Chapter 1 — Definition, Landscape and Stakes

A banker once described the moment she stops reading. It is not a bad number, she said, and it is not a weak idea. It is the sentence where the plan tells her that the market is growing fast, and there is no source under it. From that point on she reads the rest of the document differently, because she now knows that at least one number in it was written by someone who did not check. This chapter is about writing a document that survives that reader — and, more importantly, about writing one that survives the person who has to live with it afterwards, which is you.

What You'll Be Able to Do

• Define a business plan precisely, and say how it differs from a business model, a strategy, a marketing plan and a budget.
• Name the three jobs a plan does, and decide which one is primary for your own document before you write a word.
• Predict what each of the five readers opens first, and what makes each of them stop reading.
• Cite what the evidence actually says about the relationship between planning and business performance, rather than repeating either of the two folk beliefs about it.
• Name the four constraints — time, money, information and skill — that shape every real plan, and size yours honestly.
• Decide whether you need a full plan, a short plan or something else entirely, using four questions rather than instinct.

Why It Matters

There is a particular kind of waste that this book exists to prevent. It happens when someone spends four weekends writing a forty-page document, sends it to a lender, receives a polite refusal that explains nothing, and never opens the file again. The work was real. The document was not useless. But it was written for the wrong job, and nobody said so.

The second reason is quieter and it compounds. A business that has written down its assumptions can tell, twelve months later, which ones were wrong — and that is the entire mechanism by which a small company gets better at judging its own market. A business that never wrote them down experiences the same twelve months as a sequence of surprises. The plan's value as a persuasion document is real but occasional. Its value as a record of what you believed and why is permanent, and it is available to every business, including the ones that will never borrow a dollar.

Finally, there is the arithmetic of the thing. Federal statistics on business survival, which §1.4 gives in full, show that most of the mortality happens between the third and the tenth year rather than in the first. The failures that matter are therefore rarely the ones caused by a bad first idea. They are caused by a business that never wrote down what it was assuming, and so could not notice that the assumption had stopped being true.

In This Chapter

• 1.1 What a Business Plan Is, and What It Is Not
• 1.2 The Three Jobs a Plan Does
• 1.3 The Five Readers, and What Each One Looks For First
• 1.4 The Landscape: What the Evidence Says About Planning
• 1.5 The Challenges That Make Plans Fail Before They Are Read
• 1.6 What a Business Actually Needs From Its Plan
• 1.7 The Constraints: Time, Money, Information and Skill
• 1.8 How This Book Is Organized

1.1 What a Business Plan Is, and What It Is Not

Start with the definition, because almost every dispute about business plans turns out to be a dispute about what the words mean.

DEFINITION Business plan A business plan is a dated, written document that states what a business intends to do over a defined horizon, what it assumes about the world in order to believe that intention is achievable, what it will cost, what it expects to earn, and how the reader can tell whether it is working. Three properties are doing the work in that sentence. It is dated, so it can be compared with reality later. It is written, so it can be read by somebody who was not in the room. And it states its assumptions, so that a disagreement about the conclusion can become a disagreement about a specific input, which is the only kind of disagreement that can be settled.

That definition deliberately excludes several things the term is often used to cover.

A business plan is not a business model. The model is the logic by which the business creates value, delivers it and captures part of it in revenue — who the customer is, what they get, what they pay, what it costs to serve them, and why they choose you. The model is the content; the plan is the document that commits the model to a horizon, a budget and a calendar. You can have a perfectly good model and no plan, and you can have a beautifully formatted plan sitting on top of a model nobody has tested. → see Book A1 (From Idea to Business Model) for the validation work that must happen before this book is useful.

A business plan is not a strategy. Strategy is the set of choices about where to compete and how to win, including the deliberate decision about what not to do. A plan records the strategy and translates it into resources and dates, but writing a plan does not produce a strategy any more than writing a recipe produces a meal. A plan whose strategy section says "we will offer superior service at competitive prices" has recorded the absence of a strategy in a full sentence. → see Book A3 (Strategy and Competitive Positioning).

A business plan is not a marketing plan. The marketing plan lives inside it, compressed to two or three pages — the segments, the channels, the budget and the expected results. The full apparatus of positioning, personas, campaigns and measurement belongs elsewhere. → see Book B1 (Marketing Fundamentals for Business Owners).

A business plan is not a budget. A budget is a single-scenario allocation of money over a period, usually twelve months, usually built from last year's actuals. A plan contains a budget, extends further out, carries the reasoning that produced the numbers, and includes everything the budget cannot express — the market argument, the operating model, the risks and the milestones. A budget answers how much; a plan answers why that much, and what happens if we are wrong.

KEY CONCEPT A plan is a decision record, not a persuasion document The most useful way to think about the artifact you are about to write is that it records a set of decisions and the evidence behind each one, in a form that lets a future reader — including you — reconstruct the reasoning. Persuasion is a consequence of that record being credible, never a substitute for it. This inversion matters because it changes what you do when you hit a weak spot. If the plan is a persuasion document, a weak spot is something to phrase around. If it is a decision record, a weak spot is something to write down, because the note "we do not yet know what the churn rate will be; we have assumed 8% monthly and will measure it by March" is more credible than a confident number, and infinitely more useful in March.

1.1.1 The three horizons, and why the middle one is the hardest

Plans are written over a horizon, and the horizon changes the nature of the document.

The operating horizon runs about twelve months. Over that span, most of a small business's costs are already committed, the customer base is largely known, and the forecast is an extrapolation with adjustments. This horizon is mostly arithmetic.

The planning horizon runs roughly three years. This is the horizon a lender expects, the one at which capacity constraints and hiring decisions become visible, and the one at which almost every forecast becomes fiction if it is built by applying a growth percentage to the previous line. Most of the craft in this book concerns this middle horizon.

The strategic horizon runs five years and beyond. Here, numbers are illustrative rather than predictive, and their function is to show the shape of the business at scale — the ratios, the structure, the eventual size — not to be right.

DEFINITION Pro forma, assumption, forecast horizon A pro forma statement is a financial statement built from assumptions rather than from recorded transactions; the term simply means "as a matter of form," and it signals that no accountant is attesting to these figures. An assumption is an input you have chosen that could turn out to be wrong, stated as a value with a source and a confidence level. The forecast horizon is the period the projections cover, stated in months rather than years for the first year, because the first twelve months are where cash actually runs out.

1.1.2 Written for whom, exactly

A useful discipline before writing anything: name the specific human being who will read this. Not "investors" — a named lender at a named institution, or your future self in eighteen months, or the operations manager you are about to hire. Documents written for an abstraction read like documents written for an abstraction, and the reader can tell.

The rest of §1.3 develops this, because the identity of the reader determines the order of the sections, the length of the document and the amount of evidence each claim must carry.

1.2 The Three Jobs a Plan Does

A business plan can do three jobs. Most plans attempt all three at once, and the ones that fail usually fail because the three jobs pull the document in incompatible directions without anyone noticing.

1.2.1 Job one: to decide

The first job is internal and it precedes everything else. Writing forces a set of decisions that conversation allows you to postpone indefinitely: what exactly are we selling, to whom, at what price, delivered how, at what cost, with what money, by when.

This job is done the moment the decisions are written down and dated. It does not require a formatted document, a cover page or a table of contents. A business that writes a plan only for this job can legitimately produce eight pages and stop.

The mechanism at work here is worth naming, because it is the reason writing beats thinking. Prose does not tolerate the gaps that thought tolerates. You can hold "we'll sell to contractors and also to property managers" in your head indefinitely; the moment you write the market sizing for both, you discover you have two businesses with two sales cycles and two price points, and you must choose or resource both.

1.2.2 Job two: to finance

The second job is external and transactional. Somebody is being asked to advance money — a bank, a credit union, a public program, a family member, an investor — and the plan is the document on which that decision is partly based.

This job imposes requirements the first job does not. The numbers must tie together across three statements. The claims must carry sources. The section order must match what the reader expects, because a reader looking for the financing request will not hunt for it. And the document must survive being read out of order by somebody who has forty minutes.

QUÉBEC & CANADA NOTE Where a plan is actually required, and by whom Several Canadian programs make the plan a formal requirement rather than a courtesy. Futurpreneur Canada requires three uploads before an application is reviewed: a completed business plan, financial forecasts covering two years with a month-by-month breakdown, and government identification; its startup program serves entrepreneurs aged 18 to 39, lends up to $25,000 itself with up to $50,000 more from a co-lending partner for a maximum of $75,000, and repays over five years with interest-only payments in year one (Futurpreneur Canada, 2026). The Canada Small Business Financing Program, administered by Innovation, Science and Economic Development Canada, does not itself prescribe a plan, but it is delivered through chartered lenders who almost always do; the program's own parameters are a term loan of up to $1,000,000, of which at most $500,000 may fund assets other than real property and at most $150,000 of that may fund intangible assets and working capital, plus a line of credit of up to $150,000, for businesses whose estimated gross annual revenues do not exceed $10 million (Innovation, Science and Economic Development Canada, 2026a). Investissement Québec offers loans, loan guarantees, equity investment and delivery of government programs; its published product pages do not state minimum amounts, and the parameters change, so verify before quoting any figure to a reader (Investissement Québec, 2026). → see Book A5 (Financing a Business) for the comparison of these sources, and Book G7 (Grants, Subsidies and Public Programs) for the public programs themselves.

1.2.3 Job three: to execute

The third job is the one most books stop short of, and it is the reason chapters 10 and 11 of this one exist. A plan that is never converted into objectives, projects, tasks and a review cadence has done two-thirds of its work and abandoned the rest.

The conversion is mechanical, and it is described in full in chapter 10. In outline, each strategic intention becomes at most three objectives; each objective becomes projects; each project becomes tasks that carry an owner, a date and an observable criterion for being finished. The plan then becomes the thing the quarterly review is held against.

The test for whether a plan has done this job is simple and unforgiving: can you name, without opening the document, the three things this quarter is for? If not, the plan is decoration.

WATCH OUT The plan written only to obtain money is the one abandoned first There is a strong empirical pattern here that anyone who has worked with small businesses will recognize. A plan written exclusively for job two — to unlock a financing decision — has no natural readership after the decision is made. The loan is approved or refused; either way the document has completed its purpose and goes into a folder. Six months later nobody can find the assumptions behind the sales forecast, so nobody can tell whether the business is behind or merely different. The cheapest insurance against this is to write for job one first and job two second, in that order, even when the deadline is a lender's.

1.2.4 Choosing the primary job

The three jobs are not equally weighted in any given document, and the weighting should be a decision rather than an accident.

METHOD Choosing your plan's primary job, in four questions Answer these four questions in writing before you draft a section. Each takes about ten minutes and the answers belong at the top of your working file, not in the plan itself.

1. Is money being requested from somebody outside the business in the next ninety days? If yes, job two is present and imposes its format requirements. If no, everything about the document is negotiable.
2. How many people must act differently because of this document? If the answer is more than two, job three is primary, and the plan must end in a task list rather than in a conclusion.
3. Which decision is currently blocked? Name it in one sentence. A plan that unblocks no decision is being written for reassurance, which is a legitimate but different purpose.
4. What is the earliest date at which this document will be measurably wrong? That date sets your refresh cycle, and it is usually much sooner than a year.

The deliverable is four written answers. The verification criterion is that somebody else in the business, reading the four answers, can predict the length and shape of the document you are about to write.

The following table presents the three jobs side by side, with what each one demands of the document and the signal that tells you the job is done. Read it as a diagnostic: if your document is failing, it is usually because it is being judged against a job it was not built for.

Table 1.1 — The three jobs of a plan: what each one demands, and how you know it is done

1.3 The Five Readers, and What Each One Looks For First

A plan is read by at most five kinds of people, and each of them opens it at a different page. Knowing the order in which each reads is worth more than any amount of polish.

1.3.1 You and your co-owners

You read the assumptions and the cash flow, in that order, because those are the only two sections where you can be surprised by your own document. Everything else you already know.

What makes you stop reading is a number you cannot trace. If the forecast says $612,000 in year two and you cannot reconstruct where it came from, the document has failed its first reader.

1.3.2 The lender

A commercial account manager reads to answer one question: will this business generate enough cash, reliably enough, to service the debt, and what happens if it does not? That question drives an unusually consistent reading order — the financing request first, then the cash flow, then the security and the guarantees, then the operator's track record, and only then the market section.

The lender is not evaluating your idea. The lender is evaluating a repayment stream and its downside. This distinction explains why plans full of vision and thin on receivables tend to fail with lenders and can still succeed with investors.

1.3.3 The investor

An equity investor reads to answer a different question: how large can this become, and what has to be true for that to happen? The reading order inverts — market and opportunity first, then the team, then the model and its unit economics, then the ask and the dilution, and last the detailed financials, which are read as an indicator of the operator's judgment rather than as a prediction (Sahlman, 1997).

Most small businesses in this book's audience will never meet the second reader, and that is not a shortcoming. → see Book A5 (Financing a Business) for who should and should not seek equity.

1.3.4 The team

Employees, contractors and the operations lead read for one thing: what does this mean for what I do on Monday? They will not read forty pages. They will read the milestones table, the operating model and anything with their name on it.

A plan that cannot be reduced to a one-page internal version has not finished being written.

1.3.5 The advisor, the partner and the public program officer

The accountant reads the assumptions and the balance sheet. The lawyer reads the ownership and the contracts. The program officer reads for eligibility first, and everything else second: is the applicant in the right region, the right sector, the right size, at the right stage, with the right documents attached.

Figure 1.1 brings the two structures of this chapter together in one view. It is laid out as three columns, one per job: decide, finance and execute. The header of each column states what that job demands of the document. Beneath each header sit the readers that job serves — the owner under decide, the lender, the investor and the advisor under finance, the team under execute — and each reader's card names the section they open first and the thing that makes them stop reading. Read down a column to see what a document optimized for that job must contain; read across to see why no two readers enter the plan at the same page.

Figure 1.1 — Original figure. The three jobs of a plan and the readers each one serves: what every reader opens first, and what makes each of them stop reading.

The table below presents the same five readers in more detail. Its columns give, for each reader, the question they are actually answering, the section they open first, and the specific thing that makes them stop reading and set the document aside. That last column is the useful one: it is a list of the failures worth spending your editing time on.

Table 1.2 — The five readers: the question each answers, where each enters, and what stops them

WHO DOES THIS Who writes the plan, in three situations Solo. You write all of it, and the honest constraint is that you will be tempted to skip the sections you find least comfortable, which are usually the financial ones. The countermeasure is to write the financials first, while you are fresh, rather than last. Small team. One person owns the document and each section has a named contributor, but the owner writes the final text so that the voice is single. A plan written by four voices reads as four plans stapled together, and readers notice. Established business. The controller or the CFO owns the financial sections, the operations lead owns the operating model, and the owner owns the summary and the strategy. The risk here is the opposite of the solo risk: a plan assembled from departmental submissions that nobody has reconciled. The reconciliation pass, described in chapter 2, is not optional.

1.4 The Landscape: What the Evidence Says About Planning

There are two folk beliefs about business plans, and both are wrong in the same interesting way.

The first belief says that writing a plan is what separates serious businesses from hobbies, and that failing to plan is planning to fail. The second says that plans are theatre for bankers, that no plan survives first contact with a customer, and that the time is better spent selling. Both are stated with great confidence by people who have run businesses, which should tell you that both describe something real.

1.4.1 What the research actually found

The most useful single source here is a meta-analysis: Brinckmann, Grichnik and Kapsa (2010) pooled the results of 46 empirical studies of the relationship between business planning and performance in small firms, published in the Journal of Business Venturing. Two findings matter for a practitioner.

First, the overall association between planning and performance is positive. Planning is, on average, associated with better outcomes rather than worse ones, which disposes of the strong version of the second folk belief.

Second — and this is the finding that makes the paper worth citing rather than the first — the association is strongly moderated by context. It is larger for established firms than for brand-new ventures, and it varies with the cultural and institutional environment. The authors' framing is that planning helps most where there is something to plan against: existing operations, existing customers, known cost structures. For a venture whose fundamental model is still unknown, the same effort spent on structured experimentation may return more than the same effort spent on projection.

The practical translation is not "plan less." It is plan at the resolution your evidence supports. A business with three years of receipts can and should forecast in detail, because the drivers are measurable. A business with no customers should write a short plan built around the experiments that will produce the missing evidence, and should say so in the document rather than manufacture precision it does not have.

1.4.2 What the survival statistics say

The federal statistics are worth knowing precisely, because they are widely misquoted in both directions.

Innovation, Science and Economic Development Canada (2026c) publishes Key Small Business Statistics, whose 2025 edition (page modified April 23, 2026) reports survival rates for businesses with 1 to 99 employees. After one year, 94.8% of them are still operating. After three years, 79.9%. After five years, 68.0%. After ten years, 48.2%. The same source reports a meaningful divergence by sector after the third year: at the ten-year mark, 52.0% of goods-producing businesses remain, against 47.2% of services-producing businesses.

Two things follow. The popular claim that most new businesses fail in the first year is simply false in Canada: nineteen in twenty survive it. And the mortality is concentrated in years three through ten, which is precisely the window in which an original set of assumptions quietly stops being true and nobody notices because nobody wrote them down.

For scale, the same publication reports 1,079,188 small businesses in Canada as of December 2024, representing 98.2% of all employer businesses, and employing 5.8 million people, or 46.6% of the private-sector labor force in 2024.

WHAT CHANGED The survival data now runs to a more recent reference year The 2025 edition of Key Small Business Statistics was published with a page modification date of April 23, 2026 and carries data through the 2022 reference year for survival, with business counts as of December 2024. Anyone quoting survival figures from an older edition is quoting a different cohort, and the differences between editions are large enough to change an argument. Re-verify at the source before putting a survival figure in a plan; the citation should name the edition, not just the agency. Read September 9, 2026.

1.4.3 What failure analyses say about causes

Cause-of-failure data is softer than survival data, because it relies on self-report and on the analyst's classification. The most widely cited recent series comes from CB Insights, which examined 431 venture-backed companies that shut down from 2023 onward (CB Insights, 2026). In that analysis, running out of capital appears in about 70% of cases, poor product-market fit in about 43%, bad timing in about 29%, and unsustainable unit economics in about 19%; the percentages exceed one hundred because most failures cite several causes.

The useful reading of that data is not the ranking. It is that running out of capital is nearly always the terminal event rather than the cause — the thing that finally happens when a business has been serving the wrong customer, or serving the right one at a loss, for long enough. That distinction is exactly what a plan with an honest assumption log makes visible early, and exactly what a plan written to persuade conceals.

Note also the population: venture-backed companies are not the population most readers of this book belong to, and their failure modes skew toward growth-stage problems. The figure is cited here for its structure, not as a base rate for a Québec service business.

DO THE MATH The cost of an unwritten plan: the decision made twice Assume a business with two owners and one operations lead, whose loaded hourly cost averages $62 per hour. A single unresolved decision — say, whether to serve two customer segments or one — surfaces in conversation roughly twice a month, consuming about 35 minutes each time across the three people. Time per occurrence: 35 minutes × 3 people = 105 minutes = 1.75 hours. Cost per occurrence: 1.75 × $62 = $108.50. Occurrences per year: 2 per month × 12 = 24. Annual cost of not deciding: 24 × $108.50 = $2,604. Now the comparison. Writing the section of the plan that settles the question — market sizing for both segments, capacity implications, a decision and a date — takes about six hours of the owner's time at $62, or $372, plus perhaps $400 of an advisor's time to check the arithmetic. Payback: $772 ÷ $2,604 per year = 0.30 years, or about eleven weeks. The point of this calculation is not the precision, which is illusory; the inputs are yours to change. The point is the structure. The recurring cost of an open decision is invisible because it is paid in fragments, and the one-time cost of closing it is visible because it is paid in a block. That asymmetry is why unwritten plans feel cheaper than they are.

1.5 The Challenges That Make Plans Fail Before They Are Read

Six failure modes account for most of the disappointment in this domain. All six are avoidable, and none of them is about writing skill.

1.5.1 The unsourced market claim

A sentence such as "the market is growing at 12% annually" with no citation does more damage than no sentence at all, because it tells the reader that the document contains at least one unchecked number and gives them no way to know which others. Chapter 4 gives the rule and the fix.

1.5.2 The forecast built from a growth percentage

A revenue line that grows 30% a year because the previous line did is not a forecast; it is a shape. Real forecasts are built from drivers — customers, price, frequency, capacity — each of which can be argued about separately. Chapter 7 builds one from scratch.

1.5.3 The sales forecast that exceeds the company's own capacity

This is the most common internal contradiction in small-business plans and the easiest for a competent reader to spot. The revenue line implies 3,800 billable hours in year two; the operations section describes two technicians. Chapter 6 does the capacity arithmetic that prevents it.

1.5.4 The financing request that is a round number

An ask of "$150,000" that does not fall out of a cash flow trough tells the reader that the number was chosen and the plan was fitted to it. Chapter 8 derives the request from the cash curve, which is the only defensible direction.

1.5.5 The risk section that says "competition"

A risk section listing three generic risks with no probability, no dollar impact, no owner and no trigger is worse than none, because it signals that the exercise was performed rather than done. Chapter 9 gives a register that a reader will believe.

1.5.6 The plan with no next Monday

The document ends with a conclusion instead of a task list, and nothing changes. Chapters 10 and 11 exist for this failure specifically.

CASE IN POINT Nordec Entretien: the plan written twice This case is fictional; the company, the people and the figures are invented, and they recur throughout this book so that one business can be followed from the first assumption to the signed loan. Nordec Entretien inc. provides preventive maintenance and on-call technical service to commercial buildings on Montréal's north shore: light HVAC, minor plumbing, service electrical, loading doors. Sophie Nadeau founded it in 2022. Four people work there — Sophie, who sells, dispatches and still spends about a third of her time in the field; two full-time technicians; and one administrator at twenty-four hours a week. The last closed year brought in a little over $600,000, split between thirty-four buildings on monthly preventive contracts and on-call interventions billed by the hour. Sophie wrote her first plan in eleven days in early 2026, because a lender asked for one. It was twenty-eight pages, it looked professional, and it was declined. The letter gave no reason. When she later asked, the account manager told her two things: the requested amount did not follow from anything in the document, and the revenue forecast for year two implied roughly twice the billable hours her own operations section said she had. She rewrote it over five weeks, in the order this book teaches — assumptions first, capacity second, cash third, and the ask derived from the cash curve last. The second version was nineteen pages and asked for less money than the first. It was approved. The part of the story that matters is not the approval; it is that the nineteen-page version is the document she now opens at the start of every quarter, and the twenty-eight-page version is one she never opened again.

1.6 What a Business Actually Needs From Its Plan

Set aside for a moment what lenders want and what books say. What does the business itself need from this document over the next three years?

1.6.1 A record of what you believed, and why

This is the need that outlives every other. In eighteen months you will want to know what you assumed about your average contract value, and whether you were wrong by a little or by a lot. Only a written, dated assumption log can tell you.

THE RULE Every figure in a plan traces to an assumption you can name This is the one rule that, applied strictly, prevents most of the failure modes in §1.5. Every number that appears anywhere in the document must be reachable by a chain of arithmetic from an assumption that is stated somewhere, with its source and its confidence. If a number cannot be traced, it must be removed or an assumption must be written for it. There is no third option, and the discipline is not aesthetic: an untraceable number cannot be updated when the world changes, which means the plan silently rots around it. This rule is the reason chapter 7 builds an assumption log before it builds a statement.

1.6.2 A shared operating picture

Where more than one person makes decisions, they need to be making them against the same picture. The plan is the cheapest way to produce one, and its most underrated function is negative: it stops two competent people from optimizing for different objectives for six months without noticing.

