supplier, and a count read as a mark invites the work to bend toward the number instead of toward the flow.
• Skipping a quiet month of the scorecard. A month with no page is a month whose counts are read late, and a drift read late has already moved a price or a yield.
• Reading a trigger only when it fires. A trigger read only when it fires was never a control, and a quarter in which nothing fires is the quarter in which the reading matters most.
• Marking a task done without a second reader. The box being ticked and the criterion being met are two different facts, and a routine that confuses them appears to run while nothing happens.
• Letting a supplier's own report replace the company's reading. A report written by the supplier is the supplier's reading, and the company's reading is drawn from the records the agreement returns.

Key Takeaways

• Sourcing work is a routine of short reviews and a rescue is a break in the work: the routine is the small set of reviews that runs all year whether or not anything has gone wrong, so that fewer readings come late.
• The routine has five rhythms — each week, each month, each quarter, once a year and the monthly upkeep of the in-house build cell — and in the case they are the weekly build release, the monthly supplier scorecard, the quarterly capability and trigger review, the yearly make-or-buy review with the audit and the check of the molds, and the upkeep of the in-house build cell.
• The ten lines take 89 hours a year, Karim Haddad 50, Laurence Bergeron 27 and the staff installer 12, which is $5,160 at the printed illustrative rates, and every hour is posed by this book and illustrative; the routine of buying work of Book I8 (Freelancers, Agencies and Outsourcing, chapter 9 of Book I8), the routine of operations of Book J1 (Operations Management Fundamentals, chapter 8 of Book J1) and the routine of quality management of Book J4 (Quality Management and Continuous Improvement, chapter 9 of Book J4) keep their own cadence and are not counted again.
• Each task is written in the ten-field task format that Book N2 (Piloting Tasks, Executions and Projects, chapter 2) treats for the plan task, with a person, a trigger, an input, an output and an observable end, and the criterion is the part of the row that matters most.
• A tool can prompt and can collect, and a person decides: reading the count, choosing what it means, deciding to keep, move or bring back an activity, and every decision that moves a promise to a customer stay with a person, and a task marked done proves that a box was ticked and nothing more.
• A routine breaks in ordinary ways — no backup for the release, a week skipped, a month with no page, a trigger read and not written, a review whose decisions are not written, a task marked done unchecked, an audit forgotten, and a supplier's report read in place of the company's — and every remedy is a decision taken before the break.

Your Action Plan

1. List the recurring sourcing tasks and give each one a rhythm. Write the ten lines of the routine in the ten-field task format that Book N2 (Piloting Tasks, Executions and Projects, chapter 2) treats for the plan task, with a person, a trigger, an input, an output and an observable end for each. Deliverable: a written routine of five to ten lines. Verification: every line names a person, a trigger and a criterion a second person could check.
2. Cost the routine in hours a year and read it against the hours of the same people. Multiply the hours each time by the times a year for each line, add the lines, and write the total beside the hours the same people have for the work. Deliverable: a table of hours a year by line and by person. Verification: no person's year asks more than the hours the business has agreed for that role.
3. Write the release template and send the first release. Put the count of kits, the version of the firmware image, the version of the end-of-line test protocol, the due date and a line for changes since last week into one template, name who sends it and who is the backup, and ask for the written confirmation. Deliverable: a release template and one sent release with its confirmation. Verification: a release and a confirmation exist for the week, and the previous week's returned test records have been read against the release.
4. Set the reminders and name a backup for every line. Put the weekly release, the monthly scorecard, the quarterly review, the yearly review, the audit, the check of the molds and the upkeep of the in-house cell on dated entries, and write the name of the backup beside each line. Deliverable: a dated calendar of the routine with a backup for each line. Verification: each line has a person, a day and a backup, and the first cycle has been read against the criteria.
5. Write the three triggers and the rule for a missed line. Write the conditions that would bring an activity back in-house or move it, with the reading that decides, and write the rule for a missed line — done on the next working day, written down with its reason, the cause found, and the gate read again 7 days later where the miss belongs to the plan of Chapter 11. Deliverable: a written set of triggers and a written rule for a missed line. Verification: each trigger has a reading written into the minutes of a quarter, reached or not, and at least one miss has been recorded and read under the rule.

In La Machine

In the application copy of September 7, 2026, a recurring sourcing review can live in Control center → Plan tasks (/taches) as a task with a required "Done when" criterion, and the Mark as done button does not check that criterion (La Machine, 2026). Control center → Tools (inventory) (/outils) lists the installation's software tools with a measured version and date, and it can serve as a model for an inventory of what a business depends on and of what it chose to buy rather than make, for software only. Control center → Today (/aujourdhui) counts what falls due today in the plan and is never a production schedule, and Control center → Audit log (/journal) traces what happened to a record, never why. A business should check each page in its own installation.

What a business needs elsewhere is the rest of the routine. La Machine computes no scorecard, sends no release to a supplier, reads no test record and keeps no register of sourcing decisions, so the reading stays with a person. A register of sourcing decisions with an owner and a trigger for each activity, a place to record the monthly counts, and a place for the minutes of a review are chosen by function, and Chapter 10 sets out the functions to look for; this book is valid with any tool. The case does not use La Machine, and the pages above are one worked example of a business platform, described and never sold.

Self-Check

1. What is the difference between a routine and a rescue, and what are the five rhythms of a sourcing routine?
2. What does the weekly build release carry, and what happens when the contract manufacturer cannot meet it?
3. What does the monthly scorecard read, and where do its counts come from?
4. What does the quarterly review read, and why is a trigger read whether or not it has been reached?
5. What are the ordinary ways a sourcing routine breaks, and what is the remedy for each?

Answers. 1. A rescue is a recall of attention because something has already gone wrong, and a routine is the small set of reviews that runs all year whether or not anything has gone wrong, so that fewer readings come late; the five rhythms are each week, each month, each quarter, once a year and the monthly upkeep of the in-house build cell, and in the case they are the weekly build release, the monthly supplier scorecard, the quarterly capability and trigger review, the yearly make-or-buy review with the audit and the check of the molds, and the upkeep of the in-house build cell. 2. The release carries the count of kits for the week, the version of the firmware image, the version of the end-of-line test protocol, the due date and a line for the changes since last week or the words "no change"; when the contract manufacturer cannot meet it, the contract manufacturer calls the same day, the lot is written into the weekly note with its reason, and the reading of the trigger on late releases is taken at the next quarterly review. 3. The scorecard reads the first-pass yield at the end-of-line test, the lots held and for how many days, the releases on time, the price per kit against the trigger, and the share of the sensor component bought from the second source; its counts come from the end-of-line test records of each supplier-built kit, which come back to the company every day under the supply agreement, so the counts are drawn from records the company owns. 4. The quarterly review reads the capacity of the contract manufacturer, of the molder and of the second source against the scorecards of the quarter, and reads the three written triggers; a trigger is read whether or not it has been reached because a trigger read only when it fires was never a control, and a trigger reached leads to a decision and never to an automatic move. 5. The breaks are a release without a backup, a week skipped, a month with no page, a trigger read and not written, a review whose decisions are not written, a task marked done with nobody checking the criterion, an audit forgotten until the agreement is renewed, and a supplier's own report read in place of the company's; the remedies are a named backup and a written template, keeping the release through the busy weeks, a page for every month even with the words "no change", a reading of each trigger written into the minutes, one decision per open item with an owner and a day, a second reader for every task marked done, a date set once a year with a note filed, and a reading drawn from the records the agreement returns.

Chapter References

La Machine. (2026). Documentation — route reference inventory. Internal product documentation, read October 16, 2026, together with the code of the pages this book describes: /taches, /outils, /aujourdhui and /journal.

Chapter 9 — Expertise, Roles and When to Get Help

Make-or-buy work needs a few roles done well and a few kinds of outside reading engaged at the right moment, so this chapter says which roles stand behind it, when to get help and how to read an advisor's proposal, then applies all of it to Hydrosentinelle inc., which engages an outside lawyer for 5 hours and an accountant for 2, names five kinds of help without engaging them and hires no one (posed by this book).

What You'll Be Able to Do

• Name the seven roles that stand behind a small firm's sourcing decisions, and say what each one holds and what each one does not decide.
• Keep the boundary yourself, by holding the grid and its weights, the verdicts and their reasons, the triggers and the means to bring an activity back.
• Write the company's own agenda before a supplier review or an advisor's engagement, so that the meeting is read against a written need rather than against a proposal.
• Describe the form of pay of an outside reader — a fixed fee, a day rate, an hourly rate, a retainer or a mandate priced by scope — without printing an amount, and say what each form buys.
• Read an advisor's proposal on sourcing against seven questions, from the scope to the outcome claims, and check each answer against the company's own written need.
• Say plainly what outside help does not do: it reads, tests and examines, and the decision to make, buy or contract stays with the company.

Why It Matters

A sourcing decision is made once and lived with for years, and the work around it is small enough that a small firm can hold it with two or three people. What makes the work hard is not its volume but its edges. The owner has to decide which activities to own, which to have made and which to contract, and that decision needs a grid, a cost comparison and a set of triggers. The operations side has to keep the bench, the protocol and the records that would let an activity come back. The commercial side has to hold the quotes, the terms and the scorecard. And at a few moments — before an agreement is signed, before a cost comparison is trusted, before a certification question is answered — the company needs a reading that it cannot give itself.

The reason this chapter exists is that the moment of asking is easy to get wrong in both directions. A firm that asks too late signs an agreement it has not read, trusts a cost comparison it has not checked, or discovers a certification question after the product is built. A firm that asks too early, or asks the wrong party, hands over a decision that was always its own: it lets a supplier's proposal set the scope, lets an advisor's framework set the verdict, or lets a program's eligibility rule stand in for a strategy. Neither failure is dramatic, and both are ordinary. The remedy is not more help; it is a small number of named roles, a written need, and a clear line between what an outside reader does and what the company decides.

The cost of getting the roles right is modest and it is worth stating in the case's own terms. Hydrosentinelle inc. is a fictional company created for this encyclopedia, and the plan of this book takes 121 hours: Laurence Bergeron, the president, works 62 hours, Karim Haddad, the head of sales, works 36, the staff installer works 16, the accountant works 2 and the outside lawyer works 5, all posed by this book. The two outside readers are engaged for named jobs and for a stated number of hours, and no one is hired. That is the shape this chapter describes: the company keeps the decisions, buys two readings at the moments they are needed, and names five other kinds of help without engaging them. Nothing in the case is credited to any help, and no source read measures what outside help returns.

In This Chapter

Section 9.1 sets out the seven roles behind make-or-buy work and the eighth group that sits outside the firm, the suppliers, with a table of who holds each role in the case. Section 9.2 covers the owner as the keeper of the boundary and the four things the owner holds before any help is engaged. Section 9.3 covers the operations and production leads and the bench that keeps an activity able to come back. Section 9.4 covers the purchasing and supplier manager, the forms of pay for outside sourcing help, and the two pages of La Machine that touch this ground. Section 9.5 covers quality, product safety and certification help. Section 9.6 covers legal help on supply, manufacturing and acquisition agreements, with the competition rules and the contract of enterprise called to the books that hold them. Section 9.7 covers the accountant and the financial advisor and the reading of the cost comparison. Section 9.8 covers the strategy advisor and publicly funded help for manufacturers. Section 9.9 gives seven questions to put to an advisor's proposal on sourcing.

9.1 The Roles Behind Make-or-Buy Work

Seven roles stand behind a small firm's sourcing decisions, and an eighth group sits outside the firm. The seven are the owner as the keeper of the boundary, the operations and production lead, the purchasing and supplier manager, quality and certification help, legal help, the accountant and the financial advisor, and the strategy advisor with publicly funded help. The eighth group is the suppliers themselves, which perform under agreements and are not advisors. This division is the book's reasoning and not a finding; no source read measures how a small firm divides this work, how many roles it fills, or how often it engages anyone from outside.

The owner as the keeper of the boundary is the role that holds the decision. The owner signs, activity by activity, the verdict to make, buy or contract, and the owner holds the grid and its weights, the reasons written beside each verdict, the triggers that would bring an activity back, and the means to bring it back. No outside reader holds any of those. A lawyer reads an agreement, an accountant reads a cost comparison, and an advisor tests a reasoning, and none of them decides which activities the company owns. Section 9.2 takes this role in full.

The operations and production lead is the role that keeps the work the company has chosen to keep. In a firm that assembles or finishes a product, this role holds the bench, the build cell, the protocol and the records, and it reads the forecast against the hours available. The roles of operations are Book J1 (Operations Management Fundamentals, chapter 9 of Book J1), and this chapter cites that treatment rather than repeating it. Section 9.3 takes this role in full.

The purchasing and supplier manager is the role that holds the relationship with a supplier. This role reads the quotes, holds the commercial terms, keeps the scorecard and writes the company's agenda before each review, so that the review is read against a written need rather than against whatever the supplier brings. Book J5 is named here and treated nowhere in this chapter, and the buying of outside work and its control are Book I8's (Freelancers, Agencies and Outsourcing, chapter 10 of Book I8). Section 9.4 takes this role in full.

Quality, product safety and certification help is the role that reads the test protocol, the first-article inspection and the records against the standards that apply to the product. In a small firm this can be a reading bought for a named job rather than a person on staff. Quality management, its standards and certifications and its roles are Book J4's (Quality Management and Continuous Improvement, chapters 6 and 10 of Book J4), and this chapter cites that treatment rather than repeating it. Section 9.5 takes this help in full.

Legal help is the role that reads the supply and quality agreement, the tooling agreement and, if a purchase is ever considered, the acquisition agreement before they are signed. Contracts are Book H1's (Contracts and Commercial Agreements), and whether a term bears on competition is Book H4's (Consumer Protection, Advertising Law, Competition and Language Requirements). Section 9.6 takes this help in full.

The accountant and the financial advisor is the role that reads the cost comparison, the premium, the break-even and the present value that Chapter 7 builds, and checks that the two sides of the comparison are on the same basis. The owner decides what the numbers mean for the business. Section 9.7 takes this role in full.

The strategy advisor and publicly funded help is the role that tests the owner's reasoning about the boundary and points to programs that a manufacturer may be able to apply to. Strategy and competitive positioning are Book A3's (Strategy and Competitive Positioning, chapter 2 of Book A3), and public programs and their applications are Book G7's (Grants, Subsidies and Public Programs). Section 9.8 takes this help in full.

The eighth group is the suppliers. A contract manufacturer, a molder and a second source perform under agreements; they are parties to a contract and not advisors, and no part of this chapter treats a supplier as a source of advice about the boundary. The case's contract manufacturer, its molder and its second source are described by kind and are never named.

Roles and delegation are Book I2's (Organizational Design, Roles, Delegation and Accountability), and hiring is Book I3's (Recruiting, Hiring and Onboarding); the case hires no one, and this chapter cites both books rather than repeating them.

Figure 9.1 sorts the roles behind the case's make-or-buy work into the people of the company, the outside help engaged, the suppliers who perform and the help named but not engaged. Inside the company sit Laurence Bergeron, the president, with the test protocol, the pilot builds, the molds, the second source and the audits, 62 hours in the plan; Karim Haddad, the head of sales, with the forecast, the quotes, the commercial terms, the weekly release and the scorecard, 36 hours; and the staff installer, with the in-house build cell, the pilot support and the installations the freed hours return to, 16 hours. Outside help engaged by the plan is the outside lawyer, who reads the supply and quality agreement and the tooling agreement for 5 hours at Book F9's illustrative rate, and the accountant, who reads the cost comparison at the first gate for 2 hours at Book G7's illustrative rate. Five kinds of help are named and not engaged, drawn in dashed boxes: a quality or certification consultant, a purchasing or supplier-development advisor, a strategy advisor, an advisor of a public program for manufacturers, and a broker of businesses for sale. A note records that the contract manufacturer and the molder perform under agreements and are not advisors, and that no hire is made.

Figure 9.1 — Who does the make-or-buy work. Three people of the company, an outside lawyer for 5 hours and an accountant for 2; suppliers perform under agreements; five kinds of help named and not engaged; no hire (posed).

The table below sets eight roles beside what each one holds and who holds it in the case. The case is posed by this book, and the last row records the suppliers, which are parties to an agreement and not advisors. Read the third column as the distribution of the work in one small firm: the roles are the same in every business, and what changes is how many people carry them.