1.6.3 A basis for saying no

A business without a written plan says yes to too many things, because every individual opportunity looks good in isolation. A plan makes the opportunity cost visible — this project, at this moment, displaces that one — which is what converts "no" from a personality trait into a decision with a reason.

1.6.4 A credible answer to a stranger

Occasionally a stranger with money, or a large customer, or a potential partner asks what your business is and where it is going. Having a current answer in written form is worth having.

1.6.5 A structure for the review

Finally, the business needs something to hold the quarterly review against. A review with no plan behind it becomes a status meeting, which is a different and much less useful thing.

TOOLS OF THE TRADE What a plan-writing environment must provide Before naming any product — chapter 13 does that, with dated vendor pages — it is worth stating the six functions the work requires, because the functions are stable while the products change. A plan-writing environment must let you write and revise long-form text with version history, so that you can tell what changed between drafts and when. It must let you build a financial model whose formulas are visible and auditable, since a model whose logic is hidden cannot be checked by an accountant. It must let you collect and store evidence with its source and date, because a citation retrieved eight months later is worth nothing without the date it was read. It must let you produce a short presentation version, since many readers now see the deck before the document. It must let you turn decisions into owned, dated tasks, without which chapter 10 is impossible. And it must let you report progress against the plan on a schedule, which is the difference between a plan and a memory.

1.7 The Constraints: Time, Money, Information and Skill

Every real plan is written under four constraints. Naming them honestly at the outset prevents the most common form of self-inflicted failure, which is starting a document that the available time cannot finish.

1.7.1 Time

The realistic figure for a first complete plan, written by an owner who knows their business and has their numbers in reasonable order, is thirty to fifty hours of focused work. Chapter 14 lays this out as ten working sessions across thirty calendar days.

The constraint is rarely the total; it is the fragmentation. Thirty hours in three-hour blocks produces a plan. Thirty hours in twenty-minute fragments between customer calls produces a partially edited draft, because the financial model in particular requires enough continuous time to hold the whole structure in view.

1.7.2 Money

A plan written entirely in-house costs the owner's time. A plan reviewed by an accountant typically adds a professional engagement whose scope should be defined in writing — chapter 12 describes what to ask for and what each professional actually delivers. A plan written by a consultant costs substantially more and, as chapter 12 argues, is usually a poor trade for a small business, for a reason that has nothing to do with money: a plan you did not write is a plan you cannot defend in a meeting.

Following the rule set for this book, no fee ranges are printed here. Fee structures vary by market and by year, and a figure printed without a dated source is exactly the kind of number this chapter has been arguing against. Chapter 12 explains how to obtain a current quote in writing before you engage anyone.

1.7.3 Information

You will not have all the information you want. Market data for narrow sectors is often unavailable at the geography you care about; competitor revenue is private; conversion rates for a channel you have never used cannot be known.

The professional response is not to invent it and not to omit the section. It is to state the gap, state the assumption you are using in its place, and state how and when you will replace the assumption with a measurement. A plan that says "we assume a 3% response rate to direct mail, based on nothing better than the industry range published by X in 2025; we will have our own figure after the March campaign" is more credible than one asserting 3% flatly, and it is the sentence that makes the March campaign happen.

1.7.4 Skill

Six skills are involved: writing clearly, reading a financial statement, building a model, gathering evidence, estimating operations, and negotiating. Almost nobody has all six at a professional level.

The two that most often need outside help are the financial model and the reading of a lender's terms. Chapter 12 covers who to call and, more usefully, what to ask them for.

The table below presents the four constraints with the honest question to ask about each, the most common self-deception associated with it, and the countermeasure that works. It is meant to be filled in for your own situation before you begin, in writing, in about twenty minutes.

Table 1.3 — The four constraints: the honest question, the common self-deception, the countermeasure

1.8 How This Book Is Organized

The fourteen chapters follow the order in which the work is actually done, which is not the order in which the finished document is read.

Chapters 2 and 3 build the container. Chapter 2 gives the full anatomy — the nine sections, what each must answer, how long each should be, and the crucial distinction between the order of writing and the order of reading. Chapter 3 writes the two sections readers see first and writers should write last: the executive summary and the company section.

Chapters 4 through 6 build the argument. Chapter 4 is the market analysis, including how to size a market from the bottom up without inventing anything. Chapter 5 covers the offer and go-to-market sections, with customer acquisition cost and payback worked in full. Chapter 6 covers operations, team and governance, and contains the capacity arithmetic that keeps a forecast honest.

Chapters 7 and 8 build the numbers. Chapter 7 is the longest chapter in the book and constructs assumptions, sales forecast, income statement, cash flow, opening balance sheet, break-even and sensitivity on a single continuous case. Chapter 8 derives the financing request from the cash curve and explains what each source of money wants in return.

Chapter 9 stress-tests it. The risk register, the quantification of a risk in dollars, the four responses, scenarios and contingency triggers.

Chapters 10 and 11 make it live. Chapter 10 converts the plan into objectives, projects and tasks carrying verification criteria, and installs the quarterly review. Chapter 11 gives the task table in the format La Machine uses, the writing sprint, and the annual refresh.

Chapters 12 through 14 close it out. Chapter 12 covers expertise and when to get help. Chapter 13 covers the six tool functions and the popular tools within each, with dated pricing models. Chapter 14 gives the fifteen pieces of advice, the costly mistakes and the thirty-day plan.

IN LA MACHINE Where a plan becomes work La Machine does not write the plan and does not model the financials; §In La Machine below says exactly what it does and does not do. What it does hold is the third job. A plan's objectives become projects at /projets, each one either a chain or a one-off; a project's work becomes tasks at /taches, each carrying a verification criterion, required at creation and not checked by the Mark as done button; decisions that could not be settled during the review go to arbitrages; and the numbers you chose to watch appear at /tableau-de-bord. Chapter 10 does this conversion in detail.

Costly Mistakes

1. Writing the plan for a lender before writing it for yourself. The document acquires a persuasive voice early, and the persuasive voice is the one that phrases around weak spots instead of recording them. The cost is not the refusal; it is that you no longer know which of your own numbers you actually believe.
2. Treating the plan as a document rather than a decision record. Every hour spent on formatting before the assumptions are settled is an hour spent making a possibly wrong document look right. Settle the assumptions, then format once.
3. Quoting a survival or market statistic from memory. The federal survival figures are widely misquoted in both directions, and a reader who knows the real number will discount everything else in your document. Cite the edition and the date read, or leave the statistic out.
4. Choosing the horizon by convention rather than by purpose. A three-year plan is not a rule of nature; it is what most lenders ask for. If your decision is about the next nine months, a nine-month plan with a monthly cash view is a better document and takes a third of the time.
5. Starting the document without booking the time. The single strongest predictor of an unfinished plan is that no calendar blocks were reserved for it. Thirty hours of intention is not thirty hours.
6. Believing that a plan is a prediction. It is a set of dated assumptions with arithmetic attached. Being wrong is expected; being unable to tell where you were wrong is the actual failure, and it is caused by not writing the assumptions down.

Key Takeaways

• A business plan is a dated, written document that states intentions, assumptions, costs, expected results and the means of verification — and its three defining properties are that it is dated, written and explicit about assumptions.
• The plan is not the model, not the strategy, not the marketing plan and not the budget; it is the document that commits those to a horizon, a budget and a calendar.
• A plan does three jobs — decide, finance and execute — and the job you choose as primary determines the length, the order and the amount of evidence each claim must carry.
• Five readers open a plan at five different pages, and knowing which section each opens first is worth more than any amount of editing.
• The evidence supports planning on average and shows the benefit is strongly context-dependent, which means you should plan at the resolution your evidence supports rather than manufacturing precision.
• Canadian survival rates are 94.8% at one year, 79.9% at three, 68.0% at five and 48.2% at ten, so the dangerous window is years three through ten rather than the first year.
• Every figure in a plan must trace to a named assumption with a source; a number that cannot be traced cannot be updated, and the plan rots around it.
• Four constraints — time, money, information and skill — shape every real plan, and naming them honestly before you start prevents the most common self-inflicted failure.

Your Action Plan

1. Answer the four questions of §1.2.4 in writing. Deliverable: four written answers at the top of your working file. Timing: forty minutes. Verification criterion: a colleague reading only your four answers can predict the length and shape of the document you are about to write.
2. Name your primary reader, as a person rather than a category. Deliverable: one sentence naming the individual and the decision they must make. Timing: ten minutes. Verification criterion: the sentence contains a name or a specific role at a specific institution, not the word "investors."
3. Book the writing blocks in your calendar before drafting anything. Deliverable: ten sessions of two to three hours, placed on real dates within the next thirty days. Timing: fifteen minutes. Verification criterion: each block is in the calendar with a title naming the deliverable it produces.
4. Start the assumption log today, on its own page. Deliverable: a table with five columns — assumption, value, source, confidence, what changes if it is wrong — carrying at least the five assumptions you already know you are making. Timing: thirty minutes. Verification criterion: every row has something in the source column, including the word "estimate" where that is the truth.
5. Write down the three numbers you do not have. Deliverable: a short list naming each missing figure, what it would take to obtain it, and the date by which you will replace the placeholder. Timing: twenty minutes. Verification criterion: each line has a date, and each date is inside the next ninety days.

In La Machine

Every route named in this section was verified in the product's route inventory on 2026-09-09 (La Machine, 2026a), and the statuses and natures it names come from the product's user guide (La Machine, 2026b).

• The plan's objectives become projects. /projets (and productivity/projects in the Core menu) holds projects, each of which is either a chain — a sequence meant to continue — or a one-off that closes when the need is met. A chain carries a short code; a one-off carries an identifier prefixed with P-.
• The work becomes tasks. /taches and productivity/tasks hold tasks, each attached to a single project, each carrying an owner, a due date and a verification criterion.
• Decisions that could not be settled go on the record. arbitrages holds open decisions rather than leaving them in a thread.
• The numbers you chose to watch have a screen. /tableau-de-bord consolidates the operating figures; /aujourdhui gathers what needs attention on a given day.
• The document itself can live in the library. business/assets/books and business/assets/presentations hold long-form assets and decks for a business that distributes them.
• What it does not do. La Machine does not write the plan and holds no business-plan template. It does not model financial projections, does not consolidate financial statements, and does not produce an income statement, a cash flow forecast or a balance sheet. Those are built in the tools of chapter 13 and reviewed by the professionals of chapter 12; La Machine holds what happens after the plan is written.

Self-Check

1. In one sentence each, distinguish a business plan from a business model and from a budget. Name the property that only the plan has.
2. Your primary reader is a commercial account manager at a credit union. Name the first three sections they will read, in order, and say why that order follows from the question they are answering.
3. The meta-analysis cited in §1.4.1 found a positive association between planning and performance that is strongly moderated by context. Give the practical instruction that follows for a business with no customers yet, and for one with three years of receipts.
4. A plan states that revenue will reach $840,000 in year three. Apply the rule of §1.6.1 and describe, in two steps, what must exist elsewhere in the document for that figure to be legitimate.
5. Name the four constraints of §1.7 and, for your own situation, say which one is currently binding. What is the countermeasure for that one?

Chapter References

Brinckmann, J., Grichnik, D., & Kapsa, D. (2010). Should entrepreneurs plan or just storm the castle? A meta-analysis on contextual factors impacting the business planning–performance relationship in small firms. Journal of Business Venturing, 25(1), 24–40.

CB Insights. (2026). Why startups fail: Top reasons (analysis of 431 venture-backed company shutdowns since 2023). Retrieved September 9, 2026.

Futurpreneur Canada. (2026). Frequently asked questions (startup program: eligibility 18–39; up to $25,000 from Futurpreneur with up to $50,000 in co-lending; five-year term with interest-only payments in year one; business plan and two-year month-by-month financial forecasts required for review). Retrieved September 9, 2026.

Innovation, Science and Economic Development Canada. (2026a). Canada Small Business Financing Program — Program guidelines (term loans up to $1,000,000, of which at most $500,000 for assets other than real property and at most $150,000 of that for intangible assets and working capital; lines of credit up to $150,000; eligibility limited to businesses with estimated gross annual revenues of $10 million or less; 2% registration fee; floating rate capped at lender's prime plus 3%). Retrieved September 9, 2026.

Innovation, Science and Economic Development Canada. (2026c). Key small business statistics 2025 (page modified April 23, 2026; survival rates for businesses with 1–99 employees at one, three, five and ten years; 1,079,188 small businesses as of December 2024; 5.8 million employees, 46.6% of the private-sector labour force in 2024). Retrieved September 9, 2026.

Investissement Québec. (2026). Financial products: loans, loan guarantees, equity investment and government programs. Retrieved September 9, 2026.

La Machine. (2026a). Documentation — route reference inventory (211 routes: 191 Core, 20 Pilotage paths). Internal product documentation, read September 9, 2026.

La Machine. (2026b). Guide utilisateur [User guide] (FAQ: projects and task chains; pilotage, tasks and their statuses; dashboard and reports). Internal product documentation, read September 9, 2026.

Sahlman, W. A. (1997). How to write a great business plan. Harvard Business Review, 75(4), 98–108.

Chapter 2 — Anatomy of a Business Plan

Ask five people what belongs in a business plan and you will get five lists, all of them roughly right and none of them the same. This is not a failure of the profession; it is a consequence of the fact that a plan serves three different jobs and five different readers, so its shape is genuinely negotiable. What is not negotiable is that every section must answer a question, that the questions must be answered in an order a reader can follow, and that nothing may appear in the document that does not earn its page. This chapter gives the anatomy that satisfies all five readers, then shows you how to build it in the order that actually works — which is almost exactly backwards from the order it is read.

What You'll Be Able to Do

• Recite the nine sections of a business plan and the single question each one must answer.
• Produce the front matter — cover page, table of contents and confidentiality notice — that a professional reader expects.
• Choose the length of each section deliberately, using a page budget rather than however much you happened to write.
• Decide what goes in the body and what goes in an appendix, using one test rather than instinct.
• Write in the order of construction — financials before summary, always — instead of the order of reading.
• Version and date the document so that a reader six months from now can tell which assumptions were current when.

Why It Matters

The anatomy is what makes a plan readable by someone in a hurry, and every reader worth having is in a hurry. A lender who has to search for the financing request will find it, once. A program officer who cannot locate the eligibility facts on page two will set the file aside and pick up the next one. None of this is about elegance. It is about the reader's ability to answer their own question without contacting you, which is the single best predictor of whether a document produces a decision.

There is a second reason, and it is the one that saves the most time. A plan built section by section, in an order where each section consumes the output of the previous one, is roughly half the work of a plan written front to back. Written front to back, the executive summary is drafted before the numbers exist, then rewritten when the numbers arrive, then rewritten again when the operations section changes the numbers. Written in construction order, it is drafted once, at the end, from finished inputs. That difference — one draft instead of three — is most of the thirty hours this book budgets for the whole exercise.

In This Chapter

• 2.1 The Nine Sections, and Why That Order
• 2.2 The Cover, the Table of Contents and the Confidentiality Notice
• 2.3 What Each Section Must Answer
• 2.4 Length, Format and the Two Documents You Actually Need
• 2.5 The Appendices, and What Belongs There Instead of the Body
• 2.6 The Order of Writing Is Not the Order of Reading
• 2.7 Versioning, Dating and the Assumption Log
• 2.8 Adapting the Anatomy: Startup, Existing Business, Acquisition, Non-Profit
• 2.9 The Anatomy in Three Situations

2.1 The Nine Sections, and Why That Order

A complete business plan has nine sections in the body, plus front matter and appendices. The list is stable across lenders, public programs and the major template providers, with minor differences of naming that do not matter (Business Development Bank of Canada, 2026a).

1. Executive summary — the whole plan, compressed to one or two pages.
2. The company — legal identity, what the business does, mission, history, ownership.
3. Market analysis — the industry, the trends, the segments, the competition, the regulation.
4. The offer — the products or services, what they include, how they are priced.
5. Go-to-market — how the offer reaches the customer: marketing, sales process, channels.
6. Operations — how the promise is delivered: process, capacity, suppliers, location, equipment.
7. Team and governance — who does the work, who decides, who advises.
8. Financials — assumptions, forecast, income statement, cash flow, balance sheet, break-even.
9. Risks and mitigation — what could go wrong, how likely, how bad, and what you would do.

2.1.1 The logic of the sequence

The order is not arbitrary, and understanding its logic is what lets you adapt it safely.

The sequence follows a chain of dependency that a reader can follow without back-tracking. Sections 3 through 5 establish that demand exists and can be reached. Sections 6 and 7 establish that the promise can be delivered by this particular company. Section 8 turns all of it into money over time. Section 9 asks what if this is wrong. And section 2 supplies the identity that makes the rest attributable to a real legal entity.

Move a section and you break a dependency. A financials section placed before operations asks the reader to accept revenue figures before knowing whether the company can produce the work. A risk section placed early reads as anxiety rather than as rigour, because the reader does not yet know enough to judge whether the risks named are the important ones.

DEFINITION Executive summary, appendix, assumption log The executive summary is a compressed version of the entire plan, written last and placed first, containing every element of the argument in miniature — including the numbers and the ask. It is not an introduction, a preface or a teaser, and the difference is testable: a reader who reads only the summary should be able to make a preliminary decision. An appendix is supporting material referenced from the body but not required to follow the argument: résumés, quotes, detailed schedules, letters of intent, technical specifications. The assumption log is a working document, usually not part of the delivered plan, listing every input to the financial model with its value, its source, your confidence in it, and what would change if it were wrong.

2.1.2 The one section people invent, and should not

Business owners frequently want a tenth section for something the anatomy seems not to cover: the technology, the sustainability commitments, the intellectual property, the exit strategy, the community impact. Almost always, the material belongs inside an existing section rather than beside it.

Technology that delivers the service belongs in operations. Technology that is the product belongs in the offer. Intellectual property belongs in the offer if it is what customers buy, and in risks if its protection is uncertain. Sustainability belongs in operations if it changes how work is done, and in market analysis if it changes what customers choose. An exit strategy belongs in the financing request section within financials, and only when there is an equity investor to exit.

The discipline is worth keeping, because every invented section is a section the reader was not expecting, and unexpected sections get skipped.

KEY CONCEPT Every section answers one question; a section that answers none is an appendix This is the test that settles nearly every argument about what belongs where. Write the question at the top of each section in your working draft — literally, as a line of text you will delete before delivery. If you cannot state a question the section answers, and the answer matters to at least one of the five readers, the material is supporting evidence rather than argument, and it belongs in an appendix or in the recycling. Applied strictly, this test typically removes six to ten pages from a first draft, and none of them are missed.

2.2 The Cover, the Table of Contents and the Confidentiality Notice

The three pages before the body do more work than their length suggests, because they are where a reader forms an expectation about how careful the rest of the document will be.

2.2.1 The cover page

The cover carries seven items and nothing else: the legal name of the business, the words "Business Plan," the period the plan covers, the date of the version, the version number, the name and role of the person who wrote it, and one line of contact information. A logo is fine. A photograph, a slogan and a decorative border are not, for the same reason a résumé does not carry them.

The two items most often missing are the date and the version number, and their absence causes a specific, common failure: two versions of the plan circulate, a reader comments on the old one, and nobody can tell which is which.

THE RULE A dated plan is a document; an undated plan is an opinion Every version of a business plan carries, on its cover page, the date it was completed and a version identifier. Every financial statement inside it carries the date of the assumptions it was built on. Every external statistic carries the date it was retrieved. This is not administrative fussiness; it is what makes the document falsifiable. A reader who knows the plan was written on a given date can judge which facts were knowable then. A reader who does not know is entitled to assume the worst, and generally does.

2.2.2 The table of contents

A table of contents is required above roughly twelve pages and helpful below it. It should list the nine sections with page numbers, plus the appendices by name. It should not list every sub-heading, which turns a navigation aid into a second document.

If the plan will be read on a screen — and most now are — the table of contents should carry working internal links. A reader who can jump from the contents to the financing request in one click reads more of the plan than one who scrolls.

2.2.3 The confidentiality notice

A confidentiality notice is a short paragraph, usually on the inside of the cover, stating that the document contains confidential and proprietary information, that it is provided for the purpose of evaluating a specific transaction or relationship, and that it should not be reproduced or distributed without permission.

Its practical value is modest and worth understanding precisely. It does not, by itself, create a contractual obligation on someone who did not agree to it. What it does is establish that you treated the information as confidential, which matters if the question ever arises — and it signals to a professional reader that the document should be handled carefully, which is usually all you need.

QUÉBEC & CANADA NOTE What a confidentiality notice actually protects Under Québec law, the protection of commercially sensitive information rests less on a notice than on two other mechanisms. Article 2088 of the Civil Code of Québec binds an employee "to act faithfully and honestly and not use any confidential information he obtains in the performance or in the course of his work," and provides that these obligations "continue for a reasonable time after the contract terminates." Article 1612 addresses the remedy: the loss sustained by the owner of a trade secret "includes the investment expenses incurred for its acquisition, perfection and use," and the profit of which the owner is deprived "may be compensated for through payment of royalties." Both articles were read in the LégisQuébec consolidation updated to April 7, 2026 (read September 9, 2026), and a control request for a non-existent article returned nothing, confirming the retrieval. The practical instruction is therefore layered rather than single: put the notice on the document, use a signed non-disclosure agreement when you are sending detailed financials to a party who is not already bound, and do not put your genuinely irreplaceable technical detail in a document that travels. → see Book H1 (Contracts and Commercial Agreements) for the agreement itself and Book H5 (Intellectual Property) for what can and cannot be protected.

2.3 What Each Section Must Answer

Each of the nine sections exists to answer one question. The formulations below are the ones to write at the top of your draft.

Executive summary — "Should I keep reading, and what is being asked of me?" Everything else in the summary serves those two questions.

The company — "Who exactly is this, and does it legally exist as described?" Legal name, form, registration number, address, date of formation, ownership, and what it does in one sentence.

Market analysis — "Is there enough demand, and can this business reach it?" Not "is this a big market," which is almost always answerable with a yes and therefore proves nothing.

The offer — "What is being sold, and why would somebody choose it?" Including what it costs the customer and what they get for that.

Go-to-market — "How does the offer reach a buyer, at what cost, and how long does it take?" The section that most plans under-write and most lenders quietly weight heavily.

Operations — "Can this business actually deliver what it has promised to sell?" Capacity, process, dependencies, and the physical or digital means of production.

Team and governance — "Are these the people who can do it, and who decides what?" Roles before names, and an honest account of the gaps.

Financials — "What does this produce in money, over time, and what has to be true?" Three statements, one break-even, one sensitivity, and the assumptions behind all of it.

Risks — "What would make this fail, how likely is it, and what would you do?" Specific, owned, triggered.

2.3.1 The first sentence of each section

There is a small craft habit that improves a plan more than any amount of editing, and it costs nothing: make the first sentence of each section answer its question directly, then spend the rest of the section supporting that answer.

Compare two openings for a market section. The first says: "The building maintenance industry in Québec has undergone significant changes over the past decade." That sentence answers nothing; it is a throat-clearing noise, and a reader who has read forty plans recognizes it as such. The second says: "There are roughly 1,400 commercial buildings in our service area that use an external maintenance contractor, and we currently serve 34 of them." That sentence answers the question, establishes that the writer has done arithmetic, and gives the reader a reason to read the paragraph that follows.

The same habit applied to each section produces nine sentences that, read together, are very nearly the executive summary. This is not a coincidence, and chapter 3 uses it: if the first sentences of your nine sections do not assemble into a coherent argument, the plan has a structural problem that no amount of polish will fix.

2.3.2 The characteristic failure of each section

Each section has a failure mode that recurs across almost every first plan. Naming them here means you can check for them deliberately rather than hoping a reader will be kind.

• The executive summary fails by being an introduction. It describes what the plan will cover instead of stating what the plan concludes, and a reader who finishes it knows the subject but not the argument.
• The company section fails by drifting into mission language. Three paragraphs of aspiration where a reader wanted a legal name, a form, a registration number and a date of formation.
• The market section fails by proving the wrong thing. It establishes that the market is large, which was never in doubt, and never establishes that this business can reach a specific part of it.
• The offer section fails by describing features to somebody who wants outcomes. The specification belongs in an appendix; the body should say what the customer gets and what it costs them.
• The go-to-market section fails by listing channels without arithmetic. Naming five channels is not a plan; saying what each costs, how many customers it produces and how long it takes is.
• The operations section fails by omitting capacity. It describes the process in detail and never states how much work the business can actually absorb, which is the one number the section exists to produce.
• The team section fails by hiding the gap. Every small business has a role nobody currently fills; a plan that pretends otherwise is less credible than one that names the gap and says how it will be filled.
• The financials section fails by not tying. The revenue in the cash flow does not match the revenue in the income statement, or the closing cash does not match the balance sheet, and one such break makes every other figure suspect.
• The risks section fails by being generic. "Competition," "economic conditions" and "key person risk" with no probability, no dollar figure, no owner and no trigger.

The paragraph that follows introduces Table 2.1, which puts all nine sections side by side. Its columns give the question each section answers, a realistic page budget for a plan in the twenty-five to thirty-page range, and the appendix that section typically points to. Use the page budget as a constraint rather than a target: a section that will not fit its budget is usually a section that has absorbed material belonging to another one.

Table 2.1 — The nine sections: the question each answers, its page budget and its appendix

2.4 Length, Format and the Two Documents You Actually Need

2.4.1 How long a plan should be

The honest answer is that length follows from the job and the reader, and the useful answer is a set of defaults.

For a small business seeking conventional debt financing, twenty to thirty pages of body plus appendices is the working range. Below fifteen pages, a reader begins to suspect that sections were skipped rather than compressed. Above forty, the marginal page is read by nobody, and the document's own length starts working against it, because a reader who cannot finish it in one sitting will finish it never.

For an internal planning document with no external reader, eight to fifteen pages is plenty, and the financial detail can live in the model rather than in prose.

For a public program with a prescribed form, the form governs. Fill it as specified and attach your fuller document as an appendix if the form permits.

WATCH OUT The forty-page plan nobody finishes There is a reliable inverse relationship between the length of a first-time plan and the probability that its financial section is internally consistent, and the mechanism is simple: pages spent on narrative are hours not spent reconciling statements. If your draft is running long, the sections to cut are almost never the financials. They are the industry background that any reader in that industry already knows, the history of the company beyond its material milestones, and the description of the offer at a level of detail that belongs in a specification sheet. Cutting fifteen narrative pages typically costs nothing and buys the eight hours the cash flow needs.

2.4.2 Format conventions that matter

A handful of formatting choices affect readability enough to be worth naming. Use a single serif or sans-serif family at eleven or twelve points with generous line spacing. Number the pages, including the appendices, in a continuous sequence. Number the tables and figures and refer to them by number in the text, since a reader who has jumped to page 19 needs to know which table you mean. Put every financial table in the same units, stated once at the top, and never mix thousands with units in the same statement. Deliver as a PDF, which prevents the reader's software from re-flowing your tables into nonsense.

2.4.3 The two documents you actually need

Almost every business that writes a plan ends up needing two artifacts, and knowing this in advance prevents a late scramble.

The full document is the twenty-to-thirty-page plan described above. It is the reference, the record and the thing an analyst reads.

The short version is one of two things depending on the audience: a one-page summary for internal use and for a first conversation, or a ten-to-fifteen-slide deck for an investor or a formal presentation. Both are derived from the executive summary rather than written independently, which is why chapter 3 treats them together.

A third artifact — the financial model itself, as a spreadsheet — is not part of the plan but is frequently requested alongside it. Build it so it can be sent: no hard-coded numbers buried in formulas, an assumptions tab at the front, and no sheets you would be embarrassed to have opened.

2.4.4 Writing so the document can be read aloud

A plan is increasingly read on a phone, forwarded to a colleague who skims it between meetings, or worked through by someone using a screen reader. All three cases reward the same discipline, which is worth adopting even if you expect none of them.

Every table is introduced by a sentence that says what it shows. Not "see the table below," but a sentence naming the columns and the point: "The table below gives the nine sections, the question each answers, and a realistic page budget." A reader who cannot see the table still gets the content, and a reader who can see it now knows what to look for.

Every figure is introduced and captioned. The introduction says what the figure shows; the caption says what to take away. A figure whose only explanation is inside the image is invisible to a third of the ways your document will be consumed.

Bullet points are complete sentences. A list of fragments — "Strong margins," "Experienced team," "Growing market" — is a list of assertions with no verbs, and the reader supplies the missing reasoning, usually less generously than you would have. Write "Gross margin has held above 41% for eight consecutive quarters," and the same line now carries evidence.

Acronyms are expanded on first use, in every section. A reader who enters at the financing request has not read your definition on page four.

WHAT CHANGED The deck has not replaced the document It is now common for a first contact to happen through a short deck rather than a document, and this has led some owners to conclude that the full plan is obsolete. The evidence from lending practice says otherwise, and the reason is procedural rather than aesthetic: a credit decision has to be documented in a file, and a deck does not contain enough to document it. The deck now often determines whether the document gets read, which changes its importance but not its function. Prepare both, derive the second from the first, and never let the two diverge — a figure that differs between the deck and the plan is the fastest way to lose a reader's confidence. As of September 2026.

2.5 The Appendices, and What Belongs There Instead of the Body

Appendices exist to keep evidence available without letting it interrupt the argument. The test for moving something to an appendix has one question: does a reader need this to follow the reasoning, or only to verify it? Material needed to follow stays in the body. Material needed only to verify goes to the back.

Applied to common cases, that test produces the following assignments. A summary of the competitive landscape stays in the body; the detailed feature-by-feature comparison across eleven competitors goes to the appendix. The statement that the founder has fourteen years in the trade stays in the body; the résumé goes to the appendix. The equipment cost total stays in the body; the three quotes that produced it go to the appendix. The revenue forecast by year stays in the body; the month-by-month detail goes to the appendix — unless a lender has asked for monthly detail in the body, which several programs do (Futurpreneur Canada, 2026).

Number the appendices, list them in the table of contents, and reference each one from the body at the point where it becomes relevant. An appendix nobody is pointed to is an appendix nobody opens.

2.5.1 The standard appendix set

Most plans converge on a similar set of appendices, and knowing the list in advance means you can collect the material while you write rather than in a scramble at the end. The following list gives what a lender or a program officer most often expects to find, with a note on what makes each one usable rather than merely present.

• The résumés of the principals, one page each, edited for relevance to this business rather than sent as a general job-hunting document.
• The historical financial statements, if the business has any, for two or three years, in the form your accountant produced them rather than retyped.
• The month-by-month financial detail behind the annual summaries in the body, since several programs require it and every careful reader wants it.
• The equipment or renovation quotes that produced the capital cost figures, dated and from named suppliers, because a reader will otherwise wonder where the number came from.
• The letters of intent, contracts or commitments you have in hand, since a signed commitment is the single most persuasive item a plan can carry.
• The registration extract or incorporation documents, which cost nothing to obtain and settle every question about legal identity at once.
• The lease or the offer on premises, when the plan depends on a location.
• The insurance certificates, when the risk section relies on transfer to an insurer.
• The market evidence that would clutter the body: the full source tables, the survey instrument, the interview list with dates.

Two habits make this set work. Collect each item the day it exists rather than the week it is needed, and keep a checklist of what is missing, because the last missing appendix is the reason plans sit unsent. And name each file for what it is, since an appendix folder full of documents called "scan_0043" will be opened by nobody.

2.5.2 What never belongs in an appendix

Three things get put in appendices and should not be. Material that contradicts the body — an older forecast, a superseded price list — must be removed, not filed. Material you have not read — a market report attached because it looked authoritative — must be removed, because you will be asked about it. And material that is confidential beyond the purpose of the document, such as complete customer lists or employee personal information, must be removed, because a plan travels further than you expect and personal information carries obligations of its own. → see Book H2 (Privacy and Personal Information).

2.6 The Order of Writing Is Not the Order of Reading

This section contains the single most useful procedural idea in the chapter.

2.6.1 Why front-to-back fails

Written in reading order, the executive summary comes first — before the market work, before the operations arithmetic, before the numbers. It therefore has to be written from what you currently believe, which means it will be rewritten at least twice as the actual work changes the beliefs. Worse, the early summary anchors the rest of the document: having written that revenue will reach a certain figure, you will find yourself building a forecast toward it rather than from the drivers.

2.6.2 The construction order

The order below consumes each section's output as the next section's input, so that nothing is written twice.

Step one: the assumption log. Before any prose, list what you are assuming — price, volume, conversion, cost, timing — with a source and a confidence level for each. This is a working file, not a section.

Step two: market analysis. The demand argument, built from evidence. This produces the numbers the sales forecast will use.

Step three: the offer and go-to-market. What is sold, at what price, reached how, at what cost. This produces the price and the acquisition cost.

Step four: operations and team. How it is delivered and by whom. This produces capacity and the cost structure — and, critically, the ceiling that the sales forecast must respect.

Step five: financials. Now every input exists. Build the forecast from drivers, then the income statement, then the cash flow, then the balance sheet, then break-even and sensitivity.

Step six: the financing request. Derived from the cash curve produced in step five. Never before.

Step seven: risks. Written last among the analytic sections, because you now know which assumptions the whole structure rests on, and those are the risks that matter.

Step eight: the company section. Facts of record, mission, history, ownership. Quick, and it needs nothing from the others.

Step nine: the executive summary. Written last, from finished inputs, in one pass.

Step ten: the reconciliation pass. Read the whole document in reading order and check that every number appearing in more than one place is the same number. This pass finds, on average, three to six inconsistencies in a first plan, and it is the difference between a document that survives a careful reader and one that does not.

METHOD The order of writing: financials before summary, always Use this as a checklist while writing, in this exact sequence. First, write the assumption log as a table, and stop when every row has a source. Second, write the market section, and extract from it the two or three numbers the forecast will use. Third, write the offer and go-to-market, and extract the price, the sales cycle and the acquisition cost. Fourth, write operations and team, and compute the capacity ceiling before writing another word of narrative. Fifth, build the financial model in the order revenue, costs, income statement, cash flow, balance sheet, break-even, sensitivity. Sixth, derive the financing request from the cash trough. Seventh, write the risks against the assumptions the model turned out to be most sensitive to. Eighth, write the company section. Ninth, write the executive summary in one sitting. Tenth, run the reconciliation pass with a printed copy and a pen. The deliverable is a complete draft. The verification criterion is that the reconciliation pass finds fewer than three inconsistencies — if it finds more, the model and the narrative drifted apart, and the fix is in the model.

Figure 2.1 puts the two orders side by side. The left column lists the nine sections in the order a reader encounters them, and each section carries a badge giving the number of the construction step that produces it. The right column lists the ten construction steps in writing order, each with what it produces for the step that follows. Read the badges down the left column and the point of the figure appears at once: the summary that is read first is written ninth, the risks that are read ninth are written seventh, and two of the ten steps — the assumption log and the reconciliation pass — produce no section of the document at all.

Figure 2.1 — Original figure. Reading order and construction order: the nine sections as a reader meets them, the ten steps as you write them, and which step produces which section.

2.7 Versioning, Dating and the Assumption Log

2.7.1 Versioning that survives a year

Adopt a two-number version scheme and put it on the cover: the first number increments when the plan is issued to an external reader, the second when it is revised internally. A plan at version 2.3 has been issued twice and revised three times since the second issue. Keep every issued version; the point of a plan is to be comparable with what happened, and you cannot compare against a document you overwrote. Quality-management standards ask the same of any controlled document: ISO 9001 expects documented information to carry an identification such as a title, a date and a reference number, and to be kept under change control, version by version (International Organization for Standardization, 2015).

Name files so they sort chronologically and carry the version, in the pattern 2026-09-09_Nordec_Business_Plan_v2.3.pdf. This looks pedantic until the first time somebody asks which version the bank has.

2.7.2 The assumption log

The assumption log is the working document behind the financials, and it is the single most valuable artifact the planning process produces. It has five columns: the assumption, its value, its source, your confidence, and what changes if it turns out to be wrong.

It is not usually delivered with the plan. It is what you open in six months when a number stops matching reality, and it is what lets you answer a lender's question about an input without reconstructing the model from memory.

Chapter 7 builds one in full for the running case. What matters here is that it exists from step one rather than being reverse-engineered at the end, which is how most people discover they cannot remember where a figure came from.

CASE IN POINT Nordec Entretien: the table of contents that saved two weeks This case is fictional; the company, the people and the figures are invented. When Sophie Nadeau rewrote her plan, she began with something that felt like procrastination: she wrote the table of contents, with a page budget beside each section, before writing a word of body text. The nine sections, the page counts, the appendices, and beside each one a single question in italics that the section had to answer. Three things followed from that one page. She discovered that she had been planning to write eleven pages about her service offering and three about her market, which was exactly backwards for a lender. She discovered that two of her intended sections — one about her software, one about her environmental practices — answered no question anybody had, and both were folded into operations in a paragraph each. And she discovered that the page budget summed to thirty-four pages before appendices, which was too long, so she cut before writing rather than after. Her estimate afterwards was that the exercise took ninety minutes and saved about two weeks, most of it in text she never had to write. The version she delivered was nineteen pages.

2.7.3 Where the working files live

The plan itself, the model, the assumption log, the evidence folder and the version history need a home that is not one person's laptop. The requirement is unglamorous and it is the reason plans get lost: a business that loses the model behind its plan has lost the ability to update the plan, and will write the next one from scratch.

TOOLS OF THE TRADE What a document environment must guarantee for a plan Independent of which product you choose — chapter 13 names them, with dated vendor pages — the environment holding a plan must guarantee five things. It must keep a history of versions you can return to, since the point of versioning is being able to open version 1.0 a year later. It must allow more than one person to work without overwriting each other, because the reconciliation pass is usually done by a second reader. It must keep the evidence with the claim, so a statistic and the date it was retrieved travel together. It must export to PDF with the tables intact, because a re-flowed financial table is worse than no table. And it must be backed up somewhere that is not the machine you are typing on, which is the requirement most often discovered in its absence. → see Book K3 (Cybersecurity, Backups and Business Continuity).

2.8 Adapting the Anatomy: Startup, Existing Business, Acquisition, Non-Profit

The nine sections hold for every case, but their weight and content change materially across four situations.

2.8.1 The startup with no history

Everything in the plan is forward-looking, and the reader knows it. Two adaptations follow. The team section gains weight, because in the absence of a track record for the business, the track record of the people is the evidence (Sahlman, 1997). And the market section must carry primary evidence — interviews conducted, pre-orders taken, a pilot run — rather than only published data, because published data about a market says nothing about whether this particular business can reach it.

Do not compensate for the absence of history with more precision in the forecast. A monthly forecast to the dollar for year three, from a business with no customers, reads as a failure of judgment rather than as diligence.

2.8.2 The existing business

The plan gains a section of actual results — usually two or three years of income statements and the current balance sheet — and the forecast must reconcile with them. A forecast that shows a sudden inflection with no corresponding change in the business is the most common credibility failure in this category, and the fix is to name the change explicitly: the second vehicle, the new contract, the additional technician.

The market section can be shorter, because the business has demonstrated demand. The operations section gets longer, because the constraint has moved from finding customers to serving them.

2.8.3 The acquisition

A plan to buy a business is a different document wearing the same anatomy. The company section describes two entities, the target and the acquirer. The market section is about the target's market. The financials carry the target's historical statements, the purchase price and its financing, the opening balance sheet after the transaction, and the forecast under new ownership — which must show what the new owner will do differently, since a forecast identical to the target's history raises the obvious question of why the price is justified. → see Book M6 (Buying a Business).

2.8.4 The non-profit or social enterprise

The anatomy holds, with three changes. The mission is load-bearing rather than decorative, because it defines eligibility for the funding being sought. The financials show funding sources rather than a financing request: grants, contracts, donations, earned revenue, each with its own timing and conditions. And a results and evaluation section takes the place a return-on-investment discussion would occupy in a commercial plan, stating what outcomes will be measured and how. → see Book G7 (Grants, Subsidies and Public Programs).

The paragraph that follows introduces Table 2.2, which compares the four situations. Its columns give, for each one, the section that carries the most weight, the section that can be shortest, the piece of evidence a reader will look for first, and the mistake most often made in that situation. Find your own case and read across.

Table 2.2 — Four situations: what gets weight, what gets short, and the mistake to avoid

2.9 The Anatomy in Three Situations

2.9.1 Solo

Write a fifteen-page plan and a one-page summary, and put the financial detail in the model rather than in prose. The one adaptation that matters for a solo operator is in the team section: it must address, in a sentence, what happens if you are unavailable for six weeks. Every reader wonders; almost no solo plan says.

Your realistic sequencing constraint is that you will write this in evenings and weekends over four to six weeks. Follow the construction order strictly, because it is the only thing that prevents rework, and rework is what kills a plan written in fragments.

2.9.2 Small team

One person owns the document. Each section has a named contributor who supplies material, but the owner writes the final text so the voice stays single. Hold two working sessions: one after step four, where the capacity ceiling meets the sales forecast and the argument usually happens, and one after step nine, for the reconciliation pass with fresh eyes.

The failure mode here is a plan assembled from three people's drafts, in three voices, with three different sets of numbers. The reconciliation pass exists to catch it.

2.9.3 Established business

The controller owns the financial sections and the model; the operations lead owns capacity, process and suppliers; the owner owns the summary, the company section and the strategy. Add a formal review step: someone who did not write any of it reads the whole document in reading order and lists every inconsistency.

At this size, the plan is also an internal alignment instrument, and its distribution matters as much as its content. Decide before writing who receives the full document, who receives the summary, and what is not distributed at all.

WHO DOES THIS Who owns each section when a team writes together The owner or general manager owns the executive summary, the company section and the strategic argument that runs through the market analysis, because these carry the judgments only they can make. The finance person — controller, bookkeeper or external accountant — owns the model, the three statements and the assumption log, and is the right person to run the reconciliation pass because they are the one trained to notice that a number does not tie. The operations lead owns capacity, process, suppliers and the milestone table, and holds a specific veto: if the sales forecast exceeds what operations can deliver, operations wins the argument or the plan is fiction. The person who sells owns go-to-market, the sales cycle assumptions and the conversion rates, and should be asked to state the evidence behind each rate rather than the rate alone. In a solo business all four roles are you, and the countermeasure is temporal rather than organizational: write each part on a different day, and read the model with the sales hat off.

DO THE MATH What a page costs, counted across five readers Length has a cost that is easy to ignore because nobody invoices you for it. Assume a plan is read by five people — you, a lender, an accountant, a co-owner and a program officer — and that a page of dense business prose with a table takes about 2.5 minutes to read attentively. Cost of one page, across the five readers: 5 × 2.5 = 12.5 minutes of professional attention. Ten unnecessary pages: 10 × 12.5 = 125 minutes, or a little over two hours of reading time consumed by material that changed nobody's decision. Now the writer's side. Those ten pages took roughly 45 minutes each to draft, revise and proofread: 10 × 45 = 450 minutes, or 7.5 hours. Total cost of ten unnecessary pages: 7.5 hours of your time plus two hours of everyone else's, against a benefit of zero by construction, since we defined the pages as unnecessary. The number worth remembering is 7.5 hours. That is approximately the time the cash flow forecast in chapter 7 requires, and it is the trade most first-time plans get backwards.

Costly Mistakes

1. Writing front to back. The executive summary gets drafted three times and the financial section gets whatever hours remain. The cost is roughly a week of rework and a financial section that nobody had time to reconcile.
2. Omitting the date and the version number. Two versions circulate, a reader comments on the wrong one, and a meeting is spent establishing which document everyone is looking at.
3. Inventing a tenth section. Unexpected sections get skipped, and the material in them — which was often good — is lost rather than placed where a reader would have found it.
4. Filing contradictory material in an appendix instead of deleting it. A reader who finds an older forecast in appendix D now distrusts the current one, and you will spend the meeting explaining rather than discussing.
5. Letting the deck and the document diverge. A figure that differs between the two costs more credibility than either document earns, because it proves that at least one was not checked.
6. Skipping the reconciliation pass. It takes ninety minutes and it typically finds three to six inconsistencies. Every one of them would otherwise be found by a reader, in front of you.
7. Putting the month-by-month detail in the body because it looks thorough. It pushes the argument off the page and, for most readers, is exactly what an appendix is for — with the specific exception of programs that ask for monthly detail, which several do.

Key Takeaways

• A complete plan has nine body sections plus front matter and appendices, and the sequence follows a chain of dependency that lets a reader move forward without back-tracking.
• Every section answers exactly one question; material that answers no question is evidence, and evidence belongs in an appendix.
• The cover page carries seven items including the date and the version number, and their absence causes a specific and common failure.
• Twenty to thirty pages of body is the working range for external financing; eight to fifteen is plenty for an internal document.
• Most businesses need two artifacts — the full document and a short version — plus a sendable financial model, and the short version is derived from the summary rather than written separately.
• The construction order is not the reading order: assumptions, market, offer, operations, financials, financing request, risks, company, summary, reconciliation.
• The reconciliation pass finds three to six inconsistencies in a typical first plan, and it is the cheapest quality step in the entire process.
• The anatomy holds across startups, existing businesses, acquisitions and non-profits, but the weight of each section and the evidence a reader looks for change materially.

Your Action Plan

1. Write your table of contents with a page budget, before any body text. Deliverable: one page listing the nine sections, the page budget of each, and one italic question per section. Timing: ninety minutes. Verification criterion: the page budgets sum to a number inside your target range, and every section has a question that a named reader cares about.
2. Create the assumption log as a five-column table today. Deliverable: a table with assumption, value, source, confidence and consequence-if-wrong, holding at least ten rows. Timing: one hour. Verification criterion: no row has an empty source cell, including the ones whose source is your own estimate.
3. Set up the version scheme and the file naming convention now. Deliverable: a folder with a dated, versioned file name and a place for issued versions to be kept rather than overwritten. Timing: fifteen minutes. Verification criterion: you can state which version the last external reader received.
4. Draft the cover page and the confidentiality notice. Deliverable: one page carrying the seven items, plus the notice on the reverse. Timing: thirty minutes. Verification criterion: the date and the version number are both present.
5. Schedule the reconciliation pass before you start writing. Deliverable: a calendar block of ninety minutes, placed after the drafting sessions, with a named second reader if one is available. Timing: five minutes. Verification criterion: the block exists on a real date and names what will be checked.

In La Machine

Every route named in this section was verified in the product's route inventory on 2026-09-09 (La Machine, 2026a), and what each page holds is described in the product's user guide (La Machine, 2026b).