Table 9.1 — The roles behind make-or-buy work and who holds each in the case (posed)

DEFINITION Role, outside help, acceptance criterion and the keeper of the boundary, as this book uses the words

These are this book's working words, and no source read defines them in this sense. A role is a set of tasks that one person holds, whether the person is a founder, an employee or an outside helper. Outside help is a skill or a reading that the company buys for a named job and a period. An acceptance criterion is the written condition that a deliverable must meet before it is accepted. The keeper of the boundary is the owner who signs, activity by activity, the verdict to make, buy or contract. A role is not a title, outside help is not a decision, an acceptance criterion is not a hope, and the keeper of the boundary is not an advisor. The words are this book's own, and the case's hours are posed by this book.

KEY CONCEPT Outside help buys a skill or a reading, and the decision stays with the company

A lawyer reads an agreement, an accountant reads the cost comparison, an advisor tests a reasoning and a consultant examines a process, and none of them decides which activities the company owns. The company writes the need, the acceptance criterion and the decision, and reads what it receives. This is the book's reasoning and never a finding: no source read measures what outside help returns, what it costs, how many small firms engage a lawyer, an accountant or a consultant for sourcing, or how long an engagement lasts. The case's figures are the case's own, posed by this book: two readers engaged for named jobs, 5 hours of the outside lawyer and 2 hours of the accountant, and no hire. A proposal is the advisor's own offer, and no outcome claim is printed here.

9.2 The Owner as the Keeper of the Boundary

In a small firm the owner decides which activities to own, which to have made and which to contract, and no outside helper holds those decisions. This is the book's reasoning and not a finding. The reason the role cannot be delegated is not that the owner knows best; it is that the decision binds the company's capital, its capacity and its customer promise at once, and the person who carries the consequences is the person who should sign. An advisor can widen the set of options and can point out a cost that the company has not counted, and the choice among the options remains the owner's.

The owner holds four things personally before any help is engaged. The first is the grid and its weights: the criteria, the weights given to each, and the scores that follow. The second is the verdicts and their reasons: not only the word make, buy or hybrid, but the sentence that says why, written so that a reader two years later can see what the company believed. The third is the triggers that would bring an activity back: the defect rate, the price, the late releases, or whatever else the company decides would tell it that the verdict has stopped holding. The fourth is the means to bring the activity back: the bench, the protocol and the records that would let the company resume the work if a trigger were reached. A verdict without a trigger is a decision that will be reviewed only when something has already gone wrong, and a trigger without the means to act on it is a warning that the company cannot answer.

The case shows the four things in the order they were written. Laurence Bergeron and Karim Haddad write and sign the verdicts of the grid on day 28, at the first gate of the plan, and the verdicts cover the six activities of the chain: the firmware and the alert logic are made, the enclosure tooling for the second generation is hybrid, the electronic boards are bought, the final kit assembly and the end-of-line test are hybrid, the installation in buildings is hybrid, and the cloud hosting of the monitoring service is bought. The triggers are written at the gate of day 90, beside the decision, so that the verdict and the condition that would reverse it live in the same place. The means to bring assembly back are the in-house build cell that the staff installer keeps, the end-of-line test protocol that Laurence Bergeron wrote, and the duplicate test bench that the contract manufacturer runs the company's test on. All of this is posed by this book.

The plan that carries the work takes 121 hours: Laurence Bergeron works 62 hours, Karim Haddad works 36, the staff installer works 16, the accountant works 2 and the outside lawyer works 5, all posed by this book. The two founders hold 98 of those hours between them, which is the shape of a small firm: the boundary work is the founders' work, and the outside readings are small and named. The hours are illustrative, they are the case's own, and they are never a norm or a benchmark for any reader.

The owner's role also has a limit worth stating. The owner holds the decision, and the owner does not hold every reading that the decision needs. A cost comparison that the company has built itself can be wrong in the same way twice, because the same person makes the same assumption. An agreement that the company has written itself can miss a term that a reader trained on agreements would see. The four things the owner holds are the things that no one else can hold; the readings the owner buys are the things that a second pair of eyes does better. Keeping the two apart is the whole of this chapter.

WHO DOES THIS Two founders keep the boundary, two outside readers are engaged for named jobs, and no one is hired

The roles below are the case's, posed by this book. Laurence Bergeron, the president, holds the test protocol, the bench, the pilot builds, the molds, the second source and the audits, 62 hours in the plan. Karim Haddad, the head of sales, holds the forecast, the quotes, the commercial terms, the weekly release and the scorecard, 36 hours. The staff installer keeps the in-house build cell, 16 hours. The outside lawyer reads the supply and quality agreement and the tooling agreement, 5 hours at Book F9's illustrative rate. The accountant reads the cost comparison, 2 hours at Book G7's illustrative rate. The contract manufacturer and the molder perform under agreements and are not advisors. No one is hired. In general, the owner signs the verdicts and a second person reads the cost comparison, which is this book's suggestion and not part of the model. In a solo business the owner holds every role and engages a lawyer and an accountant for named jobs; in a larger business a purchasing lead and a quality lead hold their lines. No hour and no amount is promised to a reader.

9.3 Operations and Production Leads

The operations lead runs the workshop or the line that the firm keeps, the build cell and its bench, and the production lead reads the forecast against the hours. This is the book's reasoning and not a finding. The two are different jobs even when one person holds both: the first is about the place, the equipment and the people who do the work, and the second is about the arithmetic that says whether the coming period fits. In a small firm the same person can hold both, and in a larger firm they can be two roles with two calendars.

The roles of operations are Book J1 (Operations Management Fundamentals, chapter 9 of Book J1), and this chapter cites that treatment rather than repeating it. What matters here is the part of the operations role that belongs to the boundary: the bench, the protocol and the records are not only the means of production, they are the means by which an activity that has been contracted out can be brought back. A company that sells its bench, drops its protocol and stops keeping its records has made a decision that is expensive to reverse, whether or not it wrote that decision down.

In the case, the staff installer keeps the in-house build cell for repairs, pilot builds and urgent kits, and Laurence Bergeron holds the end-of-line test protocol and the duplicate test bench, all posed by this book. The build cell is small and it is kept on purpose: it is what makes the hybrid verdict on final assembly and end-of-line test a real choice rather than a word. The duplicate bench is the second copy of the test equipment, delivered to the contract manufacturer so that the company's own test runs on the company's own equipment at the supplier's site. The protocol is the written sequence that the test follows, and it is what makes the test the same in both places.

The case hires no operations lead. The work is held by the staff installer and by Laurence Bergeron, and the hours are the hours of the plan. A solo business divides the same work differently, and a larger business divides it again: the book's suggestion is that one person holds the bench and one person reads the forecast, with a named backup for each, and that suggestion is not a finding and not part of the case.

The operations role also has a boundary of its own. The operations lead decides how the work is done inside the company's own walls, and the operations lead does not decide which work stays inside them. A production lead who is asked to make a make-or-buy decision will answer from the bench's point of view, which is the point of view the bench can see; the cost comparison, the triggers and the verdict belong to the keeper of the boundary. Keeping the two apart is what stops a capacity problem from being read as a strategy.

La Machine touches this ground only lightly: a recurring review of the bench can live in the plan tasks of the platform, and the platform measures no load, no capacity and no delay. Check it in your installation. The case does not use La Machine.

METHOD Choosing who holds each activity in five steps

First, list the activities of the chain with their verdicts, from the grid of Chapter 3. Second, name one person for each activity and one backup, because an activity whose holder is away is an activity whose line is not kept. Third, write for each activity the company's agenda for the next review, before the supplier or the advisor sets one, so that the review is read against a written need. Fourth, write what the holder may decide alone and what goes to the owner, such as signing a new agreement or changing a verdict. Fifth, after a quarter, read the register against what was decided and change the holder if the line is not kept. This is the book's reasoning and not a tested procedure, and the case follows these steps with Karim Haddad and Laurence Bergeron as the holders, posed by this book.

9.4 Purchasing and Supplier Managers

The person who holds the relationship with a supplier reads the quotes, holds the commercial terms and the scorecard, and writes the company's agenda before each review. This is the book's reasoning and not a finding. The agenda is the part that matters most and the part that is easy to skip: a review that opens with the supplier's presentation is a review that will be about the supplier's subjects, and a review that opens with the company's written list is a review that will be about the company's.

Book J5 is named here and this chapter does not treat it. The buying of outside work and its control are Book I8's (Freelancers, Agencies and Outsourcing, chapter 10 of Book I8), and this chapter cites that treatment rather than repeating it. Whether a given engagement is a service contract or employment is Book H7's (Employment Law Essentials (Québec and Canada)) and a lawyer's, and this book decides nothing about it.

The forms of pay for outside purchasing and sourcing help are described here by form and never by an amount. A fixed fee buys a scope that has been written down before the work starts. A day rate buys a person's time for a day at a time, with the scope agreed in outline. An hourly rate buys time in smaller units, which suits a reading that cannot be sized in advance. A retainer buys availability over a period, whether or not the work arrives. A mandate priced by scope buys a defined piece of work whose price is set once the scope is known. Each form shifts a different risk: the fixed fee and the mandate priced by scope put the risk of overrun on the advisor, the day rate and the hourly rate put it on the company, and the retainer puts it on the company whether or not the work arrives.

In the case, Karim Haddad holds the quotes, the commercial terms, the weekly build release and the monthly scorecard, and no purchasing manager is hired, all posed by this book. A purchasing or supplier-development advisor is named and not engaged. The two outside readers of the plan are paid by the hour at the printed illustrative rates of Books F9 and G7, and those rates are illustrative and are not a pay.

Two pages of La Machine touch this ground, and both have limits that the box at the end of this section names. The platform names no purchasing manager and keeps no register of the roles of a sourcing relationship. The case does not use La Machine.

The table below sets five forms of pay beside what each one buys and what to write down when it is used. No amount is printed, and the case's two readers are paid by the hour at the printed illustrative rates of Books F9 and G7. Read the third column as the part that protects the company: a form of pay that is not written down is a form of pay that will be argued about later.

Table 9.2 — The forms of pay for outside sourcing help

TOOLS OF THE TRADE What any workspace shared with an outside reader must offer, whatever it is called

Eight functions carry a shared workspace, and a business should look for them rather than for a product. The first is a place for the written need and the acceptance criterion, so that the reader knows what the deliverable must meet. The second is a list of tasks with a status that both sides can read. The third is a place where the agreement under review can be read and annotated. The fourth is access that is given to a named person and that can be removed the same day. The fifth is a trace of what was changed and by whom. The sixth is a place for the handover documents, so that the work does not leave with the person. The seventh is an export of everything in a form another tool can read. The eighth is a way to see what is due, so that an engagement does not run on past its end. No brand name belongs on this list and no price; the requirement is the function.

IN LA MACHINE The assistant reads tasks, a company page can record a supplier, and neither reads a quote or a contract

In the application copy read on October 16, 2026, Control center → AI Agents (/agents-ia) reads tasks, projects, alerts and alarms and can create a plan task after a click; it reads no quote and no supplier contract, and it decides nothing (La Machine, 2026). CRM → Companies (business/marketing/companies) holds one record for each company, with its name, sector, size, site, status and notes; a reader can record a supplier, a molder or an installer there as a company, and the record has no field for a contract, a price, a tooling or a performance (La Machine, 2026). So each role of this chapter is held by a person, and the assistant only prompts. The platform names no purchasing manager, keeps no register of the roles of a sourcing relationship, reads no quote and no contract and decides nothing. Check it in your installation. La Machine is described and never sold, and the case does not use it.

9.5 Quality, Product Safety and Certification Help

An advisor reads the test protocol, the first-article inspection and the records against the standards that apply to the product. This is the book's reasoning and not a finding. The reading is worth buying at two moments: before a product is built, when the protocol can still be changed cheaply, and before a supplier is qualified, when the records that will be kept can still be agreed. After both, the same reading costs more and changes less.

Quality management, its standards and certifications and its roles are Book J4's (Quality Management and Continuous Improvement, chapters 6 and 10 of Book J4), and this chapter cites that treatment rather than repeating it. What belongs to this chapter is the boundary question inside the quality question: which parts of the quality work stay with the company when an activity moves to a supplier, and which parts move with the work. The answer this book gives is that the protocol and the acceptance criterion stay with the company, the execution moves, and the records are kept in a form the company can read.

The certification of a radio device built by a contract manufacturer is not read for this book, so this section prints no rule on it. A company that builds a device with a radio in it should ask the certification body and a professional about the requirements that apply to its own product, and should do so before the design is frozen rather than after. This book describes methods and sources and is not a certification guide, and it recommends no body and no advisor.

The risk that a company which assembles kits carries a component that is missing or defective is Book G8's (Business Insurance and Risk Financing, chapter 3 of Book G8), printed there and cited here rather than recounted. The same book names the workshop, the stock of detectors and components and the test bench among the company's exposures, and rates a supplier's component shortage that stops kit assembly at $40,000 and 30 days, reduced by a second source and a safety stock. Those figures are Book G8's and are not recomputed here.

In the case, Laurence Bergeron writes the end-of-line test protocol and inspects the first article of each pilot build, the contract manufacturer runs the company's test on the company's duplicate bench, and a quality or certification consultant is named and not engaged, all posed by this book. The first-article inspection is the reading of the first unit of a build against the protocol, and it is the point at which a supplier's process is either accepted or sent back. The duplicate bench is what makes the test the same at both sites. No professional title is named in this section, and the case engages no quality advisor.

The quality role has a limit that is easy to miss. A protocol that the company writes and a supplier runs is only as good as the records that come back, and records that arrive in a form the company cannot read are records the company does not have. The company should decide, before the first build, what it will receive after each build and in what form, and should write that into the agreement. That is a boundary decision as much as a quality decision, and it belongs to the keeper of the boundary as much as to the quality reader.

WATCH OUT Choosing help by its label and signing an open-ended mandate are two traps

Two traps catch a company that is buying a reading, and each has a question that catches it. The first is choosing help by its label: a title or a brand stands in for the skill that the job needs, and the engagement starts before anyone has asked what the person has read or built that resembles the job. The question is what has this person read or built that resembles the job, and what is the acceptance criterion. The second is signing an open-ended mandate: a fee with no scope runs on, the deliverable is never accepted, and the engagement has no end. The question is what is the scope, the form of pay and the day on which the engagement ends. This book's reasoning is illustrative, and no frequency of either trap in any business is printed. The case engages two readers for named jobs, 5 hours and 2 hours, posed by this book.

9.6 Legal Help on Supply, Manufacturing and Acquisition Agreements

A lawyer reads the supply and quality agreement, the tooling agreement and, if a purchase is ever considered, the acquisition agreement before they are signed, and the decision stays with the company. This is the book's reasoning and not a finding. The reading is not a formality and it is not a guarantee: a lawyer reads the agreement and the law and advises the company, and only the Competition Tribunal or a court decides whether a given agreement is lawful. A company that treats a lawyer's reading as a decision has moved a decision it owns to a person who does not own it.

Contracts are Book H1's (Contracts and Commercial Agreements), and the contract of enterprise or for services of the Civil Code of Québec (CQLR c CCQ-1991, arts. 2098, 2103, 2125, 2126 and 2129, Québec, up to date to August 12, 2026, read on October 15, 2026) is called to Book H1: article 2098 defines the contract of enterprise or for services, article 2103 draws the line between it and a sale where the work is merely accessory to the value of the property supplied, article 2125 gives the client the right to resiliate unilaterally even while the work is in progress, article 2126 lets the contractor resiliate only for a serious reason and never at an inopportune moment, and article 2129 says what the client pays on resiliation. This book applies none of those articles to a real contract, and a company with a question about its own agreement should ask a lawyer.

The Competition Act (R.S.C. 1985, c. C-34, ss. 45, 77, 90.1, 91 and 92, Canada, current to September 30, 2026, last amended March 26, 2026, read on October 16, 2026, the consolidation date being read again on the day this chapter is written) is called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements): section 45 makes it an offense for a person, with a competitor in respect of a product, to agree to fix the price, to allocate sales, territories, customers or markets, or to fix, control, prevent, lessen or eliminate production or supply, with an ancillary-restraint defense; section 77 covers exclusive dealing and tied selling under subsection 77(2) and market restriction under subsection 77(3), which the Tribunal may act on only where the practice of a major supplier or a widespread practice is likely to lessen competition substantially, the reasonable period of paragraph 77(4)(a) reaching exclusive dealing and market restriction and never tied selling; section 90.1 allows a civil order where an agreement between persons of whom two or more are competitors prevents or lessens competition substantially, and even between non-competitors where a significant purpose is to do so; section 91 defines a merger, which includes the acquisition of control over or a significant interest in a business of a supplier or a customer; and section 92 is the order on a merger of any size. This book applies none of those provisions to a real agreement, and only the Tribunal or a court decides whether a given agreement or merger is lawful.