• The writing sprint becomes a project. /projets holds the plan itself as a project — a one-off rather than a chain, since it closes when the document is delivered. Its tasks live at /taches, one per writing session, each carrying the deliverable as its verification criterion.
• Notes and links while you gather evidence. /presse-papier holds notes that can later be attached to a project or a task; /liens holds the addresses you reopen daily, which during a writing sprint are usually the statistical agency, the registry and the lender's document list.
• Versions of things the business publishes. versions holds the version registry, and backups the backups, for the business as a system rather than for your document specifically.
• Decisions that stall the drafting. arbitrages is where a question that stops a section goes, so that the session can continue and the decision is not lost.
• What it does not do. La Machine holds no business-plan template and does not assemble, format or export the document. It does not version your manuscript, does not track changes in prose, and does not produce a table of contents. The plan is written in the tools of chapter 13; what La Machine holds is the work the plan creates.

Self-Check

1. Name the nine sections in reading order, and give the single question the market analysis section must answer. Say why "is this a big market?" is not that question.
2. You have written a two-page description of your production software. Apply the test of §2.1.2 and say where it belongs, and under what condition your answer would change.
3. Give the construction order in ten steps. Explain, in one sentence, why the financing request cannot be written before step five.
4. A reader finds a superseded price list in appendix D. Describe the two effects this has on how they read the rest of the document.
5. Your plan will be read by a program officer at a public agency and by your own operations lead. State one adaptation you would make to the anatomy for each, and say which section each of them opens first.

Chapter References

Business Development Bank of Canada. (2026a). Business plan template for entrepreneurs (downloadable template with guide, instructions, glossary and worked example). Retrieved September 9, 2026.

Civil Code of Québec, CQLR c CCQ-1991, arts. 1612 and 2088 (Québec). LégisQuébec, consolidated version updated to April 7, 2026. Read September 9, 2026.

Futurpreneur Canada. (2026). Frequently asked questions (required uploads: completed business plan, two-year month-by-month financial forecasts, government identification). Retrieved September 9, 2026.

International Organization for Standardization. (2015). ISO 9001:2015 — Quality management systems — Requirements, clause 7.5 (documented information: creation, updating and control). ISO.

La Machine. (2026a). Documentation — route reference inventory (211 routes: 191 Core, 20 Pilotage paths). Internal product documentation, read September 9, 2026.

La Machine. (2026b). Guide utilisateur [User guide] (FAQ: projects and task chains; your favorite links and quick notes; versions and migrations). Internal product documentation, read September 9, 2026.

Sahlman, W. A. (1997). How to write a great business plan. Harvard Business Review, 75(4), 98–108.

Chapter 3 — The Executive Summary and the Company Section

These are the two sections a reader meets first and the two you should write nearly last. The executive summary is the hardest page in the document, because it must contain the whole argument in about six hundred words and it must do so without becoming a list. The company section is the easiest, because most of it is facts of record — and it is also the section where more first-time plans quietly damage their own credibility than anywhere else, by drifting from verifiable facts into aspiration in the space of a paragraph.

What You'll Be Able to Do

• Write a one-page executive summary containing all seven required elements, in a single pass, from finished inputs.
• Open with a first sentence that answers the reader's question, and recognize the four openings that fail.
• State the facts of record — legal name, form, registration number, address, date of formation — so that they match the register exactly.
• Write a mission that can be tested, and distinguish it from a vision and from values without producing fog.
• Present history as dated milestones rather than as narrative, so that a reader can audit the track record.
• Describe ownership in plain language, including the cap table, without requiring the reader to be a corporate lawyer.

Why It Matters

Two facts about how plans are actually read make this chapter disproportionately important. The first is that a great many readers read the executive summary and nothing else — not because they are lazy, but because they are triaging, and the summary is what tells them whether the full document is worth an hour. The second is that the company section is where a careful reader performs their first verification: they take your legal name and your registration number, they look them up, and they find out whether the document is accurate about things that are easy to check. A discrepancy there — a company name that does not match the register, a date of formation that is off by a year — costs far more credibility than its objective importance, because it is the first test and you failed it.

There is a third reason, less obvious. Writing the summary is the last honest check on whether the plan holds together. If you cannot state the argument in six hundred words, the argument has a hole in it, and the hole is usually in the same place: between what the market section claims and what the financial section assumes.

In This Chapter

• 3.1 What the Executive Summary Is For
• 3.2 The Seven Things It Must Contain
• 3.3 Writing It Last, in One Page
• 3.4 The Opening Sentence, and Four Ways to Get It Wrong
• 3.5 The Company Section: Legal Identity and Facts of Record
• 3.6 Mission, Vision and Values Without the Fog
• 3.7 History, Milestones and Track Record
• 3.8 Ownership, Founders and the Cap Table in Plain Language
• 3.9 The Elevator Version, the Page and the Deck
• 3.10 These Two Sections in Three Situations

3.1 What the Executive Summary Is For

The executive summary has one job: to let a reader make a preliminary decision without reading the rest of the document. That framing settles most of the questions people have about it.

It settles the length question, because a preliminary decision does not require three pages. It settles the content question, because a preliminary decision requires the numbers and the ask, which many summaries omit as though they were a surprise to be revealed later. And it settles the tone question, because a preliminary decision is not made on enthusiasm.

KEY CONCEPT The summary is the plan, compressed — not an introduction to it The difference is testable in one move. Read your summary and ask: does it state what the plan concludes, or does it state what the plan will discuss? A summary that says "this plan presents our market analysis, our operating model and our financial projections" has described a table of contents. A summary that says "we serve 34 buildings under contract, we can serve 52 with one additional vehicle and two technicians, and we are asking for $132,000 to get there" has stated a conclusion. The second one lets a reader decide something. The first one requires them to read on to find out whether there is anything to decide, and many will not.

3.1.1 Why it is written last

The summary is a compression of finished material. Written before the material exists, it is a compression of your intentions, which is a different and much less useful thing — and, as chapter 2 argued, it then anchors the rest of the document to a number you invented before you did the arithmetic.

There is one legitimate exception, and it is a working exception rather than a delivery one. Some writers draft a rough summary at the start as a statement of intent, keep it in the working file, and then throw it away and write the real one at the end. The value of that practice is that it makes explicit what you currently believe, so that you notice when the analysis contradicts it. If you do this, label the draft clearly and never let it survive into the document.

3.1.2 What a reader does with it

Watch a professional reader with a plan and you see a consistent behavior. They read the summary once, quickly. They then read it a second time with a pen, and they mark two or three things: the number that seems high, the claim with no evidence behind it, and the sentence that answers the question they came with. Those marks become the questions they ask you, or the sections they turn to.

This means the summary is not only a decision aid; it is a navigation instrument. A summary that names its sources and points to sections — "the capacity arithmetic is in section 6" — converts a skeptical reader into a directed one.

3.2 The Seven Things It Must Contain

A complete executive summary contains seven blocks. They can be paragraphs or labeled sub-headings; sub-headings are more useful in a longer summary and slightly more clinical in tone.

Block one: what the business is and does. The legal name, the location, what it sells and to whom, in two or three sentences. This is where the first sentence lives, and §3.4 is entirely about it.

Block two: the demand. What problem the customer has, what they do about it today, and why that is unsatisfactory. Two to four sentences, with at least one fact rather than an assertion.

Block three: the offer, and why it is chosen. What you sell, what it costs the customer, and the specific reason a buyer picks you over their current alternative. This block fails when it lists features; it works when it names the decision criterion.

Block four: the market and the evidence. How large the reachable market is, how you know, and what traction exists — customers, contracts, letters of intent, a pilot. For an existing business, the evidence is your own revenue history, which is the strongest form there is.

Block five: the team. Who runs it, what they have done before that is relevant, and — briefly — what gap exists and how it will be filled; a reader weighs the people before the projections, because what they have already done is the best evidence of what they will do (Sahlman, 1997). Two or three sentences.

Block six: the numbers. Revenue, gross margin and net result for the forecast horizon, plus the one operational number that drives them. Not a table; three or four sentences carrying five or six figures. This is the block most often omitted, and its omission reads as evasion.

Block seven: the ask and the use of funds. How much, in what form, over what term, for what specifically, and what it buys. If the plan is internal and there is no ask, this block states the decision being requested instead.

The paragraph that follows introduces Table 3.1, which presents the seven blocks with a word budget for each, based on a summary of roughly six hundred words. Its columns give the block, what it must answer, the word budget, and the failure mode that block characteristically suffers. Use the word budget literally: it is what forces the compression that makes a summary readable, and running long on block one is the most common way to run out of room for block six.

Table 3.1 — The seven blocks of an executive summary: word budget and characteristic failure

Figure 3.1 draws the same seven blocks as a single page. Each band represents one block, sized in proportion to its word budget, with the question it answers on the left and its characteristic failure on the right. Reading top to bottom gives you the order the blocks appear on the page; reading the band widths gives you the compression each one demands. The point of drawing it this way is that a summary is a fixed-size container: every word spent on block one is a word unavailable to block six, and block six is the one a lender reads twice.

Figure 3.1 — Original figure. The executive summary as a fixed-size page: seven blocks, their word budgets, and the failure each one characteristically suffers.

3.2.1 The seven blocks, written out

Abstractions about summaries are less useful than a summary. What follows is the complete executive summary of the running case, written to the budgets of Table 3.1, with the block numbers shown in brackets so you can see the joints. The company and its figures are fictional; the structure is not.

[1] Nordec Entretien inc. provides preventive maintenance and on-call technical service to 34 commercial buildings on Montréal's north shore, under monthly contracts averaging $685 per building. The company was incorporated in Québec in 2022, operates one service vehicle and employs four people.

[2] Property managers of buildings between 2,000 and 9,000 square metres are too small to employ a building technician and too large to rely on emergency call-outs. Their usual arrangement is a patchwork of trades called individually when something fails, which means every incident starts with finding somebody available. Across our own contract base, buildings arriving from that arrangement report an average of eleven unplanned interventions in their last twelve months.

[3] We sell a monthly preventive contract covering scheduled inspection of mechanical, plumbing and door systems, with a 90-minute response commitment on contract calls and hourly billing for work beyond the scope. Managers choose us over a patchwork because one call reaches somebody who already knows their building, and because unplanned interventions fall by roughly half in the first year under contract.

[4] There are approximately 1,400 buildings in our service area in this size range that we can reach; we serve 34 of them. In the last twelve months we added nine buildings and lost four, and our contract renewal rate was 88%. Of last year's $612,100 in revenue, $279,500 was contract revenue.

[5] Sophie Nadeau, president, worked fourteen years as a building systems technician, the last four as a service supervisor responsible for a portfolio of thirty buildings. Two full-time technicians and a part-time administrator complete the team. The company has no dispatcher; that role is currently carried by the president and is the first hire this plan funds.

[6] Revenue for the year just ended was $612,100. The plan forecasts $700,500, $832,900 and $933,800 over the next three years, with gross margin moving from 39% to 41% and net income from $34,300 to $79,100. The driver is buildings under contract, which moves from 34 to 52 over the period; every other line follows from it. Cash reaches its lowest point in month five of year one, at negative $31,400 before financing.

[7] We are requesting $132,000: a $92,000 five-year term loan for a second equipped service vehicle and technician tooling, and a $40,000 operating line for the receivable gap created by commercial payment terms, against a forecast cash trough of $31,400 in month five. The vehicle lifts billable capacity by roughly 2,000 hours a year, which is what the revenue forecast requires from month four onward.

Three things are worth noticing about that page. Every paragraph contains at least one number, and none of the numbers is decorative — each one is used later in the plan. The gap is named in block five rather than hidden, and naming it makes the ask in block seven legible. And block seven derives from block six: the reader can see where $132,000 came from, which is the difference between an ask and a request.

3.3 Writing It Last, in One Page

3.3.1 The ninety-minute procedure

The summary is written in one sitting, from material that already exists. The procedure below assumes the rest of the plan is drafted.

METHOD Writing the executive summary in seven blocks, in ninety minutes Minutes 0 to 10. Open the finished plan and copy, into a scratch file, the single most important sentence from each of the nine sections. You now have nine sentences and no writing to invent. Minutes 10 to 25. Write block six first — the numbers — because it is the hardest to compress and the easiest to get wrong. Take the figures directly from the statements; do not retype them from memory. Minutes 25 to 35. Write block seven, the ask, directly from the financing section. If you cannot state what the money buys in one sentence, the financing section is not finished. Minutes 35 to 65. Write blocks one through five, in order, respecting the word budgets. Set a word counter visible while you write. Minutes 65 to 75. Write the first sentence last, when you know what the summary says. Try three versions and keep the one that contains a verifiable fact. Minutes 75 to 90. Read the whole thing aloud, once, at speaking pace. Anything you stumble over gets rewritten; anything that takes more than one breath gets split. The deliverable is a one-page summary of about six hundred words. The verification criterion is that a person who has not read the plan can state, after reading only the summary, what the business does, what is being asked, and what the money buys.

3.3.2 Reading it aloud is not optional

The read-aloud step in that procedure catches more problems than any other single technique, and it does so for a mechanical reason: the ear notices sentence length, repetition and unsupported superlatives that the eye slides over. A summary that cannot be read aloud comfortably will not be read comfortably in silence either.

This is also the step that catches the plan's remaining jargon. If you say "we leverage synergies across our vertical" out loud, in a room, to a person, you will hear it.

3.3.3 The one-page constraint is real

Keep the summary to one page for a plan under thirty pages, and to two pages only if the plan exceeds forty. The constraint is not typographic superstition. A one-page summary is read completely; a three-page summary is skimmed, which defeats the purpose of writing a summary at all.

If the material will not fit, the problem is almost never that the business is unusually complex. It is that one block has absorbed material belonging to the body — most often block three, which expands to accommodate a description of the offer that belongs in section 4 of the plan.

3.4 The Opening Sentence, and Four Ways to Get It Wrong

The first sentence of the summary is the first sentence of the plan, and it sets the reader's expectation about everything that follows. Four openings recur, and all four waste it.

3.4.1 The mission opening

"Our mission is to become the leading provider of building maintenance services in the greater Montréal region." This sentence contains no information. Every business in every plan intends to be a leading provider of something, so the sentence distinguishes nothing, and it costs the reader six seconds to learn that.

3.4.2 The industry background opening

"The commercial building maintenance industry has undergone significant transformation over the past decade." The reader either knows this — in which case you have told them nothing — or does not, in which case they are not the reader you are writing for. Industry background belongs in section 3, and even there it must earn its place.

3.4.3 The founder story opening

"After fourteen years working as a technician, I realized there had to be a better way." Founder stories are genuinely valuable and this is not where they go. A lender wants to know what the business is before knowing how it came to be; the story belongs in the company section, where it supports the track record.

3.4.4 The vague superlative opening

"We offer superior service at competitive prices." This is the worst of the four, because it is not merely uninformative — it is a claim of differentiation that explicitly declines to say what the differentiation is. A reader who has seen it before, and they all have, will read the rest of the document expecting to find more of the same.

3.4.5 What works instead

An opening sentence that works usually contains three things: who the business is, what it does for whom, and one verifiable fact. The fact is what does the work, because it signals that the document contains checkable content.

Compare the four failures above with this: "Nordec Entretien inc. provides preventive maintenance and on-call technical service to 34 commercial buildings on Montréal's north shore, under monthly contracts averaging $685 per building." Twenty-nine words. Three of them are numbers. A reader now knows what the business is, what it sells, at what scale and at what price, and knows that the writer counts things.

WATCH OUT The mission statement that could belong to any company on earth Run this test on any mission, vision or opening sentence in your draft: substitute a competitor's name into the sentence and see whether it becomes false. If it remains equally true with your competitor's name in it, the sentence describes the industry rather than your business, and it is occupying space that a distinguishing fact should have. This test is brutal on first drafts, and it is the fastest editing tool in this chapter. It also applies to values, where the failure rate is close to total.

3.5 The Company Section: Legal Identity and Facts of Record

This section states who the business is, legally and factually. It should be short, dense and exactly correct, because it is the part of the plan a reader can verify in four minutes.

3.5.1 The facts of record

Nine items belong here, and all nine must match the public register exactly:

• The legal name of the entity, in the exact form it is registered, including the corporate suffix.
• Any trade name or business name used, and its relationship to the legal name.
• The legal form — sole proprietorship, general partnership, limited partnership, corporation, cooperative, non-profit.
• The jurisdiction of formation, provincial or federal, and the statute under which it exists.
• The registration number — in Québec the Québec enterprise number, which has ten digits and is assigned by the Registraire des entreprises when the enterprise is registered.
• The date of formation and, if different, the date operations began.
• The head office address, and the addresses of any other places of business.
• The fiscal year end, which matters more than it seems because it determines when statements exist.
• The directors and officers, by name and role.

THE RULE Every fact of record must match the register Before the plan is delivered, obtain a current extract from the register and check the nine items against it, one by one, with the extract open beside the draft. This takes about ten minutes and it prevents the specific failure that costs the most credibility for the least reason: a document that misstates something a reader can verify in four minutes. In Québec, registration with the enterprise register is required as soon as a business operates under a name other than the owner's own, and every registered enterprise must file an annual updating declaration; both the enterprise number and the current information are publicly searchable, which is exactly why the check matters. Attach the extract as an appendix. → see Book A4 (Legal Structures, Registration, Permits and Licences) for the registration itself.

3.5.2 What the business does, in one paragraph

After the facts, one paragraph stating what the business actually does: the activity, the customers, the geography, the scale. This paragraph is frequently the one lifted verbatim into other documents — grant applications, supplier forms, partner agreements — so it is worth writing well once.

Write it in the present tense and in the third person, and include at least one quantity. "The company operates two service vehicles and employs four people, serving commercial buildings within a forty-kilometre radius of Sainte-Thérèse" is more useful than three sentences of description without a number in them.

3.5.3 What changes with the legal form

The facts of record are the same nine items in every case, but three of them behave differently depending on the form, and readers know it.

In a sole proprietorship, there is no separation between the business and the person. The plan must therefore say so explicitly rather than leaving the reader to infer it, because it changes what a guarantee means and what happens to the business if the owner is unavailable. Registration is still required in Québec as soon as the business operates under a name other than the owner's own.

In a partnership, the item that matters most is the one plans most often omit: whether a written partnership agreement exists, and whether any partner can bind the partnership. In most forms any partner can, and a reader assessing risk wants to know that fact and whether the partners know it.

In a corporation, the reader is looking at three things — the jurisdiction, the directors, and whether there is more than one class of shares. A federally incorporated company operating in Québec appears in two registers, and the two must agree; a discrepancy between them is a live administrative problem rather than a paperwork detail.

In a cooperative or a non-profit, the governing statute and the composition of the board carry weight the equivalent items do not carry in a commercial corporation, because they determine eligibility for much of the funding such organizations seek.

3.5.4 Three errors that recur in this section

The first is the outdated address. A business moves, updates its mail and its website, and never updates the register; the plan then carries whichever address the writer remembered. The extract check catches it.

The second is the informal name. The business is registered as one thing, everybody calls it another, and the plan uses the second without ever stating the first. Give both, and say which is which.

The third is the missing fiscal year end. It seems administrative until a lender asks for interim statements and discovers that the year end falls two months from now, which changes what documents exist and when.

QUÉBEC & CANADA NOTE The register, the number and what must agree In Québec, the Registraire des entreprises assigns a Québec enterprise number (NEQ) of ten digits to every registered enterprise, and the register is public and searchable by name or by number (Registraire des entreprises du Québec, 2026). Registration is required for most businesses operating under a name that is not the owner's own name, and each registered enterprise must file an annual updating declaration; the fees and penalties attached to it change from year to year, so verify the current amount at the source rather than quoting a figure from a document of unknown vintage. A federally incorporated company is registered with Corporations Canada and files its annual return there, and if it operates in Québec it must also be registered with the Québec register (Corporations Canada, 2026) — which means two sets of records that must agree with each other and with your plan. The practical instruction for this chapter is narrow: pull the extract, check the nine items, attach it. → see Book A4 (Legal Structures, Registration, Permits and Licences) and Book H6 (Corporate Governance, Shareholder Agreements and Minute Books).

3.6 Mission, Vision and Values Without the Fog

These three words have been abused so thoroughly that many owners now skip them, which is an over-correction. Each has a specific and useful meaning, and each is worth a sentence — not a page.

3.6.1 The mission: what the business does, for whom, and to what end

A mission states what the business does, for whom, and what result it produces. It is present tense and it is testable, in the sense that you could tell whether the business was doing it.

"We keep commercial buildings running so that our clients never have to think about their mechanical systems" is a mission: it names an activity, a customer and a result. "To be the most trusted name in building services" is not a mission; it is an ambition, and it belongs — if anywhere — in the vision.

3.6.2 The vision: what the business is trying to become

A vision states a future state, usually with a horizon (Collins & Porras, 1996). It is legitimately aspirational, and it is useful precisely because it is not the mission: it tells an employee or an investor where this is going.

A vision earns its place when it constrains something. "By 2030, half our revenue comes from contracts rather than call-outs" is a vision that shapes decisions, because it tells you which customers to pursue. "To be the leader in our field" constrains nothing.

3.6.3 Values: what the business does when it costs something

Values are the behaviors the business will maintain when maintaining them is expensive. That definition is the whole test. "Integrity" is not a value in this sense, because nobody claims the opposite. "We tell a client when a repair is not worth doing, even when we would be paid for it" is a value, because it has a price.

Three or four values, each stated as a behavior with a cost, are worth more than eight abstract nouns. And values that no one can point to an instance of should be removed, because an employee who has seen the business act contrary to a stated value will trust the whole document less.

The paragraph below introduces Table 3.2, which separates the three notions. Its columns give what each one is, the tense it is written in, the test that tells you it is not fog, and an example of each done badly and done well. The test column is the operative one: apply it to your own draft before anything else in this section.

Table 3.2 — Mission, vision and values: what each is, and the test that removes the fog

3.7 History, Milestones and Track Record

3.7.1 History as a table of dated milestones

The history of the business belongs in the company section, and it belongs as dated milestones rather than narrative. A reader is looking for evidence of execution: did this business say it would do things, and then do them?

A milestone entry has three parts: the date, what happened, and — where it exists — the number that makes it verifiable. "March 2024: first multi-building contract signed, three buildings, $1,840 per month" is a milestone. "2024: significant growth" is not.

Include the setbacks that a reader would discover anyway. A business that lost a major client in 2025 and says so, with what it did about it, is more credible than one whose history is a smooth ascent, because smooth ascents do not happen and readers know it.

3.7.2 Track record for a business with no history

A new business has no company history, which is not a problem as long as you do not pretend otherwise. What it has instead is the track record of its people, and that belongs here in the same form: dated, specific and verifiable. Fourteen years at a named employer, with the scope of responsibility stated, is a fact. "Extensive industry experience" is not.

The second substitute for company history is evidence of demand already gathered: interviews conducted, with dates and counts; a pilot run, with its result; pre-orders or letters of intent, with amounts. Chapter 4 treats this evidence in detail; the summary version belongs here.

CASE IN POINT Nordec Entretien: the mission that said something verifiable This case is fictional; the company, the people and the figures are invented. Sophie Nadeau's first plan carried this mission: "Nordec Entretien is committed to excellence in building maintenance services, delivering superior value to our clients through professionalism and integrity." Applying the substitution test of §3.4.5 to it — putting a competitor's name at the front — leaves the sentence exactly as true, which is the diagnosis. Her second version read: "Nordec keeps commercial buildings running, so that a property manager never has to think about mechanical systems until we tell them to." It is not elegant, and it does two things the first version could not. It names the customer — a property manager, not "clients" — and it names the result in the customer's terms: not thinking about it. When her operations lead later argued about whether to accept a one-off emergency call from a building with no contract, the mission actually settled the argument, because the sentence is about buildings the company already keeps running. The history section changed as well. The first version described 2024 as "a year of significant growth." The second gave four dated lines, including the one she had wanted to leave out: "September 2025: lost the Lachance portfolio, four buildings, $2,640 per month, after a response-time failure. Dispatch procedure rewritten in October; average response time on contract calls has since been 71 minutes against a 90-minute commitment." Her account manager mentioned that line specifically.

3.8 Ownership, Founders and the Cap Table in Plain Language

3.8.1 What the reader needs to know

Ownership matters to a reader for three reasons: it establishes who can commit the business, it determines who is at risk, and it reveals whether there are silent parties whose interests might diverge. State it plainly.

For a sole proprietorship, there is one line: the owner, personally liable. For a corporation, state each shareholder, their percentage, their class of shares if there is more than one class, and whether a shareholder agreement exists. For a partnership, state the partners and the nature of the partnership, because the liability consequences differ substantially.

DEFINITION Cap table, founder, shareholder agreement A capitalization table, or cap table, is the list of who owns what portion of a company: each holder, the number and class of shares held, and the resulting percentage. In a small business it fits in five lines; the term sounds more technical than the object is. A founder is a person who started the business, which is a factual matter rather than a title and does not by itself confer ownership — founders who did not receive shares are employees with a story. A shareholder agreement is a contract among the owners governing what happens in the situations that otherwise destroy small companies: a shareholder wanting out, a shareholder dying, a disagreement with no majority, a new investor arriving. Its absence is a risk worth naming in chapter 9. → see Book H6 (Corporate Governance, Shareholder Agreements and Minute Books).

3.8.2 Presenting it without jargon

Present the cap table as a small table with three columns — holder, class and percentage — and one sentence beneath it stating whether a shareholder agreement is in place and what it covers in general terms. Do not reproduce the agreement; do not summarize its clauses; do not attempt legal characterization. A reader who needs that detail will ask for the document.

If ownership is about to change because of the transaction the plan supports — a new investor, a partner buying in, a departing shareholder — show both the current and the resulting cap table. A reader who has to compute the post-transaction ownership themselves will do it, and will wonder why you did not.

WHO DOES THIS Who signs the plan, and what signing implies In a sole proprietorship, the owner signs and the signature carries personal liability for everything the business commits to, which is why the personal guarantee discussion in chapter 8 matters so much in this form. In a corporation with one owner-manager, the same person signs in two capacities and it is worth being explicit about which: the plan is signed by the director on behalf of the corporation, while a personal guarantee is signed personally. In a corporation with several shareholders, the person signing must have the authority to commit the corporation, which usually comes from a directors' resolution. A lender will ask for it; having it before the meeting is faster than obtaining it afterwards. In a partnership, check the partnership agreement for who may bind the partnership, because in most forms any partner can, which is a fact worth every partner knowing before a plan circulates. In every case, the person who signs should be the person who can answer questions about the numbers. A signature from someone who cannot defend page 19 is a signature that produces a second meeting.

3.9 The Elevator Version, the Page and the Deck

Three shorter artifacts derive from the summary, and all three should be derived rather than written independently, so that they cannot diverge.

3.9.1 The elevator version: thirty seconds

Roughly seventy-five words: what the business is, what it does for whom, one number, and what you are looking for. This is blocks one, six and seven of the summary, compressed. Write it down even though it will be spoken, because a version that has been written is a version that is consistent.

3.9.2 The one-page summary as a standalone document

The same content as the executive summary, formatted to stand alone: the business name and contact at the top, the seven blocks as short paragraphs or bullets, the key figures in a small table, and the ask at the bottom. This is what you send when someone says "send me something short."

3.9.3 The deck: ten to fifteen slides

A deck follows the same seven blocks with one idea per slide, plus a title slide and a closing slide with the ask and the contact information. The temptation to add slides is strong and should be resisted; the failure mode of a deck is not that it is too short.

The sequence below is the one that maps cleanly onto the seven blocks, and it is worth building in exactly this order so that nothing is invented for the deck that is not in the plan.

1. Title. The business name, one line saying what it does, the date and the version.
2. The customer and the problem. One slide, with the fact from block two.
3. The offer. What is sold, at what price, and the decision criterion from block three.
4. Why us. The specific reason a buyer chooses this business, with evidence.
5. The market. The reachable market, sized from the bottom up, with the source.
6. Traction. What has already happened: customers, contracts, revenue, retention.
7. How it reaches the customer. The go-to-market in one slide, with a cost per acquisition.
8. Operations and capacity. How the promise is delivered and what the ceiling is.
9. The team. Who, with one relevant fact each, and the gap.
10. The numbers. Three years, revenue, margin, net, on one slide.
11. The ask and use of funds. How much, what form, what it buys.
12. The risks. Three real ones, with what you would do.
13. Closing. Contact, and the one thing you want the reader to do next.

Two slides that people add and should not: a competitor matrix in which the writer's own column is entirely checkmarks, which no reader has ever believed, and a slide of logos of tools the business uses, which answers no question anyone asked.

Two rules keep the deck honest. Every number on a slide appears identically in the document, and no slide claims anything the document does not support. The second rule sounds obvious and is violated constantly, usually by a slide added in a hurry the night before.

WHAT CHANGED The deck is now often the first read Over the last several years the sequence of a financing conversation has shifted for many small businesses: a short deck or a one-page summary is sent first, and the full document follows only if there is interest. This changes the deck's function from a presentation aid to a screening document, which has one practical consequence worth acting on. A screening document must be self-contained — readable without a presenter — which means the slides need enough text to stand alone, contrary to presentation advice written for the case where you are in the room. Prepare a version of the deck that reads without you. As of September 2026.

DO THE MATH One page, seven blocks: the word budget, computed A summary of one page is a real constraint, and it is worth converting it into a number you can watch while writing. A page of 11-point text with normal margins and single spacing holds roughly 600 words of prose. Subtract a heading and some white space between blocks, and the working budget is about 560 words. Divide it across the seven blocks in the proportions of Table 3.1: 60 + 80 + 90 + 90 + 70 + 100 + 80 = 570 words. That is your target, with ten words of slack. Now the practical test. At an average of 17 words per sentence — a reasonable length for business prose read aloud — 570 words is about 34 sentences. Block six, the numbers, gets 100 words, which is about six sentences to carry revenue, margin, net result, the horizon and the driver. Six sentences for the entire financial argument is the compression this exercise demands, and it is why the summary is written last: you cannot compress what you have not finished computing. If block six needs twelve sentences, the model is not yet simple enough to explain, which is information about the model rather than about the summary.

3.10 These Two Sections in Three Situations

3.10.1 Solo

Your summary's block five, the team, is one sentence about you and one about the gap. Do not pad it. The gap sentence — who covers the work if you are unavailable, and what happens to client commitments — is the one every reader is silently asking about, and answering it directly is worth more than a paragraph about your experience.

For the company section, the facts of record are short and the check against the register takes five minutes. Do it anyway.

3.10.2 Small team

Block five names two or three people with one relevant fact each, and states the gap honestly. The company section carries the cap table and the shareholder agreement question, which for a small team is the single most consequential item in the section — its absence is a genuine risk and belongs in chapter 9's register.

Write the summary yourself even if others wrote sections of the plan. A summary assembled from contributed paragraphs reads exactly like one.

3.10.3 Established business

The summary gains one element the other two do not have: a line reconciling the forecast with actual results. "Revenue has grown from $4.1 million to $5.6 million over three years; the forecast assumes 9% growth in year one, below the historical rate, because the second facility does not open until month nine." That single sentence pre-empts the first question a lender will ask.

The company section becomes longer, because there are more facts of record — subsidiaries, multiple registrations, more directors — and each of them must match its own register.

TOOLS OF THE TRADE What the summary needs from your document tools Three capabilities matter for this chapter specifically, independent of which product provides them. You need a visible word count while writing, because the block budgets of Table 3.1 are only useful if you can see the number as you type. You need the ability to keep the summary and the deck in one source, or at minimum a checklist that reconciles the figures between them, since divergence between the two is the failure mode of §3.9.3. And you need a register extract you can attach, obtained from the official register rather than retyped, because a retyped extract reintroduces exactly the error the extract exists to prevent. Chapter 13 names the products and their pricing models, dated.

Costly Mistakes

1. Writing the summary first and delivering that version. It compresses your intentions rather than your findings, and it anchors the rest of the plan to a number chosen before the arithmetic was done.
2. Omitting the numbers from the summary. A reader who reaches the end of a summary without seeing revenue, margin and the ask concludes either that the numbers are weak or that the writer did not think they mattered. Both conclusions are expensive.
3. Opening with a mission statement. Six seconds of the reader's attention spent learning nothing, at the exact moment their attention is highest.
4. Stating a legal name that does not match the register. It is the first thing a careful reader verifies, it takes them four minutes, and failing it colors their reading of everything after.
5. Listing values that have never cost anything. Every reader who has worked in a business knows what a values list looks like when it is decoration, and the recognition transfers to the rest of the document.
6. Presenting history as narrative rather than dated milestones. A reader cannot audit a story. They can audit "March 2024: first multi-building contract, three buildings, $1,840 per month."
7. Letting the deck say something the document does not support. It usually happens on a slide added late, and it is the single fastest way to lose a reader who has read both.

Key Takeaways

• The executive summary exists so a reader can make a preliminary decision without reading the plan, and that purpose settles its length, its content and its tone.
• Seven blocks are required, and block six — the numbers — is the one most often omitted and the one whose omission reads as evasion.
• A one-page summary is about 560 usable words, which is roughly 34 sentences, of which the entire financial argument gets about six.
• The opening sentence should contain who the business is, what it does for whom, and one verifiable fact; the four common openings contain none of those.
• The substitution test — put a competitor's name in the sentence and see whether it becomes false — removes most of the fog from missions, visions and opening lines.
• Nine facts of record belong in the company section, and all nine must match a current register extract, which takes ten minutes to check and prevents the cheapest possible loss of credibility.
• A mission is testable in the present, a vision rules something out in the future, and a value is a behavior that costs money to maintain.
• History belongs as dated milestones with numbers, including the setbacks, because a smooth ascent is not believable and readers know it.

Your Action Plan

1. Write the nine-sentence scratch file before drafting the summary. Deliverable: one sentence copied from each of the nine sections of your plan. Timing: ten minutes. Verification criterion: the nine sentences, read in order, form an argument rather than a list of topics.
2. Write block six first, taking every figure from the statements. Deliverable: about 100 words carrying revenue, gross margin, net result, the horizon and the operational driver. Timing: fifteen minutes. Verification criterion: every figure in the block matches the financial section exactly.
3. Apply the substitution test to every mission, vision and value in your draft. Deliverable: a revised set in which each statement becomes false when a competitor's name is substituted. Timing: thirty minutes. Verification criterion: at least one value has an identifiable cost attached to it.
4. Pull a current extract from the register and check the nine facts of record. Deliverable: the extract saved as an appendix, and any correction applied to the draft. Timing: fifteen minutes. Verification criterion: all nine items match the extract, character for character on the legal name.
5. Rewrite your history as dated milestones, including one setback. Deliverable: four to eight lines, each with a date, an event and a number where one exists. Timing: forty minutes. Verification criterion: a reader could verify at least half the lines from documents you already hold.
6. Read the finished summary aloud, once, at speaking pace. Deliverable: the summary, revised where you stumbled. Timing: ten minutes. Verification criterion: no sentence required a second breath, and nothing made you wince.

In La Machine

Every route named in this section was verified in the product's route inventory on 2026-09-09 (La Machine, 2026a), and what each page holds is described in the product's user guide (La Machine, 2026b).

• The company's own record. crm/organizations and crm/organization-profile hold organizations and their profiles; crm/me/details holds the business's own details. These are where the facts of record live inside the product, and keeping them current is what makes them usable in a document.
• Brand elements the summary and the deck both use. business/branding holds the branding, and business/assets/presentations holds decks, with business/assets/presentations/:id/edit for editing one.
• Where a distributable summary can live. business/assets/books holds long-form assets; a plan summary prepared for partners can be held alongside them for a business that distributes documents.
• Decisions about identity. A question such as which trade name to use, or whether to restate the mission, is exactly the kind of item arbitrages exists for.
• What it does not do. La Machine does not draft the summary, does not check a legal name against a public register, and does not hold a cap table. It has no mission-statement editor and no document assembly. The register extract is obtained from the register, and the summary is written in the tools of chapter 13.

Self-Check

1. State the single job of the executive summary, and derive from it the answer to the length question in one sentence.
2. Name the seven blocks in order. Which one is most often omitted, and what does a reader conclude from its absence?
3. Apply the substitution test to this sentence: "We are committed to delivering exceptional value through innovative solutions." What does the test tell you, and what would you write instead?
4. List five of the nine facts of record. Explain why a mismatch on the legal name costs more credibility than its objective importance warrants.
5. Your business is three months old and has no history. Name the two things that take the place of a company history in the plan, and give the form each should take.

Chapter References

Collins, J. C., & Porras, J. I. (1996). Building your company's vision. Harvard Business Review, 74(5), 65–77.

Corporations Canada. (2026). Federal corporation information and annual return requirements. Innovation, Science and Economic Development Canada. Retrieved September 9, 2026.

Civil Code of Québec, CQLR c CCQ-1991. LégisQuébec, consolidated version updated to April 7, 2026. Read September 9, 2026.

La Machine. (2026a). Documentation — route reference inventory (211 routes: 191 Core, 20 Pilotage paths). Internal product documentation, read September 9, 2026.

La Machine. (2026b). Guide utilisateur [User guide] (FAQ: CRM and contacts; knowledge assets). Internal product documentation, read September 9, 2026.

Registraire des entreprises du Québec. (2026). Le numéro d'entreprise du Québec (NEQ) and Déclaration de mise à jour annuelle (the NEQ has ten digits and is assigned on registration; every registered enterprise must file an annual updating declaration). Retrieved September 9, 2026.

Sahlman, W. A. (1997). How to write a great business plan. Harvard Business Review, 75(4), 98–108.

Chapter 4 — Market Analysis

This is the section where most plans lose their reader, and they lose them in a specific way: by proving something nobody doubted. The market for building maintenance is large. The market for accounting services is large. The market for almost anything a small business sells is large, which is why establishing that fact persuades no one. What a market section must establish is narrower, harder and far more convincing — that a particular, countable set of customers exists within reach of this business, that they have a problem worth paying to solve, and that the business can get to enough of them to make the forecast work.

What You'll Be Able to Do

• Define your industry and its boundaries in one sentence, with the classification code that lets a reader find comparable data.
• Document a trend with evidence rather than with an adjective, and know what counts as evidence.
• Size a market from the bottom up, in five steps with the arithmetic written out, and explain why the top-down version is not accepted.
• Describe segments you can actually reach, distinguishing them from segments that merely exist.
• Map competition honestly, including the substitute that wins most often — the customer doing nothing.
• Name the regulation that applies to your activity and what compliance costs, rather than discovering it after the plan is approved.

Why It Matters

For a lender, the market section answers a question about durability: if this business loses its three largest customers, is there a pool of replacements it can reach? For an investor, it answers a question about ceiling: how large can this become before it runs out of buyers? For you, it answers the question that determines the entire forecast: how many customers can we realistically add per month, and at what cost?

There is also a defensive reason to write this section well. A market section built from bottom-up arithmetic is the hardest part of a plan to argue with, because every step is visible and each step can be checked. A market section built from a national industry figure with a percentage applied to it invites exactly one response, which is that the percentage is arbitrary — and once a reader has that thought about your market section, they carry it into your financial section.

In This Chapter

• 4.1 What a Lender or Investor Actually Reads Here
• 4.2 Defining the Industry and Its Boundaries
• 4.3 Trends: Naming Them With Evidence, Not Adjectives
• 4.4 Sizing the Market Without Inventing It
• 4.5 Target Segments and the Customer You Actually Serve
• 4.6 Competition: Direct, Indirect and the Status Quo
• 4.7 Regulation, Licensing and What It Costs to Comply
• 4.8 Sources: What Counts as Evidence in a Market Section
• 4.9 Writing the Section in Six Pages
• 4.10 Market Analysis in Three Situations

4.1 What a Lender or Investor Actually Reads Here

Before writing, it is worth knowing what a professional reader extracts from this section, because it is less than most writers imagine and different from what they expect.

They extract four things. How many potential customers are within reach, expressed as a count rather than a dollar value. What the customer currently does instead, which tells them how hard the sale is. Who else is competing for the same customers, and on what basis. And whether anything in the regulatory environment could stop the business or add cost.

Everything else in the section — the industry history, the technology trends, the demographic shifts — is context that supports those four extractions or is ignored. Writing with that in mind changes the proportions: the count and the current alternative deserve pages, and the industry background deserves a paragraph.

KEY CONCEPT A market section is an argument about demand, not a list of statistics The section has a claim, and the claim is: enough customers who will buy this, at this price, exist within reach of this business, and we can name them by category and count them. Every statistic in the section either supports that claim or does not belong. This framing has an immediate editing consequence: any statistic you cannot connect to the claim in one sentence should be deleted, however impressive it is. A national market size of eleven billion dollars supports nothing if the business can serve forty customers, and a reader who sees the eleven billion knows that.

4.1.1 The count is the deliverable

If this section produces one number that the rest of the plan uses, it is the count of reachable customers. The sales forecast in chapter 7 will be built as a share of that count, converted at a rate you can defend, which means an unreliable count makes the whole forecast unreliable.

Get the count right, show how you got it, and the rest of the section becomes supporting material.

4.2 Defining the Industry and Its Boundaries

4.2.1 Why the boundary matters

Before you can size a market you must say what is inside it and what is outside it, and this is harder than it sounds because the natural instinct is to draw the boundary generously. A business that describes itself as being in "facility services" has drawn a boundary that includes cleaning companies, security firms, landscapers and elevator maintenance contractors, none of whom it competes with. The generous boundary produces a large number and destroys the section's credibility in the same move.

Draw the boundary at the point where a customer would consider you and a competitor as alternatives for the same purchase. That is the operative definition, and it is usually narrower than the industry classification.

4.2.2 Using a classification code

Give the standard classification code for your activity, because it is what lets a reader find comparable data and because using one signals that you looked.

In Canada, the relevant system is the North American Industry Classification System (NAICS), maintained by Statistics Canada, currently at NAICS 2022 Version 1.0. Codes run from two digits (sector) to six digits (industry). Group 5617 covers services to buildings and dwellings; within it, code 561790 covers "other services to buildings and dwellings," excluding pest control, janitorial, landscaping and carpet cleaning, with illustrative examples that include ventilation duct cleaning and building exterior cleaning (Statistics Canada, 2026b). Innovation, Science and Economic Development Canada publishes Canadian Industry Statistics by NAICS code, which is where establishment counts and financial benchmarks for a code can be found (Innovation, Science and Economic Development Canada, 2026b).

Two cautions apply. A classification code describes what an establishment mainly does, and many small businesses genuinely straddle two codes; say which one you file under and which one describes your competitors, if they differ. And the data available at a six-digit code is often not available at the geography you care about, which is the subject of §4.4.

DEFINITION Industry, market, segment, addressable market An industry is a supply-side category: a set of businesses that produce similar things, defined by classification systems such as NAICS. A market is a demand-side category: a set of buyers with a shared need, which may cut across several industries. A segment is a subset of a market whose members share a characteristic that changes how they buy — size, sector, geography, buying process — and the test of a useful segment is that you would sell to it differently. The addressable market is the portion of the market that this specific business could serve given its geography, capacity, licenses and offer; it is always smaller than the market, and stating it honestly is what makes the section credible.

4.2.3 Geography is part of the boundary

For most small businesses, geography is the binding constraint on the market, and it should be stated in operational terms rather than administrative ones. "Within a forty-five-minute drive of our depot" is a more useful boundary than "the greater metropolitan area," because it is the constraint that actually governs which jobs are profitable.

Where the business is not geographically bound — software, professional services delivered remotely — the boundary is something else: a language, a regulatory regime, a platform, a certification. Name it.

4.3 Trends: Naming Them With Evidence, Not Adjectives

4.3.1 The adjective problem

Plans are full of sentences like "the industry is experiencing rapid growth driven by increasing demand for sustainable solutions." The sentence contains three adjectives and no facts, and a reader discounts it entirely — not because it is false, but because it is unfalsifiable.

A trend statement that works has four parts: what is changing, by how much, over what period, and the source. "The number of commercial buildings in our service area with a mechanical systems service contract rose from X to Y between 2020 and 2025, according to Z" is a trend. Everything shorter is an adjective.

4.3.2 How many trends to name

Three is usually right, and each should be one that changes something about your business rather than one that is generally true. A trend belongs in a plan only if you can complete this sentence: because of this, we will do the following differently.

Trends that change nothing you do are context, and context belongs in one paragraph or in an appendix.

4.3.3 When a trend is really a fashion

The distinction that matters, and the one that costs businesses money, is between a trend and a fashion. A trend has a mechanism: something structural changed — a regulation, a cost, a demographic, a technology's price — and the effect follows from it. A fashion has adoption but no mechanism: people are doing it because other people are doing it.

The test is to ask what would have to happen for the trend to reverse. If the answer is "the underlying cost would have to go back up" or "the regulation would have to be repealed," it is a trend. If the answer is "people would have to lose interest," it is a fashion, and building a plan on it is a decision you should make deliberately rather than accidentally.

CASE IN POINT Studio Kavan: the trend that was a fashion This case is fictional; the company, the people and the figures are invented. Studio Kavan, a six-person digital agency, built its 2025 plan around a service line that a great many of its prospects were asking about. The market section said the demand was growing rapidly, cited three industry articles, and forecast that the line would reach 40% of revenue within two years. The agency hired two people for it. Applying the reversal test to the trend would have taken ten minutes. What would have to happen for the demand to stop? Nothing structural — no cost had changed, no regulation required it, no technology had become cheaper. Clients were asking because their competitors were asking. Fourteen months later the requests had largely stopped, the two hires were redeployed, and the agency's actual growth had come from a service line the plan mentioned in one sentence. The lesson the founders drew was not that they should have refused the work; the work was profitable while it lasted. It was that they had resourced it as a trend rather than as a fashion, which meant permanent hires against temporary demand. A plan that had named it a fashion would have served the same clients with contractors.

4.4 Sizing the Market Without Inventing It

This is the technical core of the chapter.

4.4.1 Top-down and why it fails

Top-down sizing starts with a large published number and applies percentages to it: the national market is $4.2 billion, our province is 22% of it, our region is 8% of that, and we expect 3% share, therefore our market is $2.2 million. Every step is a multiplication and none of the percentages is defended.

The method fails for two reasons. The percentages are almost always assumptions dressed as data, and a reader knows it. And the result is insensitive to whether the business can actually reach anybody, which is the question the section exists to answer.

4.4.2 Bottom-up, and why it is believed

Bottom-up sizing starts with a countable population and applies filters that you can defend one at a time, arriving at a count of reachable customers, which is then multiplied by a price you actually charge.

Each filter is a claim that can be discussed separately, which is precisely why the method is credible: a reader who disagrees with one filter can adjust it and see the effect, rather than rejecting the whole number.

METHOD Bottom-up sizing in five steps Step one: count the population. Find a countable universe from an official or verifiable source — establishments in a NAICS code and geography, properties in a municipal assessment roll, license holders in a registry, households in a census subdivision. Write down the source and the date. Step two: apply the qualifying filter. Which members of that population have the characteristic that makes them a possible customer — the size, the sector, the equipment, the situation? State the filter and where the proportion comes from, including when it comes from your own observation. Step three: apply the reachability filter. Which of those can you actually serve, given geography, capacity, licensing and language? This is the step top-down sizing omits entirely. Step four: apply the availability filter. Which of them are not locked into a long-term arrangement with somebody else, and how many come up for renewal in a year? A market of 1,400 customers who all signed five-year contracts last year has an annual opportunity of 280. Step five: multiply by your actual price. Use the price you charge, not an industry average, and state whether it is annual or per transaction. The deliverable is a five-line table showing each step, its value, its source and the running count. The verification criterion is that a reader who changes one filter can recompute the result without asking you for anything.