The guidelines of the Bureau are its approach and never the law. The Competitor collaboration guidelines (Competition Bureau, 2021), the page read on October 15, 2026, are dated May 6, 2021, the page was modified June 27, 2024, and a notice says that the Bureau is reviewing them after recent amendments to the Act. The guidelines say that purely vertical agreements between a supplier and its customer will be assessed under the civil provisions of Part VIII rather than under section 45, and, for section 90.1, they give market-share levels as a general rule; that those levels are not safe harbors is this book's reading and not a word of the page. The page predates subsections 90.1(1.01) and (1.3) of the Act read today, so the guidelines are never the current text of the Act, and the Act is called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements).

The thresholds decide only who must notify. The Bureau's news release of March 2, 2026 (Competition Bureau, 2026), read on October 15, 2026, says that as of 2026 the transaction-size threshold for pre-merger notification stays at $93 million, measured on the assets in Canada of the business acquired or the revenues they bring from sales in, from or into Canada, and that notice is also generally required when the parties and their affiliates exceed $400 million in combined assets or revenues. The release does not say that a merger below the thresholds escapes review, and section 92 applies to a merger of any size.

Buying a business lies outside this book, and Book M6 (Buying a Business) is only named here; a broker of businesses for sale is named and not engaged. The ownership of what an outside party makes is Book H5's (Intellectual Property), called here rather than treated.

In the case, the outside lawyer reads the supply and quality agreement before it is signed on day 57 and the tooling agreement signed on day 58, 5 hours at Book F9's illustrative rate of $300 an hour, 4 in days 31 to 60 and 1 in days 61 to 90, all posed by this book. The case acquires nothing, and no question of notification is raised for it.

QUÉBEC & CANADA NOTE A contract of enterprise and the competition rules are called to Books H1 and H4, and this book recommends no vendor

The Canadian point of this section is that two bodies of published rules touch a supply or tooling agreement, and both are called to the books that hold them. The Civil Code of Québec (CQLR c CCQ-1991, arts. 2098, 2103, 2125, 2126 and 2129, Québec, up to date to August 12, 2026, read on October 15, 2026) is called to Book H1 in one sentence: the contract of enterprise or for services, the line between it and a sale, and the client's unilateral resiliation and its price. The Competition Act (R.S.C. 1985, c. C-34, ss. 45, 77, 90.1, 91 and 92, Canada, current to September 30, 2026, read on October 16, 2026) is called to Book H4 in one sentence. Only the Tribunal or a court decides whether an agreement or a merger is lawful. This book explains published rules and recommends no vendor, and it sends the reader to Book H1, Book H4 and a lawyer. Nothing here says whether the reader's own agreement complies.

THE RULE A lawyer reads the supply and quality agreement and the tooling agreement before they are signed, and the decision stays with the company

The rule is this book's own for the case, drawn from its reasoning and not from a source. Before the company signs an agreement with a supplier or a molder, a lawyer reads it, and contracts are Book H1's (Contracts and Commercial Agreements). Whether a term bears on competition is Book H4's (Consumer Protection, Advertising Law, Competition and Language Requirements), and only the Tribunal or a court decides whether a given agreement is lawful. The decision to sign stays with the company. The rule is neither a standard nor a legal duty, and it is not a substitute for advice on a given agreement. The case's outside lawyer reads the supply and quality agreement before day 57 and the tooling agreement before day 58, 5 hours in all at Book F9's illustrative rate of $300 an hour, posed by this book. No figure beyond the case's is printed here.

WHAT CHANGED The regulator's guidelines of May 6, 2021 are under review, and the Act is read again on the day of writing

The dated change comes from the page read on October 15, 2026: the Competitor collaboration guidelines (Competition Bureau, 2021) are dated May 6, 2021, and the page was modified June 27, 2024. A notice says that the Bureau is reviewing them after recent amendments to the Act. The page predates subsections 90.1(1.01) and (1.3) of the Act read on October 16, 2026, so the guidelines are never the law and never the current text. The Competition Act was read again on October 16, 2026, the day this chapter is written, current to September 30, 2026, and is called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements). The dates of sources are never dates of the case, which is counted in days and never in calendar years or months. The application code of La Machine was read on October 16, 2026 (La Machine, 2026).

9.7 Accountants and Financial Advisors for the Numbers

The accountant reads the cost comparison, the premium, the break-even and the present value that Chapter 7 builds, and the owner decides. This is the book's reasoning and not a finding. The reading has one job that the company cannot easily do for itself: checking that the two sides of the comparison are on the same basis. A comparison that puts a supplier's price beside the company's direct materials, and leaves out the company's own hours, its consumables and its oversight, will show a saving that does not exist. A comparison that puts a supplier's price beside the company's full cost, and leaves out the freight and the oversight that buying adds, will show a loss that does not exist either.

The reading of costs is Book G3's (Budgeting, Cost Control and Profitability, chapters 2 to 4 of Book G3), and ratios are Book G4's (Financial Analysis, Ratios and KPIs); this chapter cites both rather than repeating them. What belongs to this chapter is the division of labor: the company builds the comparison and writes its assumptions down, the accountant reads it and says which assumptions are doing the work, and the owner decides what the numbers mean for the business. An accountant who is asked to choose between making and buying has been asked a question that belongs to the keeper of the boundary.

The forms of pay of Table 9.2 apply to this reading as they do to any other, and they are described by form and never by an amount. A reading of a cost comparison can be small enough to be bought by the hour, and it is worth buying before the verdict is signed rather than after.

In the case, the accountant reads the cost comparison on day 30, 2 hours at Book G7's illustrative rate of $120 an hour, at the first gate of the plan, posed by this book. The comparison itself is the company's: the kit made at $175.50 and bought at $202, the premium of $26.50 a kit, the bench hours against the free hours, the duplicate bench at $9,200 and the molds at $72,000, all posed by this book. The owner keeps the grid and the verdicts, and the accountant's reading does not change them. Nothing in the case is credited to the accountant's reading.

The financial role has a limit that is worth stating plainly. An accountant reads what the company brings, and a comparison built on a wrong assumption will be read as a comparison built on a wrong assumption only if the assumption is written down. The company should write its assumptions beside its numbers, so that the reading has something to check. That is the same discipline the grid asks for: a verdict with its reason written beside it.

DO THE MATH The plan's 121 hours and what the two outside readers take, line by line

Illustrative figures; the case is fictional. Step 1 — Laurence Bergeron works 20 hours in days 1 to 30, 30 in days 31 to 60 and 12 in days 61 to 90, which is 62 hours. Step 2 — Karim Haddad works 14, 12 and 10 hours, which is 36 hours, and the staff installer works 4, 8 and 4 hours, which is 16 hours. Step 3 — the accountant works 2 hours in days 1 to 30, at Book G7's illustrative rate of $120 an hour. Step 4 — the outside lawyer works 4 hours in days 31 to 60 and 1 in days 61 to 90, which is 5 hours, at Book F9's illustrative rate of $300 an hour. Step 5 — the lines add up to 121 hours, which is $8,340 at the printed rates, and with the duplicate test bench at $9,200 and the molds at $72,000 the plan commits $89,540. The hours, the bench and the molds are posed by this book, the rates are illustrative and not a pay, nothing is credited to any help, and the sums are exact.

9.8 Strategy Advisors and Publicly Funded Help for Manufacturers

A strategy advisor tests the owner's reasoning about the boundary and does not decide it. This is the book's reasoning and not a finding. The useful form of the reading is an argument: the advisor takes the company's verdicts and their reasons and asks what would have to be true for each one to be wrong. A reading that produces a list of options without an argument about the company's own reasons has not done the work.

Strategy and competitive positioning are Book A3's (Strategy and Competitive Positioning, chapter 2 of Book A3), and this chapter cites that treatment rather than repeating it. The value chain as a map of what a business could own, make or buy is Book A3's as well, and this chapter does not redraw it.

Public programs and their applications are Book G7's (Grants, Subsidies and Public Programs), called here rather than treated. No program page is read for this book, so this section names no program, states no eligibility rule and promises no funding of a sourcing change. A program of this kind is a source of advice to ask about and never a promise of funding, and a company that plans around a program it has not read has planned around a hope. A company that wants to know what a program covers should read the program's own page and, where the page is unclear, ask the program office.

The two readings are different in kind and are worth keeping apart. A strategy advisor is bought by the company and answers to the company. A publicly funded advisor is provided under a program and answers to the program's terms as well. Both can be useful, and neither decides the boundary. The owner who has written the grid, the verdicts and the triggers has something to put in front of either reader, and that is what makes the reading worth buying.

In the case, a strategy advisor and an advisor of a public program for manufacturers are named and not engaged, and the case applies for nothing in this book, all posed by this book. The five kinds of help named and not engaged are the quality or certification consultant of section 9.5, the purchasing or supplier-development advisor of section 9.4, the strategy advisor of this section, the advisor of a public program for manufacturers of this section, and the broker of businesses for sale of section 9.6. Naming them without engaging them is itself a decision: it says which readings the company believes it may need and which it does not need yet.

The limit of this section is the limit of every section of this chapter. An advisor can widen the set of options, can point out a cost the company has not counted, and can ask what would have to be true for a verdict to be wrong. An advisor cannot sign the verdict, cannot hold the triggers and cannot keep the bench. Those stay with the company.

9.9 Reading an Advisor's Proposal on Sourcing

A proposal is the advisor's own offer and is read against the company's written need. This is the book's reasoning and not a finding. The order matters: a company that writes its need first reads a proposal as an answer to a question, and a company that reads a proposal first can end by adopting the proposal's framing of the question. The written need is short — the activity, the decision to be informed, the deliverable, the acceptance criterion and the day — and it is the company's own document.

The scoping of outside work and the management of it are Book I8's (Freelancers, Agencies and Outsourcing, chapters 5 to 7 of Book I8), and this chapter cites that treatment rather than repeating it. What belongs here is the reading of the proposal itself, which is a small and repeatable exercise.

The table below sets seven questions beside what to read in a proposal and what to check against the company's own need. It is a reading guide and never a rating of any provider, and no amount is printed. Read the third column as the part that keeps the decision with the company: each answer is checked against the need the company wrote, not against the proposal's own standard.

Table 9.3 — Seven questions to put to an advisor's proposal on sourcing

In the case, the two readings are checked against the need the founders wrote. The outside lawyer's reading of the supply and quality agreement and of the tooling agreement is checked against the agreements the company is about to sign, and the accountant's reading of the cost comparison is checked against the comparison the company built, all posed by this book. Neither reading is asked to produce a verdict, and neither produces one.

Chapter 10 reads the tools by function, one table per function, with the pricing model and no amounts, and this chapter names no tool. Chapter 11 gives the plan of ninety days in which the roles of this chapter are held and the two readings are bought.

CASE IN POINT Hydrosentinelle inc. engages an outside lawyer for 5 hours and an accountant for 2, and hires no one

Hydrosentinelle inc. is a fictional business created for this encyclopedia; its founders, its employees, its customers, its partners, its suppliers, its figures and the events described are invented, and any resemblance to a real business is coincidental. Laurence Bergeron and Karim Haddad keep the boundary and sign the verdicts of the grid on day 28. The outside lawyer reads the supply and quality agreement before it is signed on day 57 and the tooling agreement signed on day 58, 5 hours at Book F9's illustrative rate of $300 an hour. The accountant reads the cost comparison on day 30, 2 hours at Book G7's illustrative rate of $120 an hour, at the first gate of the plan. A quality or certification consultant, a purchasing or supplier-development advisor, a strategy advisor, an advisor of a public program for manufacturers and a broker of businesses for sale are named and not engaged. No one is hired. All of it is posed by this book, illustrative and not a forecast, and nothing is credited to any help.

Costly Mistakes

• Asking too late, after the agreement is signed. A reading bought after signature is a reading of a document the company is already bound by, and the moment when the reading could change the document has passed.
• Asking the wrong party to decide. A lawyer reads an agreement, an accountant reads a comparison and an advisor tests a reasoning, and a company that lets any of them sign the verdict has moved a decision it owns to a person who does not own it.
• Choosing help by its label. A title or a brand stands in for the skill the job needs, and the engagement starts before anyone has asked what the person has read or built that resembles the job.
• Signing an open-ended mandate. A fee with no scope runs on, the deliverable is never accepted, and the engagement has no day on which it ends.
• Writing the need after reading the proposal. A company that reads a proposal first can end by adopting the proposal's framing of the question, and the company's own activity and criterion are never written down.
• Treating a lawyer's reading as a decision that an agreement is lawful. Only the Tribunal or a court decides whether a given agreement or merger is lawful, and a reading is advice to the company and not a ruling.
• Treating a program as a promise of funding. A program is a source of advice to ask about, and a company that plans around a program it has not read has planned around a hope.

Key Takeaways

• Seven roles stand behind a small firm's sourcing decisions — the owner as the keeper of the boundary, the operations and production lead, the purchasing and supplier manager, quality and certification help, legal help, the accountant and the financial advisor, and the strategy advisor with publicly funded help — and an eighth group, the suppliers, performs under agreements and is not an advisor.
• The owner holds four things before any help is engaged: the grid and its weights, the verdicts and their reasons, the triggers that would bring an activity back, and the means to bring it back, which in the case are the in-house build cell, the end-of-line test protocol and the duplicate test bench.
• Outside help buys a skill or a reading, and the decision stays with the company: a lawyer reads an agreement, an accountant reads the cost comparison, an advisor tests a reasoning and a consultant examines a process, and none of them decides which activities the company owns.
• The forms of pay for outside sourcing help are described by form and never by an amount — a fixed fee for a scope, a day rate, an hourly rate, a retainer and a mandate priced by scope — and each form shifts a different risk between the company and the advisor.
• The case engages an outside lawyer for 5 hours at Book F9's illustrative rate and an accountant for 2 hours at Book G7's illustrative rate, names five kinds of help without engaging them, and hires no one; the plan takes 121 hours and commits $89,540 with the duplicate test bench and the molds, all posed by this book.
• A proposal is the advisor's own offer and is read against the company's written need, through seven questions from the scope to the outcome claims, and no outcome claim is printed as a fact.

Your Action Plan

1. Write the roles of your own make-or-buy work on one page. List the seven roles of section 9.1 and write beside each one the person who holds it and the person who backs it up. Deliverable: a one-page register of roles with a holder and a backup for each. Verification: every role has a named holder and a named backup, and no role is held by a supplier.
2. Write the four things the owner holds. Put the grid and its weights, the verdicts and their reasons, the triggers and the means to bring each activity back into one document, activity by activity. Deliverable: a written verdict for each activity with its reason, its trigger and the means to act on the trigger. Verification: every verdict has a reason and a trigger, and every trigger has a means the company actually holds.
3. Write the company's need before you read any proposal. For each reading you intend to buy, write the activity, the decision to be informed, the deliverable, the acceptance criterion and the day. Deliverable: a written need of one page per reading. Verification: each need names a deliverable that a second person could check against the criterion.
4. Read each proposal against the seven questions. Take the proposal you receive and answer the seven questions of Table 9.3 in writing, checking each answer against your own need rather than against the proposal's standard. Deliverable: a written reading of the proposal, question by question. Verification: each of the seven questions has a written answer, and the form of pay and the day the engagement ends are both written down.
5. Decide what you will not buy, and write it down. Name the kinds of help you have considered and not engaged, and the condition under which you would engage each one. Deliverable: a written list of the help named and not engaged, with a condition for each. Verification: each named kind of help has a condition that would change the decision, and the list is read at the next quarterly review.

In La Machine

In the application copy read on October 16, 2026, Control center → AI Agents (/agents-ia) and CRM → Companies (business/marketing/companies) are the two pages nearest to this chapter, and the box of section 9.4 describes them (La Machine, 2026). What they offer a business that is deciding what to own is a task with an observable end and a place to keep a company's name.

What a business needs elsewhere is the rest of the work. La Machine names no purchasing manager, keeps no register of the roles of a sourcing relationship, reads no quote and no contract and decides nothing, and a company record holds no contract, no price, no tooling and no performance. A place for the register of the roles of each sourcing relationship, a place for the supplier's quote and contract, and a way to compare quotes are chosen by function in Chapter 10, and this book is valid with any tool. Check it in your installation. La Machine is described and never sold, and the case does not use it.

Self-Check

1. What are the seven roles behind make-or-buy work, and what is the eighth group that sits outside the firm?
2. What four things does the owner hold before any help is engaged, and why can none of them be delegated?
3. What does the purchasing and supplier manager hold, and what are the five forms of pay for outside sourcing help?
4. What does a lawyer read before a supply or tooling agreement is signed, and who decides whether a given agreement is lawful?
5. What are the seven questions to put to an advisor's proposal on sourcing, and what does each one check?

Answers. 1. The seven roles are the owner as the keeper of the boundary, the operations and production lead, the purchasing and supplier manager, quality and certification help, legal help, the accountant and the financial advisor, and the strategy advisor with publicly funded help; the eighth group is the suppliers, which perform under agreements and are not advisors. 2. The owner holds the grid and its weights, the verdicts and their reasons, the triggers that would bring an activity back, and the means to bring it back; none of them can be delegated because the decision binds the company's capital, its capacity and its customer promise at once, and the person who carries the consequences is the person who should sign. 3. The purchasing and supplier manager reads the quotes, holds the commercial terms and the scorecard and writes the company's agenda before each review; the five forms of pay are a fixed fee for a scope, a day rate, an hourly rate, a retainer and a mandate priced by scope, each described by form and never by an amount. 4. A lawyer reads the supply and quality agreement, the tooling agreement and, if a purchase is ever considered, the acquisition agreement before they are signed, and the decision to sign stays with the company; only the Tribunal or a court decides whether a given agreement or merger is lawful, and contracts are Book H1's while competition law is Book H4's. 5. The seven questions are the scope, the deliverables and the acceptance criterion, the form of pay, who owns the work and the data, who else pays the advisor, the handover and the way out, and the outcome claims; each one is checked against the company's own written need rather than against the proposal's own standard, and no outcome claim is printed as a fact.

Chapter References

Civil Code of Québec, CQLR c CCQ-1991, arts. 1708, 1726, 1729, 2098, 2099, 2100, 2103, 2125, 2126 and 2129 (Québec). LégisQuébec, up to date to August 12, 2026 — read October 15, 2026. https://www.legisquebec.gouv.qc.ca/en/document/cs/CCQ-1991

Competition Act, R.S.C. 1985, c. C-34, ss. 45, 77, 90.1, 91, 92, 109 and 110 (Canada). Justice Laws Website, current to September 30, 2026 (last amended March 26, 2026) — read October 16, 2026. https://laws-lois.justice.gc.ca/eng/acts/C-34/

Competition Bureau. (2021). Competitor collaboration guidelines. Government of Canada. (Enforcement guidelines dated May 6, 2021; page modified June 27, 2024.) Retrieved October 15, 2026, from https://competition-bureau.canada.ca/en/how-we-foster-competition/education-and-outreach/competitor-collaboration-guidelines

Competition Bureau. (2026, March 2). Pre-merger Competition Bureau notification threshold to remain at $93M in 2026 [News release]. Government of Canada. Retrieved October 15, 2026, from https://www.canada.ca/en/competition-bureau/news/2026/03/pre-merger-competition-bureau-notification-threshold-to-remain-at-93m-in-2026.html

La Machine. (2026). Documentation — route reference inventory. Internal product documentation, read October 16, 2026, together with the code of the pages this book describes: /agents-ia and business/marketing/companies.

Chapter 10 — Tools: Functions and Popular Tools

A tool is a function before it is a product, and the register of make-or-buy decisions is the record a business keeps whatever tools it buys, so this chapter sets out what the tools around sourcing and integration do by function, gives each function one table of examples read on the day the chapter is written, places La Machine among the functions, and says how to choose a tool and read the pricing and contract terms behind it, with the rule of free and already-paid first, applied to Hydrosentinelle inc., which chooses no tool for its plan and keeps its decisions in a written register posed by this book.

What You'll Be Able to Do

• Name the six functions that sourcing and integration tools serve, from value chain and process mapping to inventories of tools, assets and dependencies, and say what each function does and what each misses.
• Describe what a mapping tool, a cost model, a supplier repository, a purchasing module, a quality tool and an inventory each see and each miss, and why no tool holds the whole make-or-buy decision.
• Read a vendor's page for the function, the pricing model and the terms, without an amount, and know which questions belong to the contract rather than to the page.
• Apply the rule of free and already-paid first, and add a tool only when a written need remains that the spreadsheet or the module of a core system does not meet.
• Place La Machine among the six functions, saying what its tools inventory, its company records, its plan tasks and its calculator do and what the platform does not hold.
• Keep the register of make-or-buy decisions as the one record that survives every tool change, with each activity's verdict, its reason, its owner, its triggers and its next review.

Why It Matters

A business that decides what to own, to have made or to contract does not need a tool to make the decision, and it does need a place to keep the decision. Each decision of the earlier chapters produces a line: an activity, a verdict, a reason, an owner, a trigger and a next review. A line that lives only in a conversation is a line that will be taken again next year on the same facts. The register is the record that holds the lines together, and it is the one thing in this chapter that no vendor sells.

Around that register sit six functions, and each of them sees part of the work and misses another: a mapping tool does not decide which stage to own, a cost model does not choose the drivers or the rate, a repository does not read the terms it keeps, a purchasing module acts after the decision, a quality tool does not say what a record means, and an inventory does not measure dependence. The functions come first and the products second, and a business that starts from a product starts in the wrong place.

The rule of this chapter is the rule of free and already-paid first: a spreadsheet the business already runs, a module of a core system it already pays for, and a drawing it already keeps are read before any new tool is considered. No source read measures what a sourcing or integration tool returns, how many businesses use one, what it costs or whether it improves a decision, and this chapter prints no such figure. The examples in its tables are read on the day the chapter is written, they are the vendors' own pages, and they are examples of a function and never a recommendation. Hydrosentinelle inc. chooses no tool for its plan, and the plan of ninety days commits $89,540 without buying one, all posed by this book.

In This Chapter

Section 10.1 sets out what sourcing and integration tools do, by function, and places six functions around one register of make-or-buy decisions. Section 10.2 covers value chain and process mapping tools. Section 10.3 covers cost models and ownership calculators. Section 10.4 covers supplier records, scorecards and contract repositories. Section 10.5 covers purchasing, inventory and production planning modules. Section 10.6 covers quality, traceability and document control tools. Section 10.7 covers inventories of tools, assets and dependencies. Section 10.8 places La Machine among the six functions. Section 10.9 sets out how to choose a tool and what to read on a vendor's page and in a contract.

10.1 What Sourcing and Integration Tools Do, by Function

The tools that touch a sourcing decision serve six functions, and the functions are stable while the products change. Value chain and process mapping draws the stages of a chain and who does each. Cost models and ownership calculators put a made unit and a bought unit on one basis. Supplier records, scorecards and contract repositories keep each supplier's terms and readings. Purchasing, inventory and production planning modules release, receive and plan quantities. Quality, traceability and document control tools keep test records, lots and versions. Inventories of tools, assets and dependencies list what the business relies on and who holds it. A business should be able to name its need in one of those six sentences before it reads a vendor's page.

The written decision of section 3.9 is the register's content: one line per activity, with the verdict, the reason, the owner, the triggers and the next review. A spreadsheet, a document or a module of a core system can hold the register, provided it can hold a reason and a trigger and not only a quantity. What makes it a register is that a second person can read a line and see why the business decided what it decided.

The rule of free and already-paid first belongs here, before any table of examples: the spreadsheet the business already runs, the module of the core system it already pays for, and the drawing it already keeps are read first, and a new tool is added only when a written need remains that none of them meets. The rule is this book's own, drawn from its reasoning and not from a source, and it is neither a standard nor a legal duty.

Figure 10.1 sets six functions of sourcing tools around one register of make-or-buy decisions, and places La Machine among the six.

Figure 10.1 — Six functions around one register. The register of make-or-buy decisions is the record a business keeps whatever tools it buys; La Machine is described as of the application copy of September 7, 2026, without a purchase order, a supplier scorecard or a cost model.

The table below sets six functions and the register beside what each does, what each misses and how the case covers it by function only. The case's column is posed by this book, and no product is named in it. Read the table as a map of the chapter: each of the six functions has its own section and its own table of examples, and the register is the record that none of them replaces.

Table 10.1 — Six functions around the register of make-or-buy decisions, and how the case covers each

DEFINITION Function, example row, pricing model and register of make-or-buy decisions, as this book uses the words

These are this book's working words, and no source read defines them in this sense. A function is what a tool does for a business, whatever the tool is called. An example row is a tool named in a table of this chapter, read on a vendor's page on the day the chapter is written, as an example of its function and never as a recommendation. A pricing model is the way a vendor charges as its page states it — a free tier, a trial, a subscription by seat or by tier, a usage-based plan, a plan on request — without an amount. The register of make-or-buy decisions is the one record that a business keeps for each activity, with its verdict, its reason, its owner, its triggers and its next review, whatever tools it buys.

KEY CONCEPT No tool holds the whole decision, and each function sees only part of the work

This is the book's reasoning and never a finding. A mapping tool draws the stages and misses the decision about which to own. A cost model puts made and bought on one basis and misses the choice of the drivers and of the rate. A repository keeps the terms and the readings and misses the reading of them. A purchasing module releases and receives and misses the decision that comes before. A quality tool keeps the records and misses what a record means. An inventory lists what the business relies on and misses how much it depends on each. The register of make-or-buy decisions holds them together, and it holds nothing that a person has not written into it. The case buys no tool.

10.2 Value Chain and Process Mapping Tools

A mapping tool draws the stages of a chain and who does each, and it shows the handoffs between stages. It answers a question that comes before the make-or-buy decision: what are the stages of this business, and who does each one today? A chain drawn on a page makes the stages visible, and a stage that is visible is a stage that can be scored on the grid of Chapter 3. The tool does not score the stage or decide which stage to own: it draws, and the owner decides.

The value chain is Book A3's (Strategy and Competitive Positioning, chapter 2 of Book A3), and this chapter cites it rather than redrawing it. The mapping of processes and its tools are Book J1's (Operations Management Fundamentals, chapters 2 and 10 of Book J1), and this chapter cites them rather than repeating them. A mapping tool is the first function of the six, because a business cannot score an activity it has not named.

A mapping tool has three limits. It draws what a person tells it to draw, so a chain drawn from memory has the gaps of the memory; a stage on a page is not a cost; and a map ages, which is why Chapter 8 places a yearly redrawing in the routine and why a second person should read the map against the work.

In the case, the chain of the company's activities is mapped on day 9, and six activities are listed: the firmware and the alert logic, the enclosure tooling for the second generation, the electronic boards, the final kit assembly and the end-of-line test, the installation in buildings, and the cloud hosting of the monitoring service. No tool is chosen for the mapping, posed by this book. The six activities are the ones the grid scores on day 24, and the firmware and the alert logic stay in-house.

The table below presents examples of the function of value chain and process mapping, with the situation of use each is best for, its pricing cell and a note. This chapter is the only place of the book where a product is named, and the rows are set as of October 16, 2026, the day the vendor pages were read. The vendor pages were read on that day, for the function and the pricing model only, without amounts, with no account opened and no condition accepted, and the pricing cell is the form of charging and never an amount.

Table 10.2 — Examples of value chain and process mapping tools

WATCH OUT A cost model that decides for you and a scorecard that predicts are two traps

Two traps catch a business that buys a tool for a sourcing decision, and each has a question that catches it. The first is letting a cost model decide: the drivers and the rate are the tool's defaults, nobody has written the business's own, and the result is read as the answer. The question is which drivers this comparison carries, who chose the rate, and where they are written. The second is reading a scorecard tool as a predictor: a dashboard of counts is taken for a forecast of a supplier's failure, and the counts are read as a warning rather than as a call for action. The question is what action each reading calls for, and who owns it. The book's reasoning is illustrative, and no frequency of either trap in any business is printed. In the case, the founders write their own cost comparison and read their own scorecard, and the case chooses no tool, posed by this book.

10.3 Cost Models and Ownership Calculators

A cost model puts a made unit and a bought unit on one basis, adds the cost lines of total cost of ownership, and reads a break-even volume and a net present value from an outlay, a saving and a rate. It is the function that turns the numbers of Chapter 7 into a page a person can read. It does not choose the drivers, the rate or the life, and it does not say what a figure means. A model is a way of writing assumptions down, and a model whose assumptions are not written is a model that will be read as a fact.

Total cost of ownership looks beyond the price of a purchase to many other purchase-related costs, and the abstract of the study read for this book reports case studies of 11 firms, with dollar-based and value-based models used for supplier selection or supplier evaluation (Ellram, 1995, abstract only, read on October 15, 2026). The abstract reports no typical share of hidden costs, and this chapter prints none. The second reading is at the same level: the abstract reports that companies use total cost of ownership models, that a generic model is not appropriate, and that a model built on a core set of cost drivers, with auxiliary drivers chosen for the purchase at hand, is suggested (Ferrin & Plank, 2002, abstract only, read on October 15, 2026). No list of drivers beyond the abstract is printed here. This book draws one sentence from them and offers it as its own reasoning and not as a finding: the model is the business's own, and a model copied from another business carries that business's drivers.

The tools for budgets, costs and profitability are Book G3's (Budgeting, Cost Control and Profitability, chapter 11 of Book G3), and the tools of valuation are Book M2's (Business Valuation, chapter 12 of Book M2), and this chapter cites each rather than repeating it. The numbers of Chapter 7 are the book's method, and this chapter cites them rather than recomputing them. A cost model is the function that puts made and bought on one basis.

In the case, the founders write the cost comparison and the accountant reads it at the gate of day 30, and no tool is chosen, posed by this book. The comparison puts the kit made at $175.50 and the kit bought at $202 on one basis, and it is the founders' own model rather than a template. The accountant's reading is the second reader that a model needs, a service at the printed illustrative rate and not a tool. The molds are compared on the same basis, at $72,000 of investment against a difference of $70 a kit, and the break-even count of 1,028.6 kits is read from the founders' own page.

The table below presents examples of the function of cost models and ownership calculators, with the situation of use each is best for, its pricing cell and a note. The vendor pages were read on October 16, 2026, for the function and the pricing model only, without amounts, and the pricing cell is the form of charging and never an amount.

Table 10.3 — Examples of cost models and ownership calculators

TOOLS OF THE TRADE What any cost model and ownership calculator must offer, whatever it is called

Nine functions carry a cost model, and a business should look for them rather than for a product. The first is a place to write the unit, the period and the volume, because a cost per unit without a volume has no meaning. The second is the core cost lines and a place for auxiliary lines, the shape the abstract read for this book suggests (Ferrin & Plank, 2002, abstract only). The third is hours valued at a rate the business can change. The fourth is the cost per unit made and bought side by side. The fifth is an investment, a saving per unit and a break-even count of units. The sixth is a rate and a life, with the net present value shown beside the break-even. The seventh is a way to change the volume, the price and the rate and see the result move. The eighth is a record of who changed an assumption and when. The ninth is an export in a form another tool can read. No brand name belongs on this list and no price; the requirement is the function.

10.4 Supplier Records, Scorecards and Contract Repositories

A supplier record, a scorecard and a contract repository keep each supplier's terms, contacts, renewal dates and readings, and store the agreements. They answer a question that comes after the make-or-buy decision: what did the business agree to, what has the supplier done since, and when does the agreement come up again? A repository that holds the terms and the readings is a repository that a person can read, and it does not read the agreement, decide whether a supplier performs, or tell the company what to do with a reading.

The tools for buying outside work are Book I8's (Freelancers, Agencies and Outsourcing, chapter 11 of Book I8), and this chapter cites them rather than repeating them. The contract behind a tool is Book H1's (Contracts and Commercial Agreements), and this chapter calls it rather than interpreting any clause. What this chapter adds is the place of a repository in a sourcing decision: it is the function that keeps the terms and the readings, and it is the function whose value depends entirely on a person reading what it holds.

A repository has three limits. It stores what it is given, so an agreement never uploaded is unknown to it; a renewal date is a date and not a decision; and a scorecard holds counts and dates, so a count of late deliveries is not a forecast of a supplier's failure, which the monthly reading of Chapter 8 handles by giving each reading past its threshold an owner and a day.