DO THE MATH Nordec's addressable market, counted from the bottom The figures below are fictional and belong to the running case; the arithmetic is the point. Step one — the population. The municipal assessment rolls of the fourteen municipalities within a forty-kilometre radius of the depot list 4,800 commercial and institutional buildings. Source: the assessment rolls, consulted in the year stated in the plan. Step two — the qualifying filter: size. Buildings between 2,000 and 9,000 square metres are large enough to need scheduled maintenance and too small to employ a technician. That band is 41% of the roll, taken from the roll's own area field: 4,800 × 0.41 = 1,968 buildings. Step three — the reachability filter: no internal technician. From our own record of the 61 buildings we have quoted in three years, 78% had no internal technician: 1,968 × 0.78 = 1,535 buildings. Step four — the availability filter: usage categories we serve. We serve offices, multi-tenant retail, light industrial and clinics, which are 91% of the remaining set: 1,535 × 0.91 = 1,397, which we state as ≈ 1,400 buildings. The value. Our average contract is $685 per month, or $8,220 per year: 1,400 × $8,220 = $11,508,000 of annual contract revenue in the addressable market. The share this plan actually claims. The forecast takes buildings under contract from 34 to 52 in three years. That is 52 ÷ 1,400 = 3.7% of the addressable market, worth 52 × $8,220 = $427,440 of contract revenue. Two observations make this useful rather than decorative. The plan claims 3.7% of a market it counted, which is a far easier claim to accept than 0.9% of a provincial figure nobody can verify. And the fourth filter is the one that matters most for the sales forecast: if most of those 1,400 buildings are locked into contracts with competitors, the annual opportunity is the renewal flow rather than the whole set — which is exactly the question chapter 5 has to answer with a conversion rate.

Figure 4.1 draws that arithmetic as a funnel. Each band is one step of the method, labeled with the filter applied, the proportion, and the count that survives it; the width of each band is proportional to the count, so the compression from population to reachable customers is visible rather than merely stated. The last two bands show the conversion into dollars and the share this plan actually claims, which is the pair of numbers a reader will quote back to you.

Figure 4.1 — Original figure. Bottom-up sizing as a funnel: a countable population, three defended filters, the reachable count, and the share the plan actually claims.

4.4.3 What to do when the population cannot be counted

Sometimes there is no countable universe: a new category, a diffuse consumer market, a business-to-business market with no register. Three substitutes work, in descending order of strength.

The first is a proxy population: count something correlated that can be counted. A business selling to restaurants can count restaurant licenses; a business selling to landlords can count rental units.

The second is a sample extrapolation: measure a small area completely and scale it by a known ratio. Count every possible customer in one municipality, then scale by population or by establishment counts.

The third is a demand-side estimate: how many units of the thing are consumed per year in the geography, from import data, association figures or supplier statements. Weakest of the three, and worth stating as such.

In all three cases, say which one you used and why the direct count was unavailable. A reader forgives an estimate that announces itself; they do not forgive one that pretends to be a count.

4.4.4 Four arithmetic errors that recur

The funnel is simple arithmetic, which is exactly why it is worth checking. Four errors recur often enough to be worth naming.

Mixing a stock with a flow. A count of buildings is a stock: it is how many exist at a moment. A count of contracts signed per year is a flow. Multiplying a stock by an annual price gives an annual market value, which is correct; multiplying a flow by an annual price gives something that has no meaning. State for every number whether it is a stock or a flow, and never multiply two flows together.

Double counting across segments. A business that serves offices and also serves property managers will, if it is not careful, count the same building twice — once as an office and once as part of a managed portfolio. Segments must partition the population, meaning every member belongs to exactly one. If two segments overlap, either merge them or define the overlap explicitly and subtract it.

Confusing the total market with the annual opportunity. A market of 1,400 buildings whose contracts run three years has an annual opportunity of roughly 467 buildings, not 1,400 — the rest are unavailable this year at any price. This is the fourth filter of the method, and it is the one most often skipped, because it shrinks the number. It also happens to be the filter that makes the sales forecast realistic.

Applying a percentage to a filtered set as though it applied to the original. Having filtered to 1,535 buildings, it is tempting to apply the next percentage to the 4,800. Every filter applies to the output of the previous one, which is why the funnel is written as a running count with each line showing what it was computed from.

A fifth habit prevents all four: write the units beside every number in the working spreadsheet — buildings, buildings per year, dollars per building per year, dollars per year. Unit errors become visible the moment the units of a multiplication do not resolve.

4.5 Target Segments and the Customer You Actually Serve

4.5.1 A segment must change how you sell

Segmentation is only useful if the segments behave differently. If you would send the same message, at the same price, through the same channel, to two groups, they are one segment with two names.

The characteristics that usually produce real segments in a small business are size, sector, buying process, and urgency of the need. The characteristics that usually produce fake segments are demographic categories that do not affect the purchase, and internal categories that reflect how you organize rather than how they buy.

4.5.2 The beachhead

Name the segment you will serve first and say why. A plan that targets four segments simultaneously is a plan with four go-to-market motions, four sets of proof material and four sales cycles, which no small business has the capacity for. → see Book A1 (From Idea to Business Model) for the beachhead selection method.

The criteria that matter for a beachhead are that the segment is reachable through one channel, that its members talk to each other, and that a win in it is visible to the next prospect.

4.5.3 The customer profile

For each segment you keep, write a short profile: who decides, who else is involved, what triggers the purchase, what they are comparing you against, what they are afraid of, and what evidence changes their mind. Six lines each.

That profile is what the go-to-market section of chapter 5 consumes, and writing it here saves that section from being invented.

4.5.4 The three segments of the running case, written out

Abstract segmentation advice is less useful than a worked example. The three segments below belong to the fictional running case, and they illustrate the test of §4.5.1: each one is sold to differently, which is what makes them separate segments rather than three names for the same buyer.

Segment one — the independent owner-operator building. A single owner holds one or two buildings, usually a mixed retail-and-office property of 2,000 to 4,000 square metres, and manages them personally. The decision-maker is the owner and there is nobody else involved. The purchase is triggered by an incident — a failure that cost a tenant a day of business — rather than by a planning cycle. They are comparing us against calling trades individually. What they are afraid of is being locked into a contract they cannot leave. What changes their mind is a twelve-month term with a sixty-day exit, which removes the fear at almost no cost to us. This segment is 457 buildings of the 1,400 and is the current core of the business.

Segment two — the small managed portfolio. A property management firm holds five to twenty buildings for third-party owners. The decision-maker is a portfolio manager, but the owner must approve anything above a threshold, and the building superintendent is consulted informally and can veto in practice. The purchase is triggered by a renewal cycle, usually annual, on a predictable date. They are comparing us against two or three named competitors, on price and on response time. What they are afraid of is a contractor who makes them look bad to the owner. What changes their mind is a monthly one-page report they can forward without editing. This segment is 612 buildings, is the beachhead of the plan, and is the reason the reporting feature exists.

Segment three — the institutional building. A school board, a municipality or a health institution, with a formal procurement process, a tender, and requirements around insurance, safety certification and language. The decision is made by a committee against published criteria. The cycle is six to eighteen months. This segment is 331 buildings and the plan explicitly defers it, because winning one requires a bid capability the business does not have and a bonding capacity it cannot currently obtain. Naming a segment and declining it is a stronger statement than including it, because it shows the reader that the count and the forecast were built with a filter applied.

Notice that the three counts — 457, 612 and 331 — sum to 1,400, which is the reachable count from §4.4.2. Segments that do not sum to the population have a double-counting problem or a missing residual, and a reader who adds them up will find it.

4.6 Competition: Direct, Indirect and the Status Quo

4.6.1 The three kinds

Direct competitors sell the same thing to the same buyers. Name three to six by name, with what they do well. A plan that claims no direct competitors is almost always wrong, and the claim reads as a failure to look rather than as an opportunity.

Indirect competitors and substitutes solve the same problem differently: the customer who hires a general handyman rather than a maintenance contractor, or who buys software instead of a service. Substitutes put a ceiling on the price you can charge, because a buyer measures you against them as well as against your direct rivals (Porter, 2008), which is why they belong in the section.

The status quo is the alternative that wins most often, in every market: the customer keeps doing what they do now. Naming it explicitly, and stating what it costs the customer to keep doing it, is the single most useful paragraph in most competition sections, because that cost is the argument your sales process will actually make.

The paragraph below introduces Table 4.1, which presents the three kinds of competition. Its columns give what each one is, what to write about it in the plan, and the mistake most often made with it. Note that the row for the status quo is the one most plans omit entirely, and it is the row that most directly feeds the sales argument.

Table 4.1 — Direct, indirect and the status quo: what to write about each

4.6.2 What to write about a competitor

Write three things about each named competitor: what they genuinely do well, who they serve best, and where the gap is that you fill. That order matters. A competitive section that opens with weaknesses reads as marketing; one that opens with strengths reads as analysis, and a reader trusts analysis.

Avoid the feature matrix in which your column is entirely checkmarks. Nobody has ever believed one, and it costs more credibility than the space it occupies.

4.6.4 Pricing the status quo

The status quo is the competitor you must beat most often, and it is the only one whose price you can compute rather than guess — because the customer is paying it today, in their own units.

The computation has three parts. What they spend directly: the invoices they already pay for the piecemeal alternative. What they spend indirectly: the hours their own people spend arranging, supervising and following up, valued at a loaded rate. What it costs them when it fails: the incidents per year, times the cost of an incident in their business.

For the running case, a building arriving from the piecemeal arrangement reports eleven unplanned interventions in its last twelve months. If the average intervention invoices $340, direct spending is 11 × $340 = $3,740. If each one consumes an hour and a half of the manager's time at a loaded $58 per hour, that is 11 × 1.5 × $58 = $957. If two of the eleven caused a tenant complaint that took a further three hours to settle, that is 2 × 3 × $58 = $348. The status quo therefore costs about $5,045 a year, against a contract at $8,220 — which means the contract is not cheaper, and the argument cannot be a price argument. It has to be an argument about the two incidents, the predictability, and the hours returned.

That conclusion is worth more than any competitor comparison in the section, and it is the reason to do the arithmetic: it tells the sales process in chapter 5 what argument it is actually allowed to make.

4.6.3 Where competitive information legitimately comes from

Published prices, public websites, job postings, regulatory filings, industry directories, customer reports and your own lost-deal record are all legitimate sources. Say where a claim came from when it is not obvious.

What is not legitimate is worth stating plainly, because the boundary occasionally gets crossed in the enthusiasm of research: obtaining information through misrepresentation, inducing someone to breach a confidentiality obligation, or using a competitor's confidential information obtained from a former employee. Beyond the legal exposure, an employee under a duty of loyalty is bound not to use confidential information obtained in the course of their work — a duty that continues for a reasonable time after the employment ends. → see Book H5 (Intellectual Property) and Book H1 (Contracts and Commercial Agreements).

4.7 Regulation, Licensing and What It Costs to Comply

4.7.1 What belongs here

The market section is where regulation appears, for a simple reason: regulation determines who may serve this market and at what cost, which makes it a market fact rather than a legal one.

Four things belong. The licenses or certifications required to perform the activity, with who issues them. The rules that govern how you sell — advertising, consumer protection, language, privacy, anti-spam. The rules that govern how you operate — safety, environmental, sectoral. And the cost of compliance, in dollars and in hours, because that cost belongs in the financial model.

4.7.2 Regulation as a barrier and as a moat

Regulation cuts both ways in a plan, and saying so demonstrates that you understand your market. A licensing requirement is a cost to you and a barrier to the next entrant. A certification that took you eighteen months to obtain is eighteen months a competitor must also spend.

Where regulation is a genuine barrier to entry in your market, say so here and quantify it; it is one of the few defensible answers to the question of what stops somebody else from doing this.

4.7.3 Pricing compliance so it reaches the model

Compliance is a cost, and the failure mode is not that businesses ignore it but that they mention it in the market section and never carry it into the financial section. Price it in four buckets, and give each bucket a line in the assumption log.

One-time costs to enter: licenses, certifications, initial training, initial equipment required by a rule rather than by the work. These become capital or start-up costs in chapter 7.

Recurring costs to stay: renewals, annual filings, mandatory insurance, periodic inspections, continuing education. These become fixed operating costs.

Per-transaction costs: anything a rule requires you to do on every sale — a document to provide, a consent to obtain, a record to keep. Multiply by the transaction count; a two-minute obligation on four hundred transactions a year is thirteen hours.

The cost of the person who holds it: somebody must know the rules, track the renewals and answer the inspector. In a small business this is the owner, and the hours belong in the capacity arithmetic of chapter 6 rather than being treated as free.

Two habits make this reliable. Verify each requirement at the issuing body's own page rather than at an intermediary, and record the date you read it. And put the renewal dates in the calendar the day you write this section, because the most common compliance cost in a small business is not the fee — it is the penalty for a filing that everybody assumed somebody else was watching.

QUÉBEC & CANADA NOTE Where to find the official count and the official rule Three sources do most of the work for a Canadian market section. Statistics Canada maintains the North American Industry Classification System, currently NAICS 2022 Version 1.0, and publishes establishment counts and business data by code and geography; group 5617 and code 561790 in the running case were read there on September 9, 2026. Innovation, Science and Economic Development Canada publishes Canadian Industry Statistics, which turns those codes into establishment counts, financial benchmarks and employment figures, and it publishes Key Small Business Statistics — whose 2025 edition, page modified April 23, 2026, is the source for the survival rates quoted in chapter 1 (Innovation, Science and Economic Development Canada, 2026c). The Institut de la statistique du Québec provides the provincial equivalents (Institut de la statistique du Québec, 2026). For regulation, the applicable rules depend entirely on the activity: consumer-facing sales engage the Consumer Protection Act, CQLR c P-40.1; electronic marketing engages Canada's anti-spam legislation; the handling of personal information engages the Act respecting the protection of personal information in the private sector, CQLR c P-39.1; and language obligations engage the Charter of the French language, CQLR c C-11. Name the ones that apply to you, with their official titles, and give the cost. → see Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements), Book H3 (Anti-Spam and Telemarketing Compliance) and Book H2 (Privacy and Personal Information).

4.8 Sources: What Counts as Evidence in a Market Section

4.8.1 The four types, ranked

Not all evidence is equal, and a reader ranks it whether or not you do.

Primary evidence that a customer acted is strongest: a signed contract, a deposit, a purchase order, a completed pilot with a result. It proves demand rather than describing it.

Primary evidence that a customer said something is next: interviews, surveys, quote requests, with the count and the date. Weaker than action, and still far stronger than published data, because it is about your customers.

Official statistics — statistical agencies, regulators, registries — are next: reliable, verifiable, and usually not specific enough to answer your question directly, which is why §4.4 uses them as a starting population rather than as an answer.

Industry and vendor publications are last: useful for context and for the direction of a trend, and never sufficient to support a claim your forecast depends on.

The paragraph that follows introduces Table 4.2, which presents these four evidence types with what each one can prove, what it cannot prove, and how to cite it. The right-hand column is the practical one: it says exactly what a reader needs to see beside the claim.

Table 4.2 — Four kinds of evidence: what each proves, and how to cite it

THE RULE Every market figure carries its source and its year, or it is removed There is no third option, and the rule is stricter than it sounds: a figure needs the publisher, the publication, the reference year of the data, and the date you retrieved it. The reference year and the retrieval date are different things and both matter, because statistical publications lag: a 2026 publication commonly reports 2023 or 2024 data, and a reader who cannot tell which will assume the least favorable interpretation. Where the figure is your own estimate, say "our estimate" and give the method in one clause. An estimate that announces itself is evidence; an estimate dressed as a statistic is a defect.

4.8.2 Doing primary research on a small budget

The most persuasive market evidence for a small business is nearly always the cheapest to obtain: talk to twenty potential customers and count what they say. Twenty conversations, with the questions written in advance and the answers recorded in a table, produce a market section that no published report can match, because it is about your specific buyers.

Record four things per conversation: what they do today, what it costs them, what would make them change, and whether they asked what you charge. The fourth is the useful one. → see Book A1 (From Idea to Business Model) for the interview method itself.

4.9 Writing the Section in Six Pages

The market section fits in six pages, allocated as follows.

Page one: the boundary and the count. What industry, what geography, what classification code, and the headline: how many reachable customers, from what population, by what filters. This page carries the funnel table.

Pages two and three: the segments. Two or three segments, each with its profile, its size within the count, and why it is or is not the beachhead.

Page four: the trends. Three trends, each with what is changing, by how much, over what period, from what source, and what you will do differently because of it.

Page five: the competition. Three to six direct competitors, the substitutes, and the status quo with its cost to the customer.

Page six: regulation and evidence. What licenses apply, what compliance costs, and a short statement of what your evidence is and what it does not cover.

WATCH OUT The hockey-stick market chart with no source under it A chart showing a market growing steeply, with axis labels but no source line, is worse than no chart. It occupies half a page, it makes a claim the plan does not support elsewhere, and it triggers the one question you least want: where did this come from? If you have a real source with real data, cite it under the chart, including the reference year. If you do not, the space is better spent on the funnel table from §4.4, which nobody can dismiss because every step is visible.

WHO DOES THIS Who gathers the evidence, and who verifies it In a solo business, you do both, and the countermeasure for the obvious conflict of interest is temporal: gather the evidence one week, and check it the next with the specific question would a stranger accept this? In a small team, the person who sells gathers the customer evidence, because they are already having the conversations, and somebody else checks the arithmetic of the funnel. Separating the two roles matters more than who fills them. In an established business, the market work is often assigned to someone junior, which is fine, with one condition: the person who signs the plan must be able to reconstruct the funnel from memory. If they cannot, they will not be able to defend it in a meeting, and the meeting is where it matters. External help is worth buying for one thing specifically: access to paid industry data you cannot otherwise obtain. It is rarely worth buying for the interviews, which are more valuable when conducted by the person who will later sell to those people.

4.10 Market Analysis in Three Situations

4.10.1 Solo

Your market section is three pages, not six, and its strongest element will be primary evidence rather than published data. Twenty conversations and a countable population from a public register will outperform any report you could buy.

The specific risk for a solo business is defining the market as everyone who could possibly need the service. Apply the reachability filter honestly: how many customers can one person actually serve, and what does that make the relevant population?

4.10.2 Small team

You have enough people to segment and enough constraint that you must. Use the section to make the beachhead decision explicit, because in a small team the segment question is usually being answered differently by different people without anyone noticing.

4.10.3 Established business

Your own customer data is the best market evidence you have, and it is usually underused. Win rate by segment, average contract value by sector, churn by customer size — these are facts about your market that no external source can supply, and they make the section unusually strong.

The risk at this size is describing the market as it was when the business entered it. Re-run the funnel with current figures rather than updating last year's percentages.

TOOLS OF THE TRADE What a market research toolkit must include Independent of any product name, the work of this chapter requires five capabilities. You need access to the official statistical portals for classification codes, establishment counts and geography, which are public and free. You need a way to collect and count interviews, which can be a spreadsheet as long as it records the date, the person's category and the four answers. You need a place to keep evidence with its citation attached, so that a figure and the date it was read never separate. You need a spreadsheet for the funnel, built so that changing one filter recomputes the count, because a reader will ask you to change one. And you need a record of competitors that is updated rather than rewritten, since the useful thing about competitive information is the change over time. Chapter 13 names products and their pricing models, dated.

WHAT CHANGED What is knowable about a small market has improved, unevenly Two things have moved in the last few years and they pull in opposite directions. Official statistical portals now publish more at finer geographies and with better documentation, which makes step one of the funnel easier than it was. At the same time, several sources of competitive information that small businesses relied on — public pricing pages, open directories, freely searchable listings — have become less complete, as more vendors move pricing behind a contact form. The practical consequence for a plan written now is that the population count is easier and the competitor price comparison is harder, so budget your research time accordingly and expect to obtain competitor pricing from customers rather than from websites. As of September 2026.

Costly Mistakes

1. Sizing top-down. A national figure with three percentages applied to it invites the reader to reject the percentages, and once they do, the entire section is gone. The funnel takes two hours and cannot be dismissed the same way.
2. Drawing the industry boundary generously. A boundary that includes businesses you do not compete with produces a larger number and a smaller amount of credibility, which is a bad trade in both directions.
3. Claiming there are no direct competitors. It is nearly always false, and the reader's conclusion is not that you found a gap but that you did not look.
4. Omitting the status quo. The alternative that wins most often is the customer doing nothing, and a plan that never names it has not thought about what its sales process must actually overcome.
5. Treating a fashion as a trend and resourcing it permanently. The cost is not the wasted marketing; it is the permanent hires made against temporary demand, which take a year to unwind.
6. Quoting a statistic without its reference year. Publications lag by two to three years; a reader who cannot tell which year the data describes will assume the interpretation least favorable to you.
7. Building a segment list that reflects your organization rather than the customer's buying behavior. Segments that you would sell to identically are one segment, and treating them as several multiplies the go-to-market work in chapter 5 for no benefit.

Key Takeaways

• A professional reader extracts four things from this section: the reachable count, what the customer does today, who else competes, and what regulation could stop or cost you.
• The section's deliverable is a count of reachable customers, because the sales forecast is built as a share of it.
• Draw the industry boundary where a customer would consider you and a competitor as alternatives for the same purchase, which is usually narrower than the classification.
• Bottom-up sizing in five steps — population, qualifying filter, reachability filter, availability filter, price — is credible precisely because each step can be argued with separately.
• A trend has a mechanism and a fashion has only adoption; the reversal test separates them in ten minutes and prevents permanent hires against temporary demand.
• Competition has three kinds, and the one most often omitted — the status quo — is the one that wins most often and feeds the sales argument.
• Evidence is ranked: a customer who acted beats a customer who spoke, which beats official statistics, which beats industry publications.
• Every figure carries its publisher, its publication, its reference year and the date you read it, because reference year and retrieval date are different facts and both matter.

Your Action Plan

1. Write the boundary sentence and find your classification code. Deliverable: one sentence defining what is in and out of your market, plus the NAICS code you file under. Timing: forty minutes. Verification criterion: the sentence excludes at least one adjacent business type that a generous boundary would have included.
2. Build the five-step funnel in a spreadsheet. Deliverable: a five-line table with population, three filters, the reachable count and the value at your actual price, each line carrying its source. Timing: two to three hours. Verification criterion: changing any one filter recomputes the result without manual editing.
3. Conduct twenty customer conversations and count the answers. Deliverable: a table with one row per conversation recording what they do today, what it costs them, what would make them change, and whether they asked your price. Timing: two to three weeks in parallel with other sections. Verification criterion: you can state the count that asked about price.
4. Name three trends and apply the reversal test to each. Deliverable: three trend statements with what changes, by how much, over what period, from what source, plus a one-line mechanism. Timing: ninety minutes. Verification criterion: each one names what you will do differently because of it.
5. Write the competitor set, strengths first. Deliverable: three to six named direct competitors, the substitutes, and one paragraph on the status quo with its cost to the customer. Timing: two hours. Verification criterion: each competitor entry begins with something they do well.
6. List the regulation that applies and price the compliance. Deliverable: the licenses and rules by official title, with the dollar and hour cost of compliance carried into the financial model. Timing: two hours, plus verification at the source. Verification criterion: each cost appears as a line in your assumption log.