In the case, the supply and quality agreement is signed on day 57 after the outside lawyer's reading, the tooling agreement on day 58, and the monthly scorecard is live on day 88, and no tool is chosen, posed by this book. The scorecard carries the counts and the dates that the supply and quality agreement requires, and the triggers written at the gate of day 90 are read against it. The end-of-line test records of each supplier-built kit come back to the company every day under the supply agreement, and the company owns them, posed by this book.

The table below presents examples of the function of supplier records, scorecards and contract repositories, with the situation of use each is best for, its pricing cell and a note. The vendor pages were read on October 16, 2026, for the function and the pricing model only, without amounts, and the pricing cell is the form of charging and never an amount.

Table 10.4 — Examples of supplier records, scorecards and contract repositories

QUÉBEC & CANADA NOTE A repository holds personal information, and the language of its interface and of support is a question for Book H4

The Canadian point of this section is that a repository of contracts, a scorecard tool or a log can hold personal information about the people named in an agreement or a minute, and where a tool stores that information and who may read it are questions for Book H2 (Privacy and Personal Information) and Book K3 (Cybersecurity, Backups and Business Continuity). The language of the interface and of the support that a Québec business offers its staff is Book H4's (Consumer Protection, Advertising Law, Competition and Language Requirements). The contract behind a tool is Book H1's (Contracts and Commercial Agreements). This book explains published rules and recommends no vendor, and it sends the reader to Book H1, Book H2, Book H4, Book K3 and an advisor for a given agreement or a given data flow. Whether the reader's own tool or contract complies is a question for an advisor, never decided by this book. In the case, no tool is chosen, posed by this book.

10.5 Purchasing, Inventory and Production Planning Modules

A purchasing, inventory and production planning module releases an order, receives goods, tracks stock and plans quantities against a bill of materials. It is the function that runs the flow after the make-or-buy decision has been made: a module that releases a purchase order for a component has already been told that the component is bought, and the telling happened in the register rather than in the module. The decision rests on a verdict about what the business should own, and a module cannot make it.

Core systems and their modules are Book K2's (CRM and ERP Systems, chapter 2 of Book K2), and this chapter cites them rather than repeating them. The purchasing of goods and the control of stock lie outside this book, in Book J5. A module is the function that acts after the decision.

A module has three limits. A bill of materials that is out of date produces a plan that is out of date; a stock level is a count and not a judgment about whether the stock is in the right place; and a receipt records that goods arrived and not that they are acceptable, which is the quality function of the next section.

In the case, the kit forecast and the weekly build release are written by Karim Haddad, and no module is chosen, posed by this book. The forecast of 230 kits is the input to the release, and the release goes to the contract manufacturer each week. The company buys 207 kits and makes 23 in-house, and the module that would release and receive those quantities is not bought. The weekly build release is a line of the routine of Chapter 8, written by hand.

The table below presents examples of the function of purchasing, inventory and production planning modules, with the situation of use each is best for, its pricing cell and a note. The vendor pages were read on October 16, 2026, for the function and the pricing model only, without amounts, and the pricing cell is the form of charging and never an amount.

Table 10.5 — Examples of purchasing, inventory and production planning modules

10.6 Quality, Traceability and Document Control Tools

A quality, traceability and document control tool keeps test records, lots and serial numbers, the versions of a protocol, a firmware image or a drawing, and the approvals of a change. It is the function that answers a question a customer or a regulator may ask: which unit is this, what was tested on it, and which version of the protocol was in force when it was tested? The tool keeps the records, and it does not decide whether a record is good enough or whether a supplier's quality is acceptable.

The tools of quality management are Book J4's (Quality Management and Continuous Improvement, chapter 11 of Book J4), and this chapter cites them rather than repeating them. A quality tool is the function that keeps the records of both sides when work moves to a supplier, and it makes the difference between a record the company holds and a record the supplier holds.

A quality tool has three limits. A record that a unit passed says nothing about whether the test was the right test; traceability runs only as far as the data that entered it; and an approval workflow records that a person approved a change and not whether the change was a good idea.

In the case, the end-of-line test records of each supplier-built kit come back to the company every day under the supply agreement and the company owns them (section 6.10, posed by this book), and no tool is chosen. The contract manufacturer flashes the firmware image that the company supplies and runs the company's end-of-line test on the duplicate bench, and the records of that test are the company's.

The table below presents examples of the function of quality, traceability and document control, with the situation of use each is best for, its pricing cell and a note. The vendor pages were read on October 16, 2026, for the function and the pricing model only, without amounts, and the pricing cell is the form of charging and never an amount.

Table 10.6 — Examples of quality, traceability and document control tools

WHO DOES THIS Laurence Bergeron reads what a tool does, Karim Haddad reads the terms behind it, and the case buys no tool

The roles below are the case's, posed by this book. Laurence Bergeron, the president, holds the technical assessment of each candidate, the end-of-line test protocol and the records. Karim Haddad, the head of sales, holds the reading of the terms, the quotes and the scorecard. The accountant reads the cost comparison, and the outside lawyer reads the agreements. The case chooses no tool. In general, the person who will use a tool reads what it does, the person who signs reads the contract, and a second person tries the tool first; that is this book's suggestion and not part of the case. In a solo business the owner reads both and asks an advisor to read the contract. No hour and no amount is promised to a reader.

10.7 Inventories of Tools, Assets and Dependencies

An inventory of tools, assets and dependencies lists what the business relies on, who holds it, what state it is in and when it is reviewed. It answers a question that a sourcing decision creates: after this activity moved to a supplier, what does the business now depend on, and who holds each thing it depends on? The inventory lists, and it does not measure how much the business depends on each item. A list with an owner for each line is a list a person can act on, and a list without an owner is a list that will be read once.

The mapping of a technology stack is Book K1's (The SME Technology Stack, chapter 2 of Book K1), and this chapter cites it rather than repeating it. The dependence of a business on platforms and its register are Book A9's (Platform Strategy: Integrating with Dominant Offers and Partner Ecosystems), and this chapter cites it rather than repeating it. An inventory is the function that keeps the list of what the business relies on after the boundary has moved.

An inventory has three limits. It lists what a person remembered to list; a line is not a measure of dependence, so a list of ten items is not ten equal dependencies; and it ages, so the review date on each line is what keeps it alive.

In the case, no tool is chosen, and the register of make-or-buy decisions is the record that holds each activity with its reason, its owner, its triggers and its next review, posed by this book. The case's nearest record is the grid with the verdicts signed by both founders on day 28 and the triggers written at the gate of day 90. The second source of the sensor component is qualified on day 120 and holds 30% of the component by day 365, and the dependence on a single source is what the inventory would list if the case kept one.

The table below presents examples of the function of inventories of tools, assets and dependencies, with the situation of use each is best for, its pricing cell and a note. The vendor pages were read on October 16, 2026, for the function and the pricing model only, without amounts, and the pricing cell is the form of charging and never an amount.

Table 10.7 — Examples of inventories of tools, assets and dependencies

WHAT CHANGED The tool tables are read on the day the chapter is written, and a vendor's page is read again on the reader's own date

The dated reading is this one: the rows of Tables 10.2 to 10.7 are set as of the day on which the vendor pages were read for this chapter, October 16, 2026, and a tool's features, pricing model and terms change on the vendor's own date, so a reader reads each page again and never assumes a feature. The application code of La Machine was read at its copy of September 7, 2026, on October 16, 2026 (La Machine, 2026). This chapter prints no earlier version of any product and no change between versions, because no earlier version was read. None of these dates is a date of the case, which is counted in days and never in calendar years or months. No tool is named in this box, and no figure of any vendor is printed anywhere in this chapter.

10.8 La Machine: The Tools Inventory, Companies and Plan Tasks in Their Place

In the application copy of September 7, 2026, read on October 15, 2026, the pages that touch this chapter are these. Control center → Tools (inventory) (/outils) lists the installation's software tools with a name, a role, a category, a state, a cost, a count of tasks and an identity measured by a script, a version, a location and a date, and it can serve as a model for an inventory of what a business depends on, for software only (La Machine, 2026). Control center → Workflows (/registre) shows one line for each workflow or agent route, with its code, type, trigger, entry point and state, and only the state can be changed on the screen, so it is an example of a register and not the reader's register of make-or-buy decisions. Control center → Plan tasks (/taches) holds recurring or one-off tasks with a "Done when" criterion required at creation, and the Mark as done button does not check it. Control center → AI Agents (/agents-ia) can create a plan task after a click and reads no quote and no supplier contract. CRM → Companies (business/marketing/companies) holds one record for each company and can record a supplier or a molder as a company, with no field for a contract, a price, a tooling or a performance. Productivity → Calculator & Tools (productivity/tools) offers a general calculator, timers and a timesheet by project, with no template of unit cost, break-even or present value. The plan tasks are Book N2's (Piloting Tasks, Executions and Projects), the company and contact records Book N4's (CRM, Contacts and Prospect Data), and the assistant Book N11's (The AI Assistant, Agents and Orchestration), and this chapter cites each rather than repeating it.

What the platform does not hold is the rest of this chapter's subject, and the list is worth reading in full sentences. There is no supplier record, no supplier evaluation and no supplier scorecard. There is no request for quotation, no comparison of quotes, no purchase order and no receiving. There is no bill of materials, no routing, no production order and no production planning. There is no stock, no reorder point and no inventory of parts. There is no register of tooling, of molds or of assets held at a supplier. There is no traceability of lots or serial numbers. There is no total cost, break-even, present-value or make-or-buy model. There is no register of sourcing decisions and their triggers, and there are no supplier contracts. A business should check it in its own installation, and the case does not use La Machine.

The table below sets the functions of this chapter beside what La Machine offers and where to do it instead. It is one worked example of a business platform, described and never sold, and no product is named in it. Read the middle column as what the platform shows and the right column as the function a business needs elsewhere.

Table 10.8 — La Machine against the functions of this chapter

IN LA MACHINE An inventory, plan tasks, an assistant, company records and a calculator exist, and no purchasing, supplier or cost function does

In the application copy of September 7, 2026, Control center → Tools (inventory) (/outils) measures a version and a date and no dependence on a supplier of goods (La Machine, 2026). Control center → Workflows (/registre) is a register of routes with no owner column, an example of a register and not the reader's register of make-or-buy decisions. Control center → Plan tasks (/taches) holds a recurring review with a required "Done when" that the Mark as done button does not verify. Control center → AI Agents (/agents-ia) creates a plan task after a click and reads no quote and no supplier contract. CRM → Companies (business/marketing/companies) can record a supplier as a company and has no field for a contract, a price or a performance. Productivity → Calculator & Tools (productivity/tools) has no template of unit cost, break-even or present value. La Machine holds no purchase order, bill of materials, supplier scorecard or inventory of parts; check it in your installation. La Machine is described and never sold, and the case does not use it.

10.9 Choosing Tools, and the Pricing and Contract Terms Behind Them

Choosing a tool starts from the function and from what the business already has and has already paid for. The rule of free and already-paid first is the rule of this chapter, and it is applied before any vendor's page is opened. A business writes the function in one sentence, writes the need that remains after the spreadsheet and the module of the core system have been read, and reads a vendor's page only when a written need remains.

The choice of a technology stack and the buying of software are Book K1's (The SME Technology Stack, chapters 3 and 4 of Book K1), and this chapter cites them rather than repeating them. Backups and the security of access are Book K3's (Cybersecurity, Backups and Business Continuity), and personal information is Book H2's (Privacy and Personal Information), and this chapter calls each rather than deciding any question of compliance. The contract behind a tool is Book H1's (Contracts and Commercial Agreements), and the language of an interface and of support for a Québec business is Book H4's (Consumer Protection, Advertising Law, Competition and Language Requirements), and this chapter calls each rather than interpreting any clause.

The terms to read on a pricing page and in a contract are a short list, and the amounts are left to the vendor's page and are never printed here. The pricing model is the form of charging: a free tier, a trial, a subscription by seat or by tier, a usage-based plan, or a plan on request. What a free tier or a trial includes, and the length of a trial, are questions for the page. The term and the renewal belong to the contract, and so do the export of the business's data, where the data are stored, the language of the interface and of support, the handling of personal information, and security and access. A business that reads the pricing page and not the contract has read half of what it is buying.

In the case, no tool is chosen and none is bought, posed by this book. The plan of ninety days commits $89,540, which is 121 hours at the printed illustrative rates, the duplicate test bench at $9,200 and the molds at $72,000, and none of the three buys a tool. The case keeps its decisions in a written register, and the register is the record that survives every tool change.

The table below sets eight questions beside what to read on a vendor's page and what to check in the contract. It is a reading guide and not a rating of any provider, and no amount is printed in it. Read the two right-hand columns as two different documents: the page states the vendor's own offer, and the contract states what the business has agreed to.

Table 10.9 — What to read on a vendor's page and in a contract

THE RULE Start with what is free and already paid for, and add a tool only when a written need remains

The rule of this chapter is this book's own, drawn from its reasoning and not from a source: the spreadsheet, the module of a core system and the tools the business already pays for are read first, and a new tool is added only when a written need remains that none of them meets. The choice of a technology stack is Book K1's (The SME Technology Stack), and the contract behind a tool is Book H1's (Contracts and Commercial Agreements). The rule is neither a standard nor a legal duty, and it never says that a tool is needed. It says only that the order of reading is the free and the already-paid first. In the case, no tool is chosen and none is bought, posed by this book, and no figure beyond the case's is printed here.

METHOD Choosing a sourcing tool in five steps

First, write the function in one sentence and the need that remains after what the business already has and has already paid for. Second, read the vendor's pricing page and the terms — the pricing model, the free tier or trial, the term and renewal, the export of the data, where the data are stored, the language of the interface and of support, personal information, security and access — and write the date of the reading. Third, try the tool on one activity with no customer data. Fourth, decide with a written criterion and a person who owns the decision, and note the way out. Fifth, add the tool to the register of make-or-buy decisions, and read it again at the yearly review of Chapter 8. This is the book's reasoning and not a tested procedure. In the case, steps one and four are followed and no tool is bought, posed by this book.

DO THE MATH Six functions, six tables of examples and no tool chosen for the case

Illustrative figures; the case is fictional. Step 1 — Sourcing and integration tools serve six functions, from value chain and process mapping to inventories of tools, assets and dependencies. Step 2 — Each function has one table of examples, Tables 10.2 to 10.7, read on the day the chapter is written. Step 3 — The case chooses 0 tools, posed by this book. Step 4 — The plan commits $89,540, which is 121 hours at the printed illustrative rates, $8,340, the duplicate test bench at $9,200 and the molds at $72,000. Step 5 — None of the three buys a tool, since the hours pay for people's time, the bench is equipment for the end-of-line test and the molds are tooling. The counts and the amounts are posed by this book, are the case's and not a norm, the amounts are the plan's and not tool prices, and the sums are exact.

CASE IN POINT Hydrosentinelle inc. chooses no tool and keeps its decisions in a written register

Hydrosentinelle inc. is a fictional business created for this encyclopedia; its founders, its employees, its customers, its partners, its suppliers, its figures and the events described are invented, and any resemblance to a real business is coincidental. The six functions of this chapter are covered by function only. The chain of the company's activities is mapped on day 9, and six activities are listed. The founders write the cost comparison and the accountant reads it at the gate of day 30. The supply and quality agreement is signed on day 57 after the outside lawyer's reading, and the tooling agreement on day 58. The monthly scorecard is live on day 88. The end-of-line test records of each supplier-built kit come back to the company every day under the supply agreement, and the company owns them. The grid with the verdicts signed by both founders on day 28 and the triggers written at the gate of day 90 are the record that the case keeps. No tool is chosen, priced or bought, and none is named. All of it is posed by this book, illustrative and not a forecast.

Chapter 11 assembles the plan of ninety days, its gates and its triggers, and reads what the case did with the boundary it drew.

Costly Mistakes

• Starting from a product instead of a function. A business that opens a vendor's page before it can name its need in one of the six functions will buy a tool for a function it cannot describe.
• Letting a cost model decide. The drivers and the rate are the tool's defaults, nobody has written the business's own, and the result is read as the answer.
• Reading a scorecard as a forecast. A count of late deliveries is a record of what happened and not a prediction of a supplier's failure.
• Buying a purchasing module to make a make-or-buy decision. A module releases, receives and plans quantities after the decision, and it cannot make a decision that rests on a verdict about what the business should own.
• Skipping the free and already-paid tools. A business that skips the spreadsheet, the module and the drawing it already has pays twice for the same function.
• Reading the pricing page and not the contract. The pricing model is one line of what is being bought, and the term, the renewal, the export of the data, the place of storage, the language and the handling of personal information belong to the contract.
• Letting a tool stand for the register. A repository, a module or an inventory holds part of the work, and the register of make-or-buy decisions is the one record that holds each activity with its verdict, its reason, its owner, its triggers and its next review.

Key Takeaways

• Sourcing and integration tools serve six functions, from value chain and process mapping to inventories of tools, assets and dependencies, and the functions come first while the products come second.
• No tool holds the whole make-or-buy decision, because each function sees part of the work and misses another, and the register of make-or-buy decisions is the one record that holds them together.
• The rule of free and already-paid first is applied before any vendor's page is opened: the spreadsheet, the module of a core system and the tools the business already pays for are read first, and a new tool is added only when a written need remains that none of them meets.
• The examples in the tables of this chapter are read on the day the chapter is written, carry the pricing model without an amount, and are examples of a function and never a recommendation; no source read measures what a sourcing or integration tool returns, how many businesses use one, what it costs or whether it improves a decision.
• La Machine, as of the application copy of September 7, 2026, shows an inventory of software tools, a register of workflows, plan tasks, an assistant that creates a plan task after a click, company records and a general calculator, and it holds no purchase order, no bill of materials, no supplier record or scorecard, no inventory of parts, no register of tooling and no make-or-buy model.
• Hydrosentinelle inc. chooses no tool for its plan, and the plan of ninety days commits $89,540 without buying one, all posed by this book; the case keeps its decisions in a written register, and no tool is named for the case.

Your Action Plan

1. Write the function before you read a page. Name the need in one of the six functions of this chapter, in one sentence, and write what the business already has and has already paid for against it. Deliverable: a written need of one sentence and a list of the free and already-paid tools that were read first. Verification: the need names a function, and the list shows at least one tool the business already runs.
2. Build the register of make-or-buy decisions. Write one line per activity, with its verdict, its reason, its owner, its triggers and its next review, in a place a second person can read. Deliverable: a written register with one line per activity of the chain. Verification: every line carries a verdict, a reason, an owner, a trigger and a next review, and a second person has read it.
3. Read the pricing page and the contract for any tool you consider. Read the pricing model, the free tier or trial, the term and renewal, the export of the data, the place of storage, the language of the interface and of support, the handling of personal information, and security and access, and write the date of the reading. Deliverable: a written reading of the eight questions of Table 10.9 for each tool considered. Verification: every question has an answer from the page or from the contract, and the date of the reading is written beside it.
4. Try the tool on one activity with no customer data. Run the tool on a single activity of the chain, with no customer data in it, and write what it produced and what it did not. Deliverable: a written trial note of one page. Verification: the trial names the activity, the output and the limit the tool showed.
5. Add the tool to the register and read it again at the yearly review. Record the tool beside the activity it serves, with the person who owns it and the date of the next review, and read it again at the yearly make-or-buy review of Chapter 8. Deliverable: a line in the register for each tool in use. Verification: every tool in use has an owner and a next review date, and the register has been read once at the yearly review.

In La Machine

In the application copy of September 7, 2026, read on October 15, 2026, six pages touch this chapter: Control center → Tools (inventory) (/outils), Control center → Workflows (/registre), Control center → Plan tasks (/taches), Control center → AI Agents (/agents-ia), CRM → Companies (business/marketing/companies) and Productivity → Calculator & Tools (productivity/tools), described in section 10.8 (La Machine, 2026). A business should check them in its own installation.

What a business needs elsewhere is the rest of this chapter's subject. La Machine holds no purchase order, no bill of materials, no supplier record or scorecard, no inventory of parts, no production order, no register of tooling and no cost, break-even, present-value or make-or-buy model, and it holds no register of sourcing decisions and their triggers. Each of the six functions is chosen in the tables of this chapter, and this book is valid with any tool. La Machine is one worked example of a business platform, described and never sold, and the case does not use it.

Self-Check

1. What are the six functions that sourcing and integration tools serve, and what does each of them miss?
2. Why does no tool hold the whole make-or-buy decision, and what is the register of make-or-buy decisions?
3. What does the rule of free and already-paid first say, and when is a new tool added?
4. What does a cost model do, and what does the abstract read for this book report about a generic model?
5. What should a business read on a vendor's page and what should it check in the contract?

Answers. 1. The six functions are value chain and process mapping, cost models and ownership calculators, supplier records, scorecards and contract repositories, purchasing, inventory and production planning modules, quality, traceability and document control, and inventories of tools, assets and dependencies; each misses a part of the decision: the choice of the stage to own, of the drivers and the rate, of the reading of the terms, of the make-or-buy decision that comes before, of what a record means, and of how much the business depends on each item. 2. The decision is a judgment about what the business should own, and a judgment is written by a person; the register is the one record kept for each activity, with its verdict, its reason, its owner, its triggers and its next review, whatever tools the business buys. 3. The spreadsheet, the module of a core system and the tools the business already pays for are read first, and a new tool is added only when a written need remains that none of them meets; the rule is this book's own, neither a standard nor a legal duty. 4. A cost model puts a made unit and a bought unit on one basis and reads a break-even volume and a net present value from an outlay, a saving and a rate; the abstract read for this book reports that companies use total cost of ownership models, that a generic model is not appropriate, and that a model built on a core set of cost drivers, with auxiliary drivers chosen for the purchase at hand, is suggested (Ferrin & Plank, 2002, abstract only). 5. On the page a business reads the function, the pricing model without an amount, what a free tier or a trial includes, the export of the data, where the data are stored, the languages, the personal information the tool holds and the sign-in and access it offers; in the contract it checks the term, the renewal and the notice, the right to export, the place of storage, the language promised, the measures that protect confidentiality and the obligations on security and the notification of an incident.

Chapter References

Ellram, L. M. (1995). Total cost of ownership: An analysis approach for purchasing. International Journal of Physical Distribution & Logistics Management, 25(8), 4–23. https://doi.org/10.1108/09600039510099928

Ferrin, B. G., & Plank, R. E. (2002). Total cost of ownership models: An exploratory study. Journal of Supply Chain Management, 38(2), 18–29. https://doi.org/10.1111/j.1745-493X.2002.tb00132.x

La Machine. (2026). Documentation — route reference inventory. Internal product documentation, read October 16, 2026, together with the code of the pages this book describes: /outils, /registre, /taches, /agents-ia, business/marketing/companies and productivity/tools.

Chapter 11 — Practical Advice, Pitfalls and Your Action Plan

Deciding what to own takes a map, a forecast, a grid, quotes, pilot builds, written agreements, a scorecard, triggers and a plan with gates that records can show, so this chapter says what matters most, names the ten most costly mistakes, lays out the plan in three slices — days 1 to 30 for the map, the forecast and the grid, days 31 to 60 for the pilot builds, the agreements and the molds, and days 61 to 90 for the first supplier-built kits, the scorecard and the triggers — each ending at a gate on days 30, 60 and 90 with a missed gate read again 7 days later, then reads the few readings that show the boundary is right, how to keep the practice alive and the case on day 365, entirely posed by this book, saying plainly that the plan is the book's own method, that its 121 hours and the $89,540 it commits are posed by this book and that no result is credited to it.

What You'll Be Able to Do

• Rank the six priorities that matter most in a boundary decision, from the map of the chain to the triggers that bring a decision back for review.
• Name the ten most costly mistakes of a make-or-buy decision and say which chapter of this book treats each one.
• Run a ninety-day plan in three slices, each ending at a gate that is passed by the records its condition names.
• Write the tasks of each slice with a person, a day and a record, and read a missed gate again 7 days later while nothing new starts.
• Read the six readings that show whether the boundary is right, as counts and dates and never as a score.
• Keep the practice alive after day 90 with the routine of Chapter 8, and say plainly what a plan, a gate and a trigger can and cannot prove.

Why It Matters

A book about the boundary of the firm can leave an owner with a great deal of understanding and no idea what to do on Monday. Each chapter of this book answers a question: what vertical integration is, what the value chain maps, what the research on firm boundaries supports and what it does not, how backward differs from forward integration, how one grid scores a make-or-buy decision, what outsourcing and insourcing move, where partnerships sit, how the boundary reaches digital businesses and what the numbers say. None of them says in what order the questions are asked, who asks them, what record proves that an answer was reached, or when the answer is read again. That is what this chapter adds, and it adds no rule that the earlier chapters did not already hold.

The reason order matters is that a boundary decision made out of sequence is a decision made on the wrong evidence. A business that gets quotes before it has mapped its chain will compare the price of an activity it has not defined. A business that signs an agreement before it has run a pilot build will write terms for a process nobody has watched. A business that sets no trigger will read its decision again only when something has already gone wrong. The ninety days of this chapter put each step in an order in which it rests on the record the step before it produced.

The plan of this chapter is the book's own method, and it is posed by this book. No source read measures how long it takes to decide what to own, what a plan returns, how many hours a business should give to its sourcing or what trigger is the right one, and none is cited for one. The 121 hours of the plan, the $8,340 those hours come to at the printed illustrative rates, the $89,540 the plan commits with the duplicate test bench and the molds, the 89 hours a year of the routine that follows, and the three triggers of the case are all posed by this book and illustrative, and never a norm, a benchmark or a target for any reader. What a business takes from this chapter is the order of the work and the shape of the records, not the numbers of a fictional company.

In This Chapter

Section 11.1 sets out the six priorities that matter most and states the book's own sentence about deciding what to own. Section 11.2 names the ten most costly mistakes, with why each costs and the chapter that treats it. Section 11.3 lays out the ninety-day plan in three slices with a gate at the end of each. Section 11.4 covers days 1 to 30, the map, the forecast and the grid. Section 11.5 covers days 31 to 60, the pilot builds, the agreements and the molds. Section 11.6 covers days 61 to 90, the first supplier-built kits, the scorecard and the triggers. Section 11.7 reads the six readings that show whether the boundary is right. Section 11.8 says how to keep the practice alive after day 90. Section 11.9 reads the case on day 365, entirely posed by this book.

11.1 What Matters Most

Six priorities carry the whole of this book, and each rests on a chapter that has already done the work. They are stated in the order in which they are used, because the order is part of the advice: a business that starts with the fourth priority and works backward will pay for the mistake at the first gate.

The first priority is to map the chain and score each activity on one grid. The value chain is the map of what a business could own, make, buy or hold in between, and Book A3 (Strategy and Competitive Positioning, Chapter 2) holds the definition of the value chain and the activity system that this book cites and never rewrites. Chapter 1 places the boundary of the firm against that map, and Chapter 3 gives the one grid. The grid of the case carries six criteria with weights — core 25, capability 20, specificity 20, cost 15, risk 10 and control 10 — and the rule is make at 75 or more, hybrid from 55 to 75 and buy below 55. The grid is the book's own method, and it forces a business to write down why it keeps an activity or lets it go, in a form that a second person can read and disagree with.

The second priority is to compare made and bought on one basis, with the hours the business actually has. Chapter 7 holds the financial analysis of an integration decision: the same volume, the same quality, the same freight, the same oversight and the same hours on both sides. The case compares a kit made in-house at $175.50 with a kit bought at $202, a difference of $26.50 more per kit for buying, and the reason to buy is not price but capacity: the bench needs 713 staff installer hours in the year and the staff installer has 600 free after installations.

The third priority is to keep in-house what is core and specific, above all the firmware and the alert logic. Chapter 1 defines what is core and what is specific, and Chapter 3 scores both. The firmware and the mobile application are assets of the company, as Book H5 (Intellectual Property, Chapter 1) prints, and the grid of the case scores the firmware and the alert logic at 92, a verdict of make. Book I8 (Freelancers, Agencies and Outsourcing, Chapter 2) holds the core and the non-core, and this book cites that definition rather than writing a competing one.

The fourth priority is to write each agreement and to own the tooling that is specific to the business. Chapter 4 holds the agreement with a contract manufacturer or a service provider, and Chapter 5 holds the hybrid middle, where a long-term supply agreement, a dedicated capacity or a piece of tooling held at a supplier does the work that ownership would otherwise do. The case signs a supply and quality agreement and a tooling agreement, and it owns the second-generation molds, held at the molder. Book A5 (Financing a Business, Chapters 1, 3 and 6) holds the financing of the tooling and its loan.

The fifth priority is to keep the means to bring an activity back. Chapter 4 describes insourcing and the road back, and Chapter 6 describes what a business must keep when an activity leaves: the bench, the protocol and the records. The case keeps an in-house build cell for repairs, pilot builds and urgent kits, and it keeps the end-of-line test protocol and the duplicate test bench.

The sixth priority is to write each decision with its triggers and to read it again. Chapter 8 holds the cadence, and the triggers of the case are written beside the decision: assembly comes back in-house if the end-of-line defect rate exceeds 2% for 2 months running, if the price per kit exceeds $215, or if 3 releases in a year arrive late. A decision without a trigger is a decision that will be read again only when something has gone wrong.

DEFINITION Gate, trigger, register and reading, as this book uses the words

These are this book's working words, and no source read defines them in this sense. A gate is the day on which the conditions of a slice of the plan are read against records; the case's gates fall on days 30, 60 and 90, posed by this book. A trigger is a written condition that would bring an activity back in-house or move it, such as the three triggers of the case — a defect rate above 2% for 2 months running, a price above $215 a kit, or 3 late releases in a year — all posed by this book. A register is the written record that a business keeps for each activity, with its verdict, its reason, its owner, its triggers and its next review. A reading is a count or a date that a leader reads, never a score and never a feeling. The words are this book's own, and the case's figures are the case's own, posed by this book and illustrative.

KEY CONCEPT A business decides well what to own when it maps, scores, compares, keeps the core, writes its triggers and keeps the way back

A business decides well what to own when it maps its chain, scores each activity on one grid, compares made and bought on one basis, keeps in-house what is core and specific, writes each decision with its triggers, keeps the means to bring an activity back, and never promises a saving that it cannot prove. That sentence is the book's own reasoning and never a finding, and no source read measures what a plan returns. The case's figures are the case's own and are posed by this book: 121 hours of plan, $89,540 committed with the duplicate test bench and the molds, and three triggers written beside the decision to have the final assembly and the end-of-line test performed by the contract manufacturer. The sentence is read again once a year, in the yearly review of Chapter 8, and it is the standard against which the register of each activity is read.

11.2 The Ten Most Costly Mistakes

Ten mistakes are set out here, and each of them is a mistake of order, of record or of reading rather than a mistake of intelligence. Each one names the chapter of this book that treats it. No frequency of any mistake in any business is printed here, because no source read measures one, and the list is the book's own reasoning and illustrative.

The first mistake is deciding on the unit price alone. A quote is one number, and the comparison that matters is the total cost of the activity on both sides, on the same basis. The case's comparison shows why: the kit made in-house at $175.50 against the kit bought at $202 is $26.50 more per kit for buying, and a business that stopped at the price would have missed the capacity reason entirely. Chapter 7 treats this mistake.

The second mistake is outsourcing the core, the firmware and the alert logic. They are what the customer is paying for and assets of the company, as Book H5 (Intellectual Property, Chapter 1) prints, and the grid of the case scores them at 92, a verdict of make. Chapters 1 and 3 treat this mistake.

The third mistake is leaving a specific asset at a supplier without a written agreement. A mold made for someone else's product has few buyers, as Book A5 (Financing a Business, Chapter 3) prints, and a business that has paid for tooling and left it at a supplier with nothing in writing has given the supplier a hold over it. The case owns the second-generation molds and holds them at the molder under a tooling agreement. Chapter 5 treats this mistake.

The fourth mistake is forgetting the hidden costs of the transition. A move of work carries costs that are not in the quote: the duplicate equipment, the hours of the pilot builds, the lawyer's reading, the training and the weeks in which two ways of working run side by side. The case's plan commits $89,540, of which $9,200 is the duplicate test bench and $72,000 is the molds. Chapter 4 treats this mistake.

The fifth mistake is keeping no means to bring the work back. A business that has sold its bench, lost its protocol and stopped keeping its records has given up the option of insourcing. The case keeps an in-house build cell, the end-of-line test protocol and the duplicate test bench. Chapters 4 and 6 treat this mistake.

The sixth mistake is signing an exclusivity without reading it with a lawyer. An exclusive agreement can be a reasonable safeguard and it can also close a market to a business that later needs a second source. Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements, Chapter 5) holds exclusive dealing, tied selling and market restriction, and Book H1 (Contracts and Commercial Agreements, Chapters 5 and 6) holds the contract of enterprise or for services; a lawyer reads a given agreement, and only the Competition Tribunal or a court decides whether it is lawful. Chapter 5 treats this mistake.