In La Machine

Every route named in this section was verified in the product's route inventory on 2026-09-09 (La Machine, 2026a), and what each page holds is described in the product's user guide (La Machine, 2026b).

• Where the market work lands. business/marketing/companies holds companies, with business/marketing/companies/:id/edit for one of them, and this is where a counted population becomes a working list. The business explorer at business/marketing/business-explorer searches for businesses and imports them into that list, but it is not open to everyone: the page access rule allows the administrator and the reseller, and the menu shows the entry to the reseller alone, so an owner working under another role reaches it from the Explore Businesses button of the Companies page rather than from the menu. Its import checks no territory, no exclusion list and no consent, which means the consent work of Book H3 (Anti-Spam and Telemarketing Compliance) happens after the import, not during it. → see Book N4 (CRM, Contacts and Prospect Data), Chapter 9.
• Segments and audiences. business/marketing/segments and business/marketing/audiences hold segments and audiences, each with its edit route. A segment defined in this chapter becomes a segment there, and chapter 5 uses it.
• The contacts behind the count. business/marketing/contacts holds contacts, with import at business/marketing/contacts/import, and crm/clients holds clients.
• Interviews and evidence. /presse-papier holds notes that attach to a project; the interview log of §4.8.2 is a document you keep, and the tasks that produce it live at /taches.
• What it does not do. La Machine does not size a market, does not connect to statistical agencies, and holds no industry data. It will not compute your funnel, and it does not maintain a competitor database. The counting is done in the tools of chapter 13, from the public sources named in this chapter.

Self-Check

1. Give the operative test for where to draw an industry boundary, and apply it to a business you know: name one adjacent activity that falls outside.
2. Recite the five steps of bottom-up sizing. Explain in one sentence why step three is the one top-down sizing omits, and what its omission costs.
3. A plan states that its market is growing at 14% a year. List everything you would need beside that figure before you would accept it.
4. Apply the reversal test to a trend in your own market: what would have to happen for it to stop? Say whether your answer makes it a trend or a fashion.
5. Your competition section names four competitors and describes the weaknesses of each. Name two problems with that and give the fix for each.

Chapter References

Innovation, Science and Economic Development Canada. (2026b). Canadian industry statistics: Services to buildings and dwellings (NAICS 5617). Retrieved September 9, 2026.

Innovation, Science and Economic Development Canada. (2026c). Key small business statistics 2025 (page modified April 23, 2026). Retrieved September 9, 2026.

Institut de la statistique du Québec. (2026). Statistiques par secteur d'activité économique. Retrieved September 9, 2026.

La Machine. (2026a). Documentation — route reference inventory (211 routes: 191 Core, 20 Pilotage paths). Internal product documentation, read September 9, 2026.

La Machine. (2026b). Guide utilisateur [User guide] (FAQ: CRM and contacts; marketing modules; prospecting strategy). Internal product documentation, read September 9, 2026.

Porter, M. E. (2008). The five competitive forces that shape strategy. Harvard Business Review, 86(1), 78–93.

Statistics Canada. (2026b). North American Industry Classification System (NAICS) Canada 2022 Version 1.0, group 5617 (services to buildings and dwellings) and class 561790 (other services to buildings and dwellings). Retrieved September 9, 2026.

Chapter 5 — Offer and Go-to-Market Sections

These two sections carry the part of the plan that a lender quietly weights more heavily than most owners expect. The market section established that customers exist; these sections establish that you can reach them, at a cost you can name, in a time you can estimate. That is a different claim, and it is the one that most often turns out to be wrong — not because the offer was bad, but because nobody computed what it costs to acquire a customer or how long the sale actually takes.

What You'll Be Able to Do

• Describe an offer so that a stranger understands it, in one page, without a specification sheet.
• Convert a feature into a benefit and a proof line, so that every claim in the section is supported.
• Present pricing in a plan: what to show, what to justify, and what belongs in an appendix.
• Write the marketing plan section in two pages, with a budget that is derived rather than assumed.
• State a sales process as stages with conversion rates, and compute a realistic cycle length.
• Compute customer acquisition cost and payback period on your own numbers, and read what the result tells you.

Why It Matters

Three specific failures live in these sections, and each is expensive in a distinct way.

The first is the marketing budget set as a percentage of a revenue that does not exist yet. It produces a circular plan: revenue depends on marketing, marketing depends on revenue, and neither is anchored to anything. A reader who spots it stops trusting the forecast.

The second is the sales cycle assumed to be shorter than it is. A business that assumes six weeks and experiences fourteen does not lose the customers; it runs out of cash while waiting for them, which is a much worse outcome than losing them, because the business was right about the market and still failed.

The third is the acquisition cost that nobody computed. Without it, there is no way to know whether growth makes the business better or worse, and it is entirely possible to grow revenue while destroying the company. The arithmetic in §5.8 takes about an hour and prevents that.

In This Chapter

• 5.1 Describing an Offer So a Stranger Understands It
• 5.2 Features, Benefits and the Proof Line
• 5.3 Pricing in the Plan: What to Show and What to Justify
• 5.4 The Marketing Plan Section, in Two Pages
• 5.5 The Sales Process Section: Stages, Cycle Length and Conversion
• 5.6 Channels and Partners
• 5.7 The Go-to-Market Sequence and Its First Ninety Days
• 5.8 Customer Acquisition Cost and Payback, Computed
• 5.9 What to Do When You Have No History
• 5.10 Offer and Go-to-Market in Three Situations

5.1 Describing an Offer So a Stranger Understands It

5.1.1 The one-page test

The offer section has a simple test: hand the page to somebody outside your industry and ask them to explain back what you sell and who buys it. If they cannot, the page has failed, regardless of how accurate it is.

Failures come from three sources. Unexplained vocabulary, where trade terms appear without definition. Missing units, where the reader cannot tell whether you sell an hour, a month, a project or a device. And the specification substituted for the offer, where four paragraphs of technical detail arrive before the reader knows what the thing is for.

5.1.2 The structure that works

Describe each offer in five elements, in this order.

What it is, in one sentence, in ordinary language. What the customer gets, as a list of what is included and — equally important — what is not. What it costs them, with the unit stated. How it is delivered, in a sentence: who comes, how often, over what term. Why they choose it over their current alternative, which is the sentence that carries the argument.

Two or three offers presented this way take a page. Ten offers do not belong in a plan; group them into families and put the full list in an appendix.

5.1.3 Grouping many offers into families

Businesses with long catalogs face a real problem here: the plan cannot list everything, and summarizing badly hides the economics. Group by the way the customer buys, not by what the thing is.

Three grouping rules produce families a reader can follow. Group together offers that share a unit and a price logic, since a family whose members are all priced per month behaves as one line in the forecast. Group together offers sold in the same motion, meaning the same conversation, the same cycle and the same buyer, because those share an acquisition cost. And keep separate anything with a materially different margin, since merging a 60% margin service with a 12% margin resale hides the mix effect that will later explain why the gross margin moved.

For each family, the plan gives the five elements once, plus the range of prices within the family and the mix — what proportion of revenue each family represents today and is forecast to represent. The mix is the number a reader uses, because a shift in mix changes the margin without anything else changing, and it is the most common explanation for a forecast that misses while every unit assumption was correct.

DEFINITION Offer, unit, channel, pipeline An offer is a specific, purchasable combination of goods or services at a stated price, which is narrower than what a business "does" — a company that does building maintenance may have three offers. The unit is what the customer buys one of: an hour, a month of coverage, a project, a device, a seat. Naming the unit precisely is what makes the forecast computable, because volume × price only works when the volume has a unit. A channel is a path by which an offer reaches a buyer: direct sales, a partner, a marketplace, a website, a distributor. A pipeline is the set of potential sales in progress, arranged by stage, and it is the operational form of the sales process described in §5.5.

5.1.4 Naming what is excluded

Small businesses under-write exclusions in plans and over-experience them in operations. A reader who has bought services knows that scope boundaries are where disputes happen, and a plan that names them reads as experienced rather than as defensive.

For the running case, the preventive contract covers scheduled inspection of mechanical, plumbing and door systems, and it explicitly excludes parts, work beyond two hours per incident, roof and envelope, elevators, and anything requiring a licensed trade the company does not hold. Those five exclusions are why the hourly line in the revenue forecast exists at all — every one of them is billable work.

5.2 Features, Benefits and the Proof Line

5.2.1 The three-part line

A claim in a plan should have three parts: the feature (what exists), the benefit (what it does for the buyer), and the proof (why the reader should believe it). Most plans have the first two.

Consider: "We offer a ninety-minute response commitment on contract calls" — that is a feature. "Which means a tenant complaint is resolved the same day" — that is a benefit. "Our median response time on contract calls over the last twelve months was 71 minutes across 412 calls" — that is proof, and it is what converts the sentence from marketing into evidence.

5.2.2 Where proof comes from when you are small

Proof does not require scale. Four kinds are available to almost any business.

Your own operating record: response times, defect rates, retention, repeat purchase. Even one year of records beats any assertion.

A named customer's result, with permission: what changed for them, in their numbers.

A third-party fact: a certification held, an audit passed, a license maintained, an insurance limit carried.

A guarantee that costs you something if you are wrong: a service credit, a fixed price, a right to exit. A guarantee is proof precisely because it is expensive to offer falsely.

5.2.3 The claims that need proof most

Not every sentence needs a proof line; that would be unreadable. Three kinds do, without exception: any claim of speed, any claim of saving, and any claim of superiority over an alternative. Those three are the ones a reader will test, and an unsupported one in any of the three categories colors the rest.

5.3 Pricing in the Plan: What to Show and What to Justify

5.3.1 What belongs in the plan

The plan shows the price structure, not the price list. Structure means: what the unit is, what the price per unit is, how it varies (by size, by term, by volume), what discounts exist and under what conditions, and what the payment terms are.

The full price list, if you have one, goes to an appendix. Payment terms belong in the body because they drive the cash flow in chapter 7 — a business selling on net-60 terms has a fundamentally different cash profile from one collecting on delivery, and that difference is worth more space in the plan than the price itself.

5.3.2 What must be justified

Three pricing decisions require a justification in the plan, because a reader will otherwise supply their own and it will be less flattering.

Why this price and not less. The answer is not "our costs" — costs set a floor, not a price. The answer is what the customer gets, compared to what the alternative costs them, which is exactly the status quo arithmetic from the market section. That is the value-based logic the pricing literature sets against cost-plus pricing, and it is the argument a reader is looking for (Nagle & Müller, 2018).

Why this price and not more. If you are below the market, say whether it is a deliberate position or an artifact of never having raised prices. Both are acceptable; being unaware is not.

What happens if you have to move it. State the sensitivity: if the average price falls 5%, what happens to the net result? Chapter 7 computes it; the pricing section should mention that it was computed.

KEY CONCEPT A price in a plan is an argument, not a number The number itself carries almost no information to a reader who does not know your market. What carries information is the relationship between the price, what it costs you to deliver, and what the alternative costs the buyer. Present all three together and the price becomes self-explanatory. Present the price alone and every reader constructs their own comparison, usually against the cheapest thing they can think of. This is why the status quo arithmetic of the market section belongs, in one sentence, in the pricing section: "the piecemeal alternative costs a comparable building about $5,045 a year and produces eleven unplanned interventions; our contract is $8,220 and reduces those to a scheduled program."

5.3.3 Discounts and what they cost

If discounts exist, state them and state their cost, because a discount is a price change and it moves the entire margin structure. A 10% discount on a 39% gross margin removes just over a quarter of the margin: the margin falls from 39% to about 32%, which means you must sell roughly 22% more volume to end up in the same place. Put that sentence in the plan if you offer discounts; it demonstrates that you know what they cost.

→ see Book A7 (Pricing Strategy and Revenue Models) for how the price is set in the first place, which is a different exercise from how it is presented here.

QUÉBEC & CANADA NOTE What the law requires of a price you display A plan that describes prices should be written by somebody who knows the rules the prices will have to follow in market. In Québec, the Consumer Protection Act, CQLR c P-40.1 governs consumer transactions, including how prices are indicated and what an advertised price must include; the Competition Bureau enforces the federal prohibition on false or misleading representations, which covers advertised prices that are not attainable and savings claims that cannot be substantiated (Competition Bureau Canada, 2026). The Charter of the French language, CQLR c C-11 governs commercial publications, signage and contracts in Québec. Business-to-business transactions are governed by the contract and the Civil Code of Québec rather than by consumer law, which is why the running case — selling to property managers — has more latitude than a consumer-facing business would. Name the regime that applies to you in one sentence and move on; the detail belongs elsewhere. → see Book H4 (Consumer Protection, Advertising Law, Competition and Language Requirements).

5.4 The Marketing Plan Section, in Two Pages

5.4.1 What the two pages contain

A plan's marketing section is not a marketing plan; it is a summary of one, and two pages is the right length. → see Book B1 (Marketing Fundamentals for Business Owners) for the full discipline and Book B4 (The Marketing Plan, Budget and KPIs) for the standalone document.

Page one carries positioning and message: who the customer is, what you are to them, and the two or three things you say. Page two carries activities, budget and expected result: what you will actually do, what it costs, and how many customers it is expected to produce.

5.4.2 The budget must be built, not assumed

The circular budget — marketing set at a percentage of forecast revenue, forecast revenue driven by marketing — is the most common defect in this section, and the fix is mechanical.

Build the budget from the activities, not from the revenue. List what you will do, price each item, and total it. Then check the total against the acquisition arithmetic of §5.8: if the budget divided by the expected new customers gives an acquisition cost far from what your history shows, one of the two numbers is wrong and you have learned something.

For the running case, the year-one acquisition spending is $20,758, built from four lines: the owner's selling time at 276 hours, list and tool subscriptions, printed material, and referral payments. Nobody chose a percentage.

5.4.3 Naming the activities honestly

Two habits keep this section credible. Name only activities you will actually do, which for a small business is two or three, not nine. And state who does each one, because the most common reason a marketing plan does not happen is that every activity was implicitly assigned to the owner, who also sells and delivers.

5.4.4 Costing an activity properly

An activity's cost has three components, and plans routinely count only the first.

The money spent is the obvious one: the subscription, the advertisement, the printing, the event fee.

The time spent, priced is the one omitted. A weekly outreach block of six hours is not free because no invoice arrives; it is the most expensive line in most small-business marketing budgets. Price it at a loaded hourly cost — salary or draw, plus employer contributions, divided by working hours — and put it in the table.

The time to first result is not a cost but it belongs beside the cost, because it determines when the spending shows up in the cash flow. Outreach produces a meeting in weeks; content and search produce a first lead in months. A budget that mixes the two without saying so implies results that arrive earlier than they will.

The activities a small business actually runs are few, and each has a characteristic profile worth knowing before you choose. Direct outreach costs mostly time, produces results in weeks, and stops producing the week you stop doing it. Referral asks cost almost nothing, produce the highest conversion of any channel, and are capped by how many satisfied customers you have. A website and search presence cost money up front and produce nothing for several months, then produce steadily at a low marginal cost. Content behaves like search but demands a durable commitment; three articles do not work, and a plan that budgets three has budgeted a false start. Events and associations cost money and a great deal of time, produce a small number of high-quality contacts, and are worth budgeting only where the segment concentrates in one room. Paid advertising costs money immediately, produces results immediately, and stops immediately — which makes it a test instrument as much as a channel, and a reasonable way to measure demand before committing to something slower.

Choose two. Price them in the three components. Put the total in the model and the reasoning in one paragraph.

5.5 The Sales Process Section: Stages, Cycle Length and Conversion

5.5.1 Stages that are observable

A sales process is a sequence of stages, and a stage is only useful if an outsider could tell which stage a deal is in. "Interested" is not a stage; "walkthrough booked, with a date" is.

Five stages fit most small businesses: identified, contacted, meeting or demonstration held, proposal issued, signed. Name yours, define the observable event that moves a deal into each one, and use the same names in the plan, in the pipeline and in the forecast. → see Book E1 (The Sales Process: From Prospect to Signed Contract).

5.5.2 Conversion rates, and where they come from

Each transition has a conversion rate. If you have history, use your own; if you do not, §5.9 gives the alternatives.

The paragraph that follows introduces Table 5.1, which presents the running case's sales funnel for year one. Its columns give each stage, the count that reaches it, the conversion from the previous stage, and the evidence behind that rate. Read the evidence column first: it is what makes the other three believable, and it is the column most plans omit.

Table 5.1 — The five sales stages, their conversion and the evidence behind each rate

Two things about that table are worth imitating. The rates are taken from the company's own prior year, not from an industry benchmark, and the one adjustment made is stated and is downward. And the last line ties to the forecast: eleven signed is exactly the number the revenue forecast in chapter 7 uses, which means a reader can trace the revenue back to an activity count.

5.5.3 Cycle length, and why it is usually understated

The sales cycle is the time from first contact to signature, and the honest measure is the median, not the average, because a handful of very long deals distort the mean.

For the running case the cycle is eleven weeks, decomposed as three weeks from contact to walkthrough, one week from walkthrough to proposal, and seven weeks from proposal to signature. That last interval is the important one: it is not indecision, it is the customer's renewal date on their existing arrangement. A cycle driven by the customer's calendar rather than by their enthusiasm cannot be shortened by selling harder, and a plan that understands this will not assume it can.

Write the cycle down and then check one thing: does the cash flow forecast in chapter 7 assume revenue starting before the cycle allows it? A sale initiated in month one at an eleven-week cycle produces revenue in month four, not month two.

THE RULE Every conversion rate in a plan names its evidence A rate without a source is an aspiration with a decimal point, and a reader treats it as one. Each rate must name where it came from, in one of four permitted forms: your own record, with the counts and the period; an adjusted analogue, with the adjustment and its direction stated; a measured test, with the sample size and the dates; or an explicit estimate, labeled as such in the assumption log with the date it will be replaced. There is no fifth form, and in particular an industry benchmark quoted without a source is not one of them. This rule matters more than it appears, because conversion rates multiply: a forecast that uses four rates has four chances to be quietly wrong, and the errors compound in the same direction — optimistically.

METHOD Writing the sales process section in five stages First, name the stages by their observable event. Write each stage as something that either happened or did not: "a walkthrough took place on a date," not "the prospect is engaged." Five stages is right for most small businesses; more than seven means you are describing internal steps rather than the customer's progress. Second, count what happened last year at each stage. If you have no records, count what you can reconstruct from invoices, calendars and email, and say that is what you did. An imperfect reconstruction with a stated method beats a benchmark. Third, compute each transition and write the evidence beside it, in the form the rule above requires. Fourth, measure the median duration of each interval — contact to meeting, meeting to proposal, proposal to signature — separately rather than end to end, because the intervals have different causes and only some of them are within your control. Fifth, tie the last line to the forecast. The number of signatures the funnel produces must be exactly the number the revenue forecast uses. If they differ, the plan contains two sales forecasts and a reader will find the seam. The deliverable is a five-row table with counts, rates, evidence and durations. The verification criterion is that a reader can multiply your activity count by your rates and arrive at your revenue.

5.5.4 What lengthens a cycle, and what legitimately shortens it

Three things lengthen a sales cycle, and only one of them is about the seller.

The customer's calendar lengthens it and cannot be argued with. A renewal date, a budget year, a lease expiry, a tender schedule — these set a date, and effort before that date changes nothing. The correct response is to start earlier, not to push harder, which is a planning decision rather than a selling one.

The number of approvals lengthens it in proportion to how many people must say yes. Each additional approver adds their own availability to the timeline. The response is to find out early who they are, which is a question that can simply be asked in the first meeting and almost never is.

Your own follow-up gaps lengthen it, and this is the one within your control. A proposal that sits for two weeks because nobody was assigned to follow up has added two weeks to the cycle and to the cash flow. The fix is a task with a date and an owner, which is the subject of chapter 10.

The legitimate ways to shorten a cycle are correspondingly narrow: remove an approval by pre-qualifying the decision-maker, remove a step by combining the walkthrough with the proposal, or remove a wait by asking at the first meeting when the current arrangement ends and scheduling backwards from it. Discounting to accelerate is not on that list; it changes the price permanently to move a date once.

5.6 Channels and Partners

5.6.1 The channel question

A channel is how the offer reaches the buyer, and the plan should name at most three. Each one carries a cost, a cycle length and a set of requirements, and adding one is a real commitment rather than an option to keep open.

The paragraph below introduces Table 5.2, which compares the channel types available to a small business. Its columns give the reach each one provides, its cost structure, the typical cycle length, and what the channel requires of you before it produces anything. That last column is the one that determines feasibility: a channel whose requirement you cannot meet is not a channel you have.

Table 5.2 — Channels compared: reach, cost, cycle and what each requires of you

5.6.2 The channel a small business under-uses

Referral is nearly always the cheapest channel and nearly always the least deliberately worked. The reason is that it feels passive, and the fix is to make it an activity with a trigger: ask at a defined moment, in a defined way, and record the result.

For the running case, referrals are triggered at the ninety-day mark of a new contract, when the first quarterly report is delivered, and the incentive is a fixed amount rather than a percentage. Five referral payments appear in the year-one acquisition budget, which is an assumption to be checked against actuals rather than a hope.

5.6.3 Partners belong in the plan only when the terms exist

A plan that says "we will develop partnerships with property management firms" has stated an intention with no content. A plan that says "we have a written referral arrangement with one firm paying a fixed amount per signed building, and we are seeking two more on the same terms" has stated a fact and a target. If the terms do not exist yet, say what they will be. → see Book F1 (Go-to-Market and Channel Strategy).

5.7 The Go-to-Market Sequence and Its First Ninety Days

5.7.1 Sequence rather than list

The difference between a go-to-market section that works and one that does not is usually sequencing. A list of activities can be started in any order and therefore tends to be started in all orders at once. A sequence states what must be true before the next step is worth doing.

Figure 5.1 draws the sequence for a service business, from offer to first revenue. It reads as six stacked rows: the step on the left, the observable checkpoint that completes it in the middle, and the week it falls on at the right. The checkpoints are the point of the figure — each one is an event that either happened or did not, rather than a feeling — which is what makes the sequence usable directly as a set of tasks with verification criteria in chapter 10. Read the week column downward and the eleven-week cycle of §5.5.3 appears as a calendar rather than as a statistic.

Figure 5.1 — Original figure. The go-to-market sequence: six steps from offer to first revenue, the observable checkpoint that completes each one, and the week each one falls on.

5.7.2 The first ninety days

The ninety-day view is what a reader uses to judge whether the plan is executable. It should name, for each month, the activity, the volume and the expected result.

For the running case: month one is list assembly and the first forty contacts, expecting twelve walkthroughs; month two is the first twelve walkthroughs and eight proposals, expecting two signatures; month three is forty more contacts and the follow-up cycle on the proposals issued in month two, expecting three more signatures. The cumulative expectation at ninety days is five signed buildings, against a year-one target of eleven — which is deliberately front-loaded, because the two technicians arrive in month four and the work must exist before they do.

That last sentence is the kind of reasoning a plan should make visible. The hiring date and the sales sequence are linked; a reader who sees the link believes the forecast, and a reader who does not will ask why you are hiring before the revenue.

5.8 Customer Acquisition Cost and Payback, Computed

This is the arithmetic that most distinguishes a plan a lender trusts from one they do not.

DEFINITION Customer acquisition cost, payback period, conversion rate Customer acquisition cost (CAC) is everything spent to acquire customers in a period, divided by the number of customers acquired in that period. "Everything" includes the value of the owner's own selling time, which is the input most often omitted and usually the largest. The payback period is how long the customer's contribution margin takes to repay the acquisition cost; it is measured in months and it is the number that determines how fast you can grow without running out of cash. A conversion rate is the proportion of deals moving from one stage to the next; it is only meaningful when both stages are defined by observable events.