The seventh mistake is agreeing on prices with a supplier that is also a competitor. An agreement between competitors on price, on markets or on supply is a different matter from a supply agreement, and the rules that apply are those of Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements, Chapter 5) and of a lawyer. Chapter 1 treats this mistake.

The eighth mistake is integrating a stage without the volume or the skill. A stage that a business owns must be fed with enough volume to justify the fixed cost and staffed with people who can run it. The case's molds pay back after 1,028.6 kits, in year 4 of a five-year life of 1,400 kits; the numbers are thin, and control of a specific asset decides. Chapters 2 and 3 treat this mistake.

The ninth mistake is buying a company without reading the merger rules with a lawyer. The purchase of a supplier or of a customer can be a merger within the Act's definition, and Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements, Chapter 5) holds the notification thresholds and the review; a threshold decides only who must notify and is never an exemption. Buying a business lies outside this book, and Book M6 (Buying a Business) is only named here. Chapter 7 treats this mistake.

The tenth mistake is never reading the decision again. A decision that was right at one volume, one price and one level of skill can be wrong a year later, and a decision with no trigger and no yearly review will be read again only when something has gone wrong. The case writes three triggers beside the decision and reads them at the quarterly review. Chapter 8 treats this mistake.

The table below sets the ten mistakes beside why each costs and the chapter of this book that treats it. It is the book's own reasoning and not a finding, it adds no rule, and no frequency of any mistake in any business is printed. Read it as a checklist to be read once before a decision and once a year after it.

Table 11.1 — The ten most costly mistakes and where to read them

WATCH OUT Starting the second slice before the first gate passes, and reading a met gate as a result, are two traps

Two traps catch a ninety-day plan, and each has a question that catches it. The first is starting the second slice before the first gate passes: the plan runs ahead of its own records, the pilot builds are ordered before the grid is signed, and the question is which record shows that each part of the condition is met, and whether the plan waits 7 days when it is not. The second is reading a met gate as a result: the records exist, the slice is closed, and the move is called a success, when all the gate shows is that the records its condition names exist. The question is what the gate shows and what it does not show, since the figures of the case are posed by this book and nothing is measured. The book's reasoning is illustrative, and no frequency of either trap in any business is printed. The case reads each gate against its records and credits nothing to the plan, posed by this book.

IN LA MACHINE A task marked done proves nothing, and an inventory says what is installed and not how much a business depends on it

In the application copy read on October 15, 2026, Control center → Plan tasks (/taches) holds the routine as recurring tasks with a required "Done when" criterion, and the Mark as done button does not verify that criterion, so a second person reads it (La Machine, 2026). Control center → Tools (inventory) (/outils) lists the installation's tools with a measured version and date, and it measures no dependence on a supplier of goods, no total cost and no volume, so the exposure is measured elsewhere. A company record holds no contract, no price and no performance, so the sourcing decision lives in a written register that the business keeps, and no route is named for that register because the platform holds none. La Machine keeps no register of sourcing decisions and their triggers; check it in your installation. La Machine is described and never sold, and the case does not use it.

11.3 Deciding What to Own in Ninety Days

The plan of this chapter covers ninety days in three slices of thirty days, and each slice ends at a gate. The plan is the book's own method, posed by this book, and it is never a norm, a benchmark or a promise. It answers one question: in what order, by whom and with what proof of completion does a small firm, in ninety days, map its chain, forecast its volume, score its activities on one grid, get quotes, run pilot builds, sign written agreements, own the tooling that is specific to it, set its scorecard and write its triggers, without putting its firmware or its alert logic outside the business?

A gate is passed by the records its condition names, and by nothing else. The condition of each gate is written before day 1, so that the plan cannot move its own target. When every part of a condition is met, the gate is approved and the next slice starts. When a part is missing, the part is named and the gate is read again 7 days later, and nothing new starts until it passes. A gate that is met shows that the records exist; it does not show that the move works, and it does not prove that the decision was right. A ninety-day plan does not settle the boundary for good, and the decisions are read again on triggers and every year, in the routine of Chapter 8.

The plan takes 121 hours, and the hours are posed by this book. Laurence Bergeron, the president, product and engineering, works 62 hours; Karim Haddad, the head of sales, sales and operations, works 36 hours; the staff installer works 16 hours; the accountant works 2 hours; and the outside lawyer works 5 hours. At the printed illustrative rates — $60 an hour for each founder, $45 an hour for the staff installer, $120 an hour for the accountant and $300 an hour for the outside lawyer — the hours come to $8,340. With the duplicate test bench at $9,200 and the second-generation molds at $72,000, the plan commits $89,540 in ninety days. Every one of those figures is posed by this book and illustrative, and no hour and no amount is a norm for any reader.

The routine of Chapter 8 follows day 90 and is never added to the plan. It takes 89 hours a year, posed by this book, and it holds the weekly build release, the monthly scorecard, the quarterly review of the triggers, the yearly review of every activity, the yearly audit at the contract manufacturer and the yearly check of the molds at the molder. The plan builds the boundary; the routine keeps it read.

Figure 11.1 sets out the plan of ninety days, its gates and its hours, the routine and the triggers that follow, and the readings one year later.

Figure 11.1 — Ninety days, three gates. 121 hours and $89,540 committed, then 89 hours a year of routine and three triggers; one year later, 236 kits and nothing brought back, with nothing credited (posed).

The table below sets the three gates beside their day and the condition that passes each. The conditions are written before day 1 and read against records, and a gate that is not met is read again 7 days later while nothing new starts. Read it as the spine of the plan: each row is a day on which the plan either moves forward or waits.

Table 11.2 — The three gates of the ninety days

The table below sets the hours of each slice beside each person. The figures are posed by this book and illustrative, the $8,340 that the total of 121 hours comes to at the printed rates is given in the paragraph above the table, and no hour is a norm. Read the columns as the load each person carries through the plan, and read the total row as the plan's own cost in hours.

Table 11.3 — The hours of the plan by slice and by person (posed)

METHOD Running a gate in five steps

First, read the condition of the gate, written before day 1, so that the plan cannot move its own target. Second, find, for each part of the condition, the record that shows it: the map, the forecast, the signed grid, the quotes, the pilot-build acceptance, the signed agreements, the samples, the first release, the scorecard and the written triggers. Third, mark each part met or not met, with the day on which it was read. Fourth, when every part is met, approve the gate and start the next slice, and when a part is missing, name it and read the gate again 7 days later while nothing new starts. Fifth, write what the gate showed and what it did not show, because a gate that is met shows that the records exist and never that the move works. This is the book's reasoning and not a tested procedure. The case follows these steps at days 30, 60 and 90, posed by this book.

THE RULE A gate is passed by the records its condition names, and a missed gate is read again 7 days later while nothing new starts

The rule is this book's own for the plan, drawn from its reasoning and not from a source. The condition of each gate is written before day 1, a gate is passed only when the records the condition names exist, and a gate that is not met is read again 7 days later while nothing new starts. The rule is neither a standard nor a legal duty, and it promises no result. Its conditions can be written as plan tasks with an observable "Done when" criterion, and the Mark as done button does not verify that criterion (La Machine, 2026). The case's gates fall on days 30, 60 and 90, posed by this book, and no figure beyond the case's is printed here. A business that writes its conditions after the fact has written a description of what it did, not a gate.

11.4 Days 1 to 30: The Map, the Forecast and the Grid

The first slice builds what the rest of the plan stands on. Nothing is moved, nothing is signed and nothing is ordered in these thirty days; the slice produces the four records that every later decision will cite, and it ends with two quotes in hand.

The chain of the company's activities is mapped on day 9. The map follows the method of Chapter 2, and it covers the whole chain of Hydrosentinelle inc.: the design of the connected water-leak detectors, the electronic boards bought from the electronics supplier of Book A5, the tooling Book A5 (Financing a Business, Chapters 1 and 3) financed, the assembly and the end-of-line test in the company's own workshop, the installation through the staff installer and the certified installers of Book F9 (Building and Running a Partner Program, Chapter 4), and the monitoring service and the alert logic that the company owns. Laurence Bergeron, the president, product and engineering, holds the technical side of the map.

The kit forecast is written on day 18. The forecast is 230 kits for the year, posed by this book, and it is the volume against which every comparison in the plan is made: the hours at the bench, the price per kit made and bought, the break-even of the molds and the size of the first release. Karim Haddad, the head of sales, sales and operations, holds the forecast.

The six activities are scored on one grid on day 24. The grid carries the six criteria with their weights, and the rule is make at 75 or more, hybrid from 55 to 75 and buy below 55. The verdicts are these: the firmware and the alert logic score 92, a verdict of make; the enclosure tooling for the second generation scores 68, a verdict of hybrid; the electronic boards score 35, a verdict of buy; the final kit assembly and the end-of-line test score 57, a verdict of hybrid; the installation in buildings scores 71, a verdict of hybrid; and the cloud hosting of the monitoring service scores 35, a verdict of buy. The grid and its verdicts are posed by this book, and they are the ones Chapter 3 prints.

The verdicts are written and signed by both founders on day 28, and the accountant reads the cost comparison before the gate. The signature turns a discussion into a decision that a later reader can hold the founders to. The accountant's rate is the printed illustrative rate of $120 an hour.

Two quotes for the final assembly and test of the kits are received on day 30. The quotes are the input to the comparison of Chapter 7, and they are read on the same basis: the same volume, the same quality, the same freight and the same oversight. The case's comparison comes to $175.50 a kit made in-house and $202 a kit bought, a difference of $26.50 more per kit for buying, and the reason to buy is capacity rather than price. The quotes are posed by this book.

The hours of the slice are in Table 11.3. The gate of day 30 is read against the records of its condition: the map, the forecast, the signed grid, the signed verdicts and the two quotes.

The table below sets the five tasks of the first slice beside who does them and the day each is due. The hours of the slice are those of its row of Table 11.3, with no total of the slice computed here, and all of it is posed by this book. Read it as the work of the first thirty days, in the order in which the records are produced.

Table 11.4 — Days 1 to 30: the tasks of the first slice

WHO DOES THIS Laurence Bergeron and Karim Haddad share the three slices, and the outside lawyer and the accountant read for named jobs

The roles are the case's, posed by this book. Laurence Bergeron, the president, product and engineering, holds the technical side of the chain map, the end-of-line test protocol, the duplicate test bench, the pilot builds, the molds, the second source and the audits, 62 hours in the plan. Karim Haddad, the head of sales, sales and operations, holds the forecast, the quotes, the commercial terms, the weekly release, the scorecard and the triggers, 36 hours. The staff installer keeps the in-house build cell and supports the pilot builds, 16 hours. The accountant reads the cost comparison, 2 hours. The outside lawyer reads the supply and quality agreement and the tooling agreement, 5 hours. The contract manufacturer and the molder perform under agreements. No one is hired. In general, one person holds each gate with a named backup, which is the book's suggestion and not part of the case. In a solo business the owner holds every slice and engages a lawyer and an accountant for named jobs; in a larger business an operations lead and a purchasing lead hold the slices. No hour and no amount is promised to a reader.

11.5 Days 31 to 60: Pilot Builds, Agreements and the Molds

The second slice moves the final assembly and the end-of-line test under written terms. It is the slice in which the plan stops describing and starts committing, and it is the slice in which the records of the first slice are tested against a real build.

The duplicate test bench is delivered to the contract manufacturer on day 41. The bench is the company's own end-of-line test equipment, duplicated so that the contract manufacturer can run the company's test on the company's protocol, and it costs $9,200, posed by this book. It is the physical form of the fifth priority: the means by which the company could bring the assembly back in-house, and the means by which the contract manufacturer runs the test the company specifies.

Two pilot builds are run, of 10 kits and of 25 kits, and both are accepted, on days 47 and 54. Each is followed by a first-article inspection by Laurence Bergeron, the president, product and engineering, which is the record that the process produces what the drawing says. Quality management is Book J4 (Quality Management and Continuous Improvement, Chapter 6), and this book cites that book rather than describing its methods. A quote is a price for a process nobody has watched, and a pilot build is the process watched.

The supply and quality agreement is signed on day 57, after the outside lawyer's reading. It carries performance indicators on quality, production capacity, stock levels and delivery times and confidentiality clauses, as Chapter 4 describes. Scoping, contracts and the management of external work are Book I8 (Freelancers, Agencies and Outsourcing, Chapters 5 to 7), and this book cites that book rather than rewriting it. The outside lawyer's rate is the printed illustrative rate of $300 an hour, and the lawyer's reading is 4 hours in this slice.

The tooling agreement is signed and the second-generation molds are ordered on day 58. The molds cost $72,000, posed by this book, and they give an enclosure set at $64 a kit against $134 a kit if the molds are left to the molder. The difference of $70 a kit pays back the molds after 1,028.6 kits, in year 4 of a five-year life of 1,400 kits, and the net present value of owning them is $3,068 at a posed rate of 9% and minus $8,304 at 15% less volume. The numbers are thin, and control of a specific asset decides. The financing of the tooling and its loan are Book A5 (Financing a Business, Chapters 1, 3 and 6), and this book cites that financing rather than recounting it.

The move of the final assembly to the contract manufacturer can appear in the application this book describes as a chain of tasks: Control center → Projects (chains) (/projets) shows the projects of the plan as chains of tasks, in read-only mode, and no page creates one (La Machine, 2026). The hours of the slice are in Table 11.3. The gate of day 60 is read against the records of its condition: the delivery of the bench, the two accepted pilot builds, the signed supply and quality agreement, and the signed tooling agreement with the molds ordered.

The table below sets the five tasks of the second slice beside who does them and the day each is due, and all of it is posed by this book. Read it as the work of the middle thirty days, in the order in which the commitments are made.

Table 11.5 — Days 31 to 60: the tasks of the second slice

QUÉBEC & CANADA NOTE The competition, contract, employment and privacy rules are called to Books H4, H1, H7 and H2, and this book recommends no vendor

The Canadian point of this section is where the rules of a supply and tooling agreement live. The competition statute is called to Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements, Chapter 5), the contracts and the contract of enterprise or for services to Book H1 (Contracts and Commercial Agreements, Chapters 5 and 6), the employment statute and the move of work to Book H7 (Employment Law Essentials (Québec and Canada)), and personal information to Book H2 (Privacy and Personal Information, Chapter 8), in one sentence each. The supply and quality agreement and the tooling agreement of the case are signed after the outside lawyer's reading, on days 57 and 58, posed by this book. The case has no union and makes no acquisition. Only the Competition Tribunal or a court decides whether an agreement or a merger is lawful. This book explains published rules and recommends no vendor, and a business with a question about its own agreement should ask a lawyer; nothing here says whether the reader's own agreement complies.

11.6 Days 61 to 90: The First Supplier-Built Kits, the Scorecard and the Triggers

The third slice sets the routine. The commitments of the second slice are now running, and the third slice produces the records that will be read every month and every quarter for as long as the decision stands.

Samples of the second source of the sensor component are received on day 68. The second source is the answer to the exposure that Book G8 (Business Insurance and Risk Financing, Chapter 3) names, where a supplier's component shortage that stops kit assembly is rated at $40,000 and 30 days, reduced by a second source and a safety stock. The second source is qualified on day 120, after day 90, and holds 30% of the sensor component by day 365. The samples are posed by this book.

The first release of 25 supplier-built kits is delivered on day 84. It is the first production output of the contract manufacturer under the signed agreement, the record that the pilot builds were not a special case, and it is read against the same end-of-line test that the company runs on its own kits. The release is posed by this book.

The monthly scorecard is live on day 88. It carries the performance indicators that the supply and quality agreement names, and it is read every month by Karim Haddad, the head of sales, sales and operations, with the quality lines read by Laurence Bergeron, the president, product and engineering. It turns the agreement into a monthly reading and is the record that the quarterly review of the triggers will cite. The scorecard is posed by this book.

The triggers that would bring assembly back in-house are written beside the decision. There are three: assembly comes back in-house if the end-of-line defect rate exceeds 2% for 2 months running, if the price per kit exceeds $215, or if 3 releases in a year arrive late. The triggers are posed by this book, and no source read gives any trigger for any business. The triggers are read at the quarterly review of Chapter 8.

The routine of Chapter 8 is set as plan tasks with a verification criterion, and the plan tasks of the application this book describes are described in section 11.2 only. The hours of the slice are in Table 11.3. The gate of day 90 is read against the records of its condition: the samples of the second source, the first release of 25 supplier-built kits, the live scorecard and the written triggers.

The table below sets the four tasks of the third slice beside who does them and the day each is due, and all of it is posed by this book. Read it as the work of the last thirty days, in the order in which the routine is set.

Table 11.6 — Days 61 to 90: the tasks of the third slice

TOOLS OF THE TRADE What any record of a gate must offer, whatever it is called

Eight functions carry a gate, and a business should look for them rather than for a product. The first is a list of the conditions of each gate, each with a status. The second is a date for each condition and a person who owns it. The third is a link to the record that shows the condition met, whether that record is a map, a forecast, a signed grid, a quote, an acceptance, a signed agreement, a sample or a release. The fourth is a place to write the decision and what the gate did not show. The fifth is a read-again date that falls 7 days later when a part is missing. The sixth is a history that cannot be rewritten, so that a later reader can see what was known on the day. The seventh is an export in a form another tool can read, because the register of the decision will outlive the tool. The eighth is a way to name a backup for the person who owns each condition. No brand name belongs on this list and no price; the requirement is the function.

11.7 The Few Readings That Show the Boundary Is Right

Six readings show whether the boundary is right, and all six are counts or dates. None of them is a score, none of them is a feeling, and none of them is called typical of anything. They are the readings of the case, posed by this book, and they are read at the monthly scorecard, at the quarterly review of the triggers and at the yearly review of every activity.

The first reading is the staff installer hours freed, the count of hours that returned to installations; the question is whether they went where the plan said they would go, because hours freed and not redeployed are hours lost.

The second reading is the first-pass yield at the end-of-line test, the count of kits that pass the company's test on the first attempt, read on the duplicate bench and on the company's own bench; a yield that drifts down is the first sign that a trigger is approaching.

The third reading is the releases on time, the count of weekly build releases that arrived on time; the question is which were late and why, because the third trigger of the case is 3 late releases in a year.

The fourth reading is the lots held and for how many days, the count of lots set aside and the days each was held; a lot released without a change is a lot that will be held again.

The fifth reading is the share of the sensor component bought from the second source, read against its qualification; it answers the exposure that Book G8 (Business Insurance and Risk Financing, Chapter 3) names, and a second source that is qualified and never used is not a second source.

The sixth reading is the count of kits built with the second-generation molds, set against the break-even of 1,028.6 kits; the question is how far the count is from the break-even and how fast it is moving.

The table below sets six readings beside the count or the date a leader reads and the question to ask. No reading is a score and none is a feeling, and every cell is the book's own reasoning and illustrative, with no frequency in any business. Read it as the short list that a leader keeps in view between the monthly scorecard and the yearly review.

Table 11.7 — Six readings that show the boundary is right

The readings of the case on day 365 are set out in section 11.9, and every one of them is posed by this book.

11.8 Keeping the Practice Alive

After day 90 the routine of Chapter 8 runs, and it is never added to the 121 hours of the plan. The routine takes 89 hours a year, posed by this book, and it holds the work that keeps the boundary read: the weekly build release goes out to the contract manufacturer, the monthly scorecard is read, the quarterly review reads the three triggers, and the yearly review rescores the six activities on the grid. The routine is the difference between a decision that was made once and a decision that is kept.

The weekly build release is the shortest line of the routine and the one that keeps the flow visible. Karim Haddad, the head of sales, sales and operations, sends it to the contract manufacturer, and it is the record against which the third trigger — 3 late releases in a year — is read. The monthly scorecard, with the quality lines read by Laurence Bergeron, the president, product and engineering, is the record against which the first trigger — a defect rate above 2% for 2 months running — is read. The quarterly review reads the three triggers together, and the yearly review rescores the six activities on the grid, because a score that was right at one volume, one price and one level of skill can be wrong a year later.

The rules of law and the pages of the sources that the earlier chapters cite are read again on a set date, and their state after the date they were read is not stated. A business that keeps a register of its decisions should keep, beside each decision, the date on which the rules it relied on were read, so that a later reader can see what was known on the day.

The book's own sentence is read again once a year, in the yearly review of every activity: map the chain, score each activity on one grid, compare made and bought on one basis, keep in-house what is core and specific, write each decision with its triggers, keep the means to bring an activity back, and promise no saving that cannot be proved. The sentence is the standard against which the register of each activity is read, and it is the book's own reasoning and never a finding.

When the practice slips, the remedy is small and it is written down. The missed task is done on the next working day rather than folded into the next cycle, the miss is recorded with its reason, and its cause is read as a fact about the system rather than a person's fault. Where the miss belongs to the plan of this chapter, the gate is read again 7 days later, and nothing new starts until it passes. A routine that depends on a calendar and a person will not fade when the owner's attention moves.

DO THE MATH The 121 hours of the plan, the 89 hours a year of the routine and the $89,540 committed, line by line

Illustrative figures; the case is fictional. Step 1 — Laurence Bergeron works 20 hours in days 1 to 30, 30 in days 31 to 60 and 12 in days 61 to 90, which is 62 hours. Step 2 — Karim Haddad works 14, 12 and 10 hours, which is 36 hours, and the staff installer works 4, 8 and 4 hours, which is 16 hours. Step 3 — the accountant works 2 hours, and the outside lawyer works 5 hours, 4 in days 31 to 60 and 1 in days 61 to 90. Step 4 — the plan takes 62 plus 36 plus 16 plus 2 plus 5, which is 121 hours, and at the printed illustrative rates the hours come to $8,340. Step 5 — with the duplicate test bench at $9,200 and the molds at $72,000 the plan commits $89,540, and the routine that follows takes 89 hours a year. The hours, the bench and the molds are posed by this book, the rates are illustrative and not a pay, nothing is credited to the plan, and the sums are exact.

11.9 The Case, One Year Later

Day 365 is entirely posed by this book, and nothing in it is credited to the plan or to any decision. It is a set of readings, not a result, and it is read the way the readings of section 11.7 are read: as counts and dates, with the question that each one raises.

Hydrosentinelle inc. ships 236 kits by day 365, of which 213 are built by the contract manufacturer and 23 in-house, posed by this book. The 23 in-house kits are the ones the in-house build cell keeps for repairs, pilot builds and urgent kits, and the cell is the means by which the company could bring the assembly back. The first-pass yield, the lot held, the releases on time and the freed hours are in Table 11.8, posed by this book.

The first samples of the second-generation molds come on day 128 and the first production enclosures on day 150, both after day 90 and both posed by this book. No trigger is reached, nothing is brought back in-house, and no one is hired, all posed by this book.

The table below sets the readings of the case on day 365 beside what each shows. Every line is posed by this book, nothing is measured, and no result of the business is credited to the plan. Read it as the last page of the case and as the first page of the next year's review.

Table 11.8 — The case on day 365, entirely posed by this book

The chapter ends the book here, and the work it describes does not end with it. The routine of Chapter 8 holds the boundary after day 90: the weekly build release, the monthly scorecard, the quarterly review of the triggers and the yearly review of every activity. A business that keeps that routine, keeps its register, keeps its bench, its protocol and its records, and reads its decisions again on triggers and every year, has done what this book can teach. Nothing in the case is credited to the plan, and nothing in the plan is a promise.

WHAT CHANGED The application copy was read on October 15, 2026, and the plan is read again at each gate

The dated reading is this one: the application code was read at its copy of September 7, 2026, on October 16, 2026 (La Machine, 2026). In that copy the monthly scorecard and the quarterly review of Chapter 8 can live as recurring plan tasks whose "Done when" the Mark as done button does not verify, and the plan projects appear read-only. Every reading of a rule or of a source is in the chapter that holds it, and this chapter assembles them rather than restating them. None of these dates is a date of the case, which is counted in days, and the readings of day 365 belong to section 11.9, never to this box. A reader checks every page and every term on the reader's own date and reads the plan again at each gate, on days 30, 60 and 90.

CASE IN POINT Hydrosentinelle inc. passes three gates in ninety days and brings nothing back in-house in the year

Hydrosentinelle inc. is a fictional business created for this encyclopedia; its founders, its employees, its customers, its partners, its suppliers, its figures and the events described are invented, and any resemblance to a real business is coincidental. In ninety days the company maps its chain, forecasts 230 kits, scores six activities, gets two quotes, runs two pilot builds, signs a supply and quality agreement and a tooling agreement, orders the molds, receives samples of a second source, takes the first release of 25 supplier-built kits and sets its scorecard and its triggers, for 121 hours and $89,540 committed. On day 365 the readings are posed and nothing is credited to the plan. All of it is posed by this book, illustrative and not a forecast, and no verdict and no result is drawn from it.

Costly Mistakes

• Starting the second slice before the first gate passes. The plan runs ahead of its own records, the pilot builds are ordered before the grid is signed, and the gate is read afterward as a description of what was done rather than as a condition that was met.
• Reading a met gate as a result. A gate that is met shows that the records its condition names exist; it does not show that the move works, and the figures of the case are posed by this book and nothing is measured.
• Writing the conditions of a gate after the fact. A condition written after the slice has run is a description of what happened, and it cannot fail, which means it cannot tell a business anything.
• Leaving the register of a decision in a conversation. A verdict that lives in the memory of the people who were in the room will be remembered differently by each of them, and the triggers will be read against a decision nobody can quote.
• Setting no trigger and no yearly review. A decision that was right at one volume, one price and one level of skill will be read again only when something has gone wrong, which is the moment at which the reading is least useful.
• Treating the routine as part of the plan. The routine of Chapter 8 follows day 90 and takes 89 hours a year, posed by this book; adding it to the 121 hours of the plan hides the cost of keeping the boundary read.
• Crediting a result to the plan. Nothing in day 365 is credited to the plan or to any decision, and a business that reads a count as a proof of its own method has stopped reading the count.

Key Takeaways

• Six priorities carry the whole of this book: map the chain and score each activity on one grid, compare made and bought on one basis with the hours the business has, keep the core and the specific in-house, write each agreement and own the specific tooling, keep the means to bring an activity back, and write each decision with its triggers and read it again.
• Ten mistakes cost a boundary decision dearly, and each is a mistake of order, of record or of reading, from deciding on the unit price alone to never reading the decision again; Table 11.1 names the chapter that treats each one.
• The plan of this chapter covers ninety days in three slices, each ending at a gate on days 30, 60 and 90; a gate is passed by the records its condition names, and a gate that is not met is read again 7 days later while nothing new starts.
• The plan takes 121 hours and commits $89,540 with the duplicate test bench and the molds, and the routine that follows takes 89 hours a year; every one of those figures is posed by this book and illustrative, and none is a norm, a benchmark or a promise.
• Six readings show whether the boundary is right, and all six are counts or dates: the staff installer hours freed, the first-pass yield at the end-of-line test, the releases on time, the lots held and for how many days, the share of the sensor component bought from the second source, and the kits built with the new molds against the break-even.
• Day 365 is entirely posed by this book, and nothing in it is credited to the plan or to any decision: 236 kits shipped, a first-pass yield of 98.6%, 21 of 23 releases on time, no trigger reached, nothing brought back in-house and no one hired.

Your Action Plan

1. Write the six priorities into one page and read it before any boundary decision. State the map, the grid, the comparison on one basis, the core and specific activities kept in-house, the agreements and the tooling, the means to bring an activity back, and the triggers. Deliverable: a one-page statement of the six priorities, with the chapter of this book that treats each. Verification: every priority names the record it produces, and the page is read before the first quote is requested.
2. Write the three gates before day 1, with the record that shows each part of each condition. Set the conditions of days 30, 60 and 90 in writing, and name, for each part, the record that will show it met. Deliverable: a written gate sheet with three gates and their conditions. Verification: every part of every condition names a record, and no condition can be met by an intention.
3. Run the first slice and read the first gate against its records. Map the chain, write the forecast, score the activities on one grid, sign the verdicts and receive two quotes, then read the gate on day 30 and mark each part met or not met. Deliverable: the map, the forecast, the signed grid, the signed verdicts and the two quotes. Verification: every part of the day-30 condition has its record, and a missing part is named and read again 7 days later while nothing new starts.
4. Run the second and third slices, and set the routine as plan tasks with an observable criterion. Deliver the duplicate bench, run the two pilot builds, sign the supply and quality agreement and the tooling agreement, order the molds, receive the second-source samples, take the first release, put the scorecard live and write the triggers. Deliverable: the signed agreements, the pilot-build acceptances, the live scorecard and the written triggers. Verification: the day-60 and day-90 conditions are each read against their records, and each recurring task carries a criterion a second person can check.
5. Keep the register and read it again on triggers and every year. Keep, for each activity, its verdict, its reason, its owner, its triggers and its next review, and read the register at the quarterly review of the triggers and at the yearly review of every activity. Deliverable: a written register of the boundary decisions. Verification: every activity in the register has an owner, a trigger and a next review date, and at least one review has been held under the routine of Chapter 8.

In La Machine

In the application copy read on October 15, 2026, the routine of the plan can live in Control center → Plan tasks (/taches) as recurring tasks with a required "Done when" criterion, and the Mark as done button does not verify that criterion, so a second person reads it (La Machine, 2026). Control center → Tools (inventory) (/outils) lists the installation's software tools with a measured version and date, for software only. Control center → Projects (chains) (/projets) shows the projects of the plan as chains of tasks, in read-only mode, and no page creates one. Check it in your installation.

What a business needs elsewhere is the rest of the practice. La Machine keeps no register of sourcing decisions and their triggers, holds no supplier contract, no price and no performance, measures no dependence on a supplier of goods, and verifies no "Done when" criterion; it also holds no purchase order, no bill of materials, no supplier record or scorecard, no inventory of parts, no production order, no register of tooling and no cost, break-even, present-value or make-or-buy model. A written register of sourcing decisions with an owner and triggers, a place for the supplier's contract and the test records, and a way to measure the dependence on a supplier live in a spreadsheet, a database or a tool the business chooses by function, and this book is valid with any tool. La Machine is one worked example of a business platform, described and never sold, and the case does not use it.

Self-Check

1. What are the six priorities that matter most in a boundary decision, and which chapter of this book treats each?
2. What are the ten most costly mistakes, and what kind of mistake is each one?
3. What is a gate, what passes it, and what happens when a gate is not met?
4. What does the plan of ninety days take in hours and commit in money, and what does the routine that follows take?
5. What are the six readings that show whether the boundary is right, and what does day 365 show?

Answers. 1. The six priorities are to map the chain and score each activity on one grid, treated in Chapters 1 and 3; to compare made and bought on one basis with the hours the business actually has, treated in Chapter 7; to keep in-house what is core and specific, above all the firmware and the alert logic, treated in Chapters 1 and 3; to write each agreement and own the tooling that is specific to the business, treated in Chapters 4 and 5; to keep the means to bring an activity back, treated in Chapters 4 and 6; and to write each decision with its triggers and read it again, treated in Chapter 8. 2. The ten mistakes are deciding on the unit price alone, outsourcing the core, leaving a specific asset at a supplier without a written agreement, forgetting the hidden costs of the transition, keeping no means to bring the work back, signing an exclusivity without reading it with a lawyer, agreeing on prices with a supplier that is also a competitor, integrating a stage without the volume or the skill, buying a company without reading the merger rules with a lawyer, and never reading the decision again; the book calls each a mistake of order, of record or of reading, and none of them a mistake of intelligence. 3. A gate is the day on which the conditions of a slice of the plan are read against records, and the case's gates fall on days 30, 60 and 90, posed by this book; a gate is passed by the records its condition names and by nothing else, and a gate that is not met is read again 7 days later while nothing new starts. 4. The plan takes 121 hours, which is $8,340 at the printed illustrative rates, and with the duplicate test bench at $9,200 and the molds at $72,000 it commits $89,540, all posed by this book; the routine that follows takes 89 hours a year, also posed by this book, and it is never added to the 121 hours of the plan. 5. The six readings are the staff installer hours freed, the first-pass yield at the end-of-line test, the releases on time, the lots held and for how many days, the share of the sensor component bought from the second source, and the kits built with the new molds against the break-even of 1,028.6 kits; day 365 is entirely posed by this book and shows 236 kits shipped, a first-pass yield of 98.6%, 21 of 23 releases on time, no trigger reached and nothing brought back in-house, with nothing credited to the plan or to any decision.

Chapter References

La Machine. (2026). Documentation — route reference inventory. Internal product documentation, read October 16, 2026, together with the code of the pages this book describes: /taches, /outils and /projets.