DO THE MATH CAC, payback and the first-year budget, fully worked The figures below belong to the fictional running case. Step one: total the acquisition spending for year one. The owner's selling time is six hours a week over forty-six working weeks: 6 × 46 = 276 hours, at a loaded cost of $58 per hour, which is 276 × $58 = $16,008. List and tool subscriptions are $2,400. Printed material and sample reports are $1,100. Referral payments are five at $250, or $1,250. Total acquisition spending: 16,008 + 2,400 + 1,100 + 1,250 = $20,758. Step two: divide by customers acquired. Year one signs 11 buildings: $20,758 ÷ 11 = $1,887 per building. That is the CAC. Step three: compute the contribution margin per building per year. The contract is $8,220 a year at a 39% gross margin, which is 8,220 × 0.39 = $3,206. A building under contract also generates about 24 hours of intervention work a year at $96, which is $2,304 of revenue at the same margin, or 2,304 × 0.39 = $899. Total annual contribution: 3,206 + 899 = $4,105, which is $4,105 ÷ 12 = $342 per month. Step four: the payback period. $1,887 ÷ $342 = 5.5 months. Step five: lifetime value. At an 88% annual renewal rate, the average contract lasts 1 ÷ (1 − 0.88) = 8.3 years. Lifetime contribution is 4,105 × 8.3 = $34,072, and the ratio of lifetime value to acquisition cost is 34,072 ÷ 1,887 = 18.1. What the result actually means. A ratio of eighteen is very high, and the correct reading is not "our marketing is excellent." It is that acquisition is not this business's constraint — capacity is. A business with a payback of 5.5 months and a ratio of 18 should be asking whether it is spending too little on acquisition rather than too much, and the honest answer here is that it cannot spend more until it can serve more, which is precisely what the financing request in chapter 8 is for. When the ratio is this lopsided, the plan's argument moves from "we need to sell better" to "we need to deliver more," and that is a different plan.

5.8.1 Reading the number

Three readings are worth knowing.

A payback longer than twelve months means growth consumes cash faster than it produces it, and the plan must show where that cash comes from. This is normal for equipment-heavy or subscription businesses and it must be financed deliberately.

A payback under six months means acquisition is cheap relative to value, and the binding constraint is almost certainly somewhere else — capacity, delivery, or the size of the reachable market.

A ratio of lifetime value to acquisition cost below three is a warning: after allowing for the costs not in the contribution margin, the business may be acquiring customers at a loss. Check the retention assumption first, since lifetime value is very sensitive to it.

WATCH OUT The marketing budget that is a percentage of a revenue you do not have "We will invest 8% of revenue in marketing" sounds disciplined and is circular: the revenue depends on the marketing and the marketing depends on the revenue. It also produces a specific operational absurdity — the budget falls in the months when sales are weak, which are exactly the months when it should not. Build the budget from activities, price each one, total it, and then divide by expected new customers to check the implied acquisition cost against your own history. If the two disagree by more than about 30%, one of them is wrong, and finding out which is worth an afternoon.

5.9 What to Do When You Have No History

A business with no operating history cannot take conversion rates from its own record, and the temptation is to take them from an industry article. Three better options exist.

Run a small test and measure it. Forty contacts, made properly, produce a real contact-to-meeting rate in two weeks. It is a small sample and it is your sample, which makes it more useful than a benchmark from a different market.

Use an analogue with the difference stated. If the founder ran the same process at a previous employer, use those rates and say so, adjusted for what is different — a smaller list, no brand, a different price point. State the adjustment and its direction.

Use a conservative placeholder and mark it. Where no evidence is available, choose a deliberately pessimistic rate, label it in the assumption log as an estimate, and state what would replace it and when. This is the weakest option and it is honest, which makes it stronger than an unlabelled benchmark.

In all three cases, the sensitivity analysis of chapter 7 must include the conversion rate, because it is the assumption most likely to be wrong and it drives the whole forecast.

CASE IN POINT Nordec Entretien: two economies in one company This case is fictional; the company, the people and the figures are invented. The first version of Sophie Nadeau's plan described the business as "commercial building maintenance" and gave one price: $96 an hour. The second version separated two things that had always been separate in the operation and never in the document. The contract economy is recurring: $685 a month per building, sold in an eleven-week cycle, renewed at 88%, and constrained by how many buildings one technician can cover on a preventive schedule. Its acquisition cost is high and paid once; its margin is stable; its value is in the retention. The intervention economy is transactional: $96 an hour plus parts, sold in minutes rather than weeks, generated mostly by buildings that are already under contract. Its acquisition cost is effectively zero, because the customer is already yours; its margin varies with the job; its volume is constrained entirely by available hours. Writing them separately changed three things in the plan. The forecast became two lines instead of one, so a reader could see which was growing. The acquisition arithmetic became computable, because the eleven signed contracts had a cost and the interventions did not. And the capacity argument in chapter 6 became visible: the preventive schedule consumes hours that cannot be sold at $96, which is a real cost of the contract economy that the single-price version had hidden entirely.

5.10 Offer and Go-to-Market in Three Situations

5.10.1 Solo

Your acquisition cost is dominated by your own time, and the first useful thing this chapter can do for you is force you to price it. Six hours a week of selling at a loaded rate is a real number that belongs in the model, and seeing it usually changes how those six hours are spent.

Choose one channel and work it. A solo business with three channels has three half-worked channels, and the honest arithmetic of §5.8 will show it.

5.10.2 Small team

The question to settle in writing is who sells. In most small teams the answer is the owner, which caps acquisition at whatever hours the owner has left after delivering. Say so in the plan, and let the constraint be visible: it is usually the strongest argument for the hire the plan is asking to fund.

5.10.3 Established business

You have conversion data, and the plan should use it by segment rather than in aggregate. An overall 26% proposal-to-signature rate that is 40% in one segment and 12% in another is telling you where to spend, and a plan that shows the split reads as a business that measures.

The specific risk at this size is a marketing section describing activities that a department performs and a forecast that assumes results from activities nobody has scheduled. Reconcile the two before delivery.

WHO DOES THIS Who owns go-to-market when there is no marketing person The owner owns the message and the price, in every situation, because both are strategic decisions rather than executional ones and both will be defended by the owner in front of a lender. Whoever talks to customers owns the conversion rates, and should be asked to state the evidence behind each rate rather than the rate alone. A rate offered without evidence is a hope with a decimal point. Somebody must own the list, meaning the countable population from chapter 4 turned into a working file with names, dates and outcomes. In a small business this is usually the same person who sells, and it is the first task worth systematizing. Nobody should own "marketing" as an abstraction. Assign activities, not the function: this campaign, this referral ask, this content piece, each with a name and a date beside it.

TOOLS OF THE TRADE What a go-to-market section needs to be measurable The claims in this chapter are only defensible if four things are being recorded, whatever the tool. You need a list with outcomes, meaning every prospect with the stage they reached and the date they reached it, because conversion rates are computed from it. You need the selling time recorded, at least approximately, since it is the largest component of acquisition cost and the one most often left out. You need the source of every won customer, recorded at the moment of signature rather than reconstructed later, because channel comparison is impossible without it. And you need the date of each stage transition, which is what produces a median cycle length rather than an impression of one. A spreadsheet does all four; chapter 13 names the products that do them with less friction.

WHAT CHANGED Cycles have lengthened at the proposal stage Practitioners across several service sectors report the same shift over recent years: the early stages of a sale have become faster, because a buyer arrives better informed, while the final stage has become slower, because more people are involved in approving it. For a plan, the consequence is specific and it affects cash rather than volume: the gap between proposal and signature is where the forecast breaks, and it is the interval to measure rather than assume. If you have any history, compute that one interval from your own records before writing the forecast; if you do not, assume it is longer than it feels. As of September 2026.

Costly Mistakes

1. Setting the marketing budget as a percentage of forecast revenue. It makes the plan circular, and it cuts spending precisely in the months when sales are weakest.
2. Omitting the owner's selling time from acquisition cost. It is usually the largest single component, and leaving it out produces an acquisition cost that looks excellent and means nothing.
3. Using the mean sales cycle instead of the median. A few long deals pull the mean upward and hide the fact that most sales are faster, or — worse — a few fast ones hide that most are slow.
4. Assuming revenue begins before the cycle allows. A sale started in month one at an eleven-week cycle is revenue in month four; a cash flow that books it in month two is wrong by two months on every line.
5. Naming five channels. A small business works one or two; naming five means none of them has an owner, a budget or a schedule.
6. Writing "we will develop partnerships" with no terms. It is an intention, not a plan, and a reader discounts it entirely. State the terms, even if the partner does not exist yet.
7. Presenting one price for two different economies. A recurring contract and a transactional intervention have different acquisition costs, different margins and different constraints, and merging them hides the capacity problem that chapter 6 exists to find.

Key Takeaways

• The offer section passes or fails one test: a stranger reads the page and can explain back what you sell and who buys it.
• Every claim of speed, saving or superiority needs a proof line, and proof is available to small businesses from their own operating record.
• A plan shows the price structure, not the price list, and payment terms matter more to the reader than the price itself because they drive cash.
• The marketing budget is built from priced activities and then checked against the acquisition arithmetic — never set as a percentage of a revenue that does not exist.
• Sales stages must be defined by observable events, and conversion rates should come from your own record, an adjusted analogue, or a labeled conservative placeholder.
• The median cycle length, decomposed by interval, tells you when revenue can begin; the proposal-to-signature gap is the interval that most often breaks a forecast.
• Customer acquisition cost must include the owner's selling time; payback and the lifetime-to-acquisition ratio then tell you where the real constraint is.
• A very high lifetime-to-acquisition ratio is not a compliment to your marketing; it usually means the constraint is capacity, which changes what the plan should be asking for.

Your Action Plan

1. Write each offer in the five elements, and give the page to a stranger. Deliverable: one page per offer family, covering what it is, what is included and excluded, what it costs, how it is delivered, and why it is chosen. Timing: two hours. Verification criterion: someone outside your industry explains your offer back correctly.
2. Add a proof line to every claim of speed, saving or superiority. Deliverable: a revised section in which each of those claims carries a number, a source or a guarantee. Timing: ninety minutes. Verification criterion: no such claim remains unsupported.
3. Build the marketing budget from priced activities. Deliverable: a list of activities with a cost and an owner for each, totalled. Timing: two hours. Verification criterion: the total divided by expected new customers is within 30% of your historical acquisition cost, or you can explain the difference.
4. Define your sales stages by observable events and compute your conversion rates. Deliverable: a five-line table with counts, rates and the evidence for each rate. Timing: three hours if you have records, two weeks if you must run a test. Verification criterion: an outsider could look at any deal and say which stage it is in.
5. Compute your CAC, payback and lifetime ratio in a spreadsheet. Deliverable: the five-step calculation of §5.8 on your own numbers, including your selling time. Timing: one hour. Verification criterion: the acquisition spending total matches the marketing budget you built in step three.
6. Write the ninety-day sequence with a monthly expected result. Deliverable: three months, each with the activity, the volume and the expected outcome, ending in a cumulative number. Timing: one hour. Verification criterion: the ninety-day cumulative number is consistent with the year-one figure in your forecast.

In La Machine

Every route named in this section was verified in the product's route inventory on 2026-09-09 (La Machine, 2026a), and what each page holds is described in the product's user guide (La Machine, 2026b).

• The offer, as the product holds it. business/packages holds packages and business/subscription-packages holds subscription packages, whose billing period is monthly or annual and nothing else. business/promotions holds promotions and client/promotions shows them to a client; a promotion carries a whole-number percentage of discount with a start and an end date, and it carries no code, no fixed amount, no free month and no usage limit. plans and /pricing are not your pricing pages: they are the plans and the prices of La Machine itself.
• The pipeline and the people in it. business/marketing/contacts holds contacts with their detail, edit and import routes; crm/clients and crm/client-details hold clients; crm/intake holds the client intake form.
• Campaigns, sequences and funnels. business/marketing/campaigns, business/marketing/sequences and business/marketing/funnels hold the marketing activities, each with an edit route. A sequence is the one object here that can send on its own: a background engine reads the active ones, enrolls contacts when a matching event arrives and hands each message to the dispatcher, so marking a sequence active is a decision to send rather than a drawing. A funnel is a frame rather than an engine, and it runs nothing. business/marketing/log holds the page titled Action Log, which lists one-to-many outbound communications and excludes one-to-one messages and anything inbound.
• Consent and exclusions, which govern outreach. business/marketing/consent shows, channel by channel, whether one contact may be written to and on what basis; it has no menu entry, it is reached by its address, and it opens one contact at a time with no list and no export. business/marketing/do-not-reach is the single master exclusion list, shared across the businesses of the deployment; /exclusions no longer opens a page of its own and redirects to it. → see Book H3 (Anti-Spam and Telemarketing Compliance).
• Where the selling work lives. Outreach tasks live at /taches with their verification criteria, and no page creates one: a plan task is loaded through the programming interface, a script or the assistant. /aujourdhui gathers what needs a decision now, with one caution — the tasks due today appear there as a count and not as a list.
• What it does not do. La Machine does not compute customer acquisition cost, payback or lifetime value, and it does not model conversion by stage. It holds the activities and the records from which those figures are computed; the computation is done in the tools of chapter 13.

Self-Check

1. Give the one-page test for an offer description, and name the three failures that most often cause it to fail.
2. Turn this feature into a three-part line: "we answer the phone twenty-four hours a day." Supply a plausible benefit and say what would count as proof.
3. A plan sets its marketing budget at 8% of forecast revenue. Explain the circularity in one sentence, and give the operational absurdity it produces.
4. Compute the payback period for a business whose acquisition cost is $2,400 and whose customer contributes $180 of margin per month. Say what that number implies about how fast it can grow without financing.
5. Your lifetime-to-acquisition ratio is 19. Give the correct reading of that number, and say what question the plan should ask next.

Chapter References

Charter of the French language, CQLR c C-11 (Québec). LégisQuébec. Read September 9, 2026.

Civil Code of Québec, CQLR c CCQ-1991. LégisQuébec, consolidated version updated to April 7, 2026. Read September 9, 2026.

Competition Bureau Canada. (2026). Deceptive marketing practices: price representations and ordinary selling price. Retrieved September 9, 2026.

Consumer Protection Act, CQLR c P-40.1 (Québec). LégisQuébec. Read September 9, 2026.

La Machine. (2026a). Documentation — route reference inventory (211 routes: 191 Core, 20 Pilotage paths). Internal product documentation, read September 9, 2026.

La Machine. (2026b). Guide utilisateur [User guide] (FAQ: marketing modules; CRM and contacts; prospecting strategy; CASL compliance). Internal product documentation, read September 9, 2026.

Nagle, T. T., & Müller, G. (2018). The strategy and tactics of pricing: A guide to growing more profitably (6th ed.). Routledge.

Chapter 6 — Operations, Team and Governance

This is the section where a plan either proves that its revenue is deliverable or quietly reveals that it is not. The market section said customers exist. The go-to-market section said they can be reached. This one has to say that when they arrive, somebody can actually serve them — and it has to say it with arithmetic, because the single most common internal contradiction in a small-business plan is a revenue line that exceeds what the operations section describes the company as able to produce.

What You'll Be Able to Do

• Describe your delivery model step by step, so that a reader can see how a sale becomes a delivered service or product.
• Compute your capacity ceiling from hours available, utilization and price, and check the sales forecast against it.
• Name your dependencies and single points of failure, including the ones that are people rather than suppliers.
• Budget the founder's own time honestly, including the hours that are not billable and cannot be.
• Write a milestones table whose entries carry a date, an owner, a deliverable and an observable verification criterion.
• Describe governance in proportion to the business, without pretending to a board that does not exist.

Why It Matters

A lender reads this section to answer a question that is not about operations at all: if the sales forecast is right, does this company fall over? Growth kills more small businesses than stagnation does, and it kills them through operations — a company wins the work, cannot deliver it, damages its reputation, loses the customers it already had, and runs out of cash paying for capacity it acquired too late or too early.

For you, the section has a different value. The capacity arithmetic in §6.3 usually produces one uncomfortable number, and that number is often the most important finding in the whole planning exercise: the point at which the business, as currently configured, cannot grow any further. Most owners have a rough sense of it. Very few have computed it, and the difference between a sense and a number is the difference between hoping the timing works and knowing which month it breaks.

Finally, the milestones table written here becomes the execution plan of chapter 10. It is the one part of the plan that survives directly into daily work, which makes it worth writing with more care than its two pages suggest.

In This Chapter

• 6.1 What the Operations Section Proves
• 6.2 The Delivery Model, Step by Step
• 6.3 Capacity: The Arithmetic Nobody Writes Down
• 6.4 Suppliers, Dependencies and Single Points of Failure
• 6.5 Location, Equipment and What It Costs to Start
• 6.6 The Team Section: Roles Before Names
• 6.7 The Founder's Own Time, Honestly Budgeted
• 6.8 Advisors, Board and Governance in a Small Company
• 6.9 Milestones: The Section That Becomes the Execution Plan
• 6.10 Operations, Team and Governance in Three Situations

6.1 What the Operations Section Proves

The operations section has one job: to establish that the promise in the offer section can be kept at the volume in the financial section. Everything in it serves that job.

That framing tells you what to include and what to leave out. Include whatever a reader needs to believe the capacity claim: the steps, the people, the equipment, the suppliers, the constraint. Leave out the process detail that belongs in a procedure manual, the equipment specifications that belong in an appendix, and the description of activities that any business in your sector performs identically.

KEY CONCEPT The operations section is where a plan proves the revenue is deliverable Every other section makes a claim about the world: customers exist, they can be reached, they will pay this much. This section makes a claim about you: that this specific set of people, tools and arrangements can produce the volume the forecast assumes. It is the only section whose facts are entirely within your control, which means an error here is never bad luck. A reader who finds that the revenue line requires more hours than the company has is not discovering an optimistic assumption; they are discovering that the plan was not checked, and that conclusion transfers to every other number in it.

6.1.1 The three questions to answer

How does the work get done? The delivery model, in steps, with who does each one.

How much of it can you do? The capacity ceiling, computed, with the constraint named.

What could stop it? Dependencies, single points of failure, and what happens when one fails.

A section that answers those three in four pages is complete. A section that spends four pages on the first and never answers the second is the common failure.

6.2 The Delivery Model, Step by Step

6.2.1 From sale to delivered

Describe the path a unit of work takes, from the moment a customer says yes to the moment the work is complete and invoiced. Six to ten steps is typical. For each step, state what happens, who does it, how long it takes, and what it depends on.

For a service business, the steps usually include intake, scheduling, preparation, execution, verification, and invoicing. For a product business, they include ordering, receiving, storage, picking, shipping and after-sales. Neither list is universal; write yours.

The value of writing it as steps rather than as prose is that steps expose the handoffs, and handoffs are where small businesses lose time. A step that passes from one person to another has a waiting period attached to it whether or not anybody has measured it.

6.2.2 Naming the constraint

Every delivery model has one step that limits throughput. In a service business it is usually a person; in a product business it is usually a machine, a supplier lead time or a storage limit.

Name yours explicitly, because it is what chapter 7 will need in order to model growth. A business that adds sales capacity when its constraint is delivery capacity has spent money to make its problem worse, which happens often enough to be worth the sentence.

6.2.3 What quality means, operationally

Two lines are worth spending here. What defines an acceptable outcome, stated as something observable, and what you do when it is not met. A business with a defined response to a failure is a business that has thought about failures.

This is also where a plan can note, briefly, whether it works to a recognized standard. Where a quality management system is in place or planned, say which one and what it covers; a general commitment to quality is not information. The best known of those standards asks a business to plan and control the processes it needs in order to deliver, and to control its production and service provision, which is exactly the ground this section covers (International Organization for Standardization, 2015). → see Book J4 (Quality Management and Continuous Improvement).

6.2.4 The delivery model of the running case, in eight steps

The case is fictional; the structure is what to imitate. The value of writing it this way is that each step has a duration and an owner, which is what makes the capacity arithmetic of §6.3 possible at all.

1. Intake. A call or an email arrives and is logged with the building, the symptom and the urgency. Owner: the dispatcher, currently the president. Duration: four minutes. Depends on: somebody answering, which is the point of the role this plan funds.
2. Triage. The request is classified as contract-covered, billable, or outside scope. Owner: dispatcher. Duration: three minutes. This is the step that determines which revenue line the work lands in, and getting it wrong quietly costs margin.
3. Scheduling. The work is placed on a technician's day, respecting the ninety-minute response commitment for contract calls. Owner: dispatcher. Duration: five minutes. Depends on: knowing where the technicians are.
4. Preparation. Parts and access are confirmed before departure. Owner: technician. Duration: fifteen minutes. This is the step most often skipped and the one that produces the second trip.
5. Travel. Owner: technician. Duration: twenty-five minutes on average. Non-billable, and the largest single component of the gap between available and billable hours.
6. Execution. Owner: technician. Duration: the job. Billable, at the hourly rate, except where covered by the contract.
7. Verification and record. The technician records what was done, what was used and what is still outstanding. Owner: technician. Duration: eight minutes. This is what makes the monthly client report possible without anybody reconstructing the month.
8. Invoicing. Owner: administrator, weekly. Depends on: step seven having happened, which is why a missing record is a cash problem rather than a paperwork problem.

Two things become visible when the model is written this way. Steps one to three consume the dispatcher's day, which is why one person carrying dispatch and selling simultaneously is the constraint. And steps four, five and seven are non-billable and unavoidable, together accounting for roughly forty-eight minutes per intervention, which is the arithmetic behind a utilization rate of 68% rather than the 85% that a naive plan would assume.

6.3 Capacity: The Arithmetic Nobody Writes Down

This section contains the calculation that most changes a first plan.

DEFINITION Capacity, utilization, lead time, bottleneck Capacity is the maximum output the business can produce in a period with its current resources, expressed in the unit the business sells — billable hours, units shipped, clients served, seats filled. Utilization is the proportion of available time that is actually spent on billable or productive work, and for field service, professional services and trades it is almost never above about 75%; travel, quoting, administration and waiting consume the rest. Lead time is the elapsed time from order to delivery, which is a customer-facing measure and is distinct from the labor the work consumes. A bottleneck is the step whose capacity is lowest, which therefore sets the capacity of the whole system regardless of how much capacity the other steps have.

6.3.1 Building the number

Capacity is computed in four moves, and each one has a place where businesses lose track of reality.

Move one: available hours per person. Start from paid hours and subtract what is genuinely not available. In Québec, the Act respecting labour standards, CQLR c N-1.1 sets out the statutory general holidays in section 60 — seven of them: 1 January, Good Friday or Easter Monday at the employer's option, the Monday preceding 25 May, 1 July, the first Monday in September, the second Monday in October, and 25 December — and the National Holiday of 24 June is provided by the National Holiday Act, CQLR c F-1.1, for eight in total. Annual leave is governed by sections 66 to 69 of the same Act, which set the reference year and the entitlement that grows with uninterrupted service. Read in the LégisQuébec consolidation updated to April 7, 2026, on September 9, 2026. The commission that administers those standards publishes the current list of the holidays and the annual leave entitlement on its own pages, which is where to read them rather than from memory (Commission des normes, de l'équité, de la santé et de la sécurité du travail, 2026). Subtract holidays, vacation, and a realistic allowance for illness and training.

Move two: the utilization rate. This is where optimism enters. Use your own measured rate if you have one; if you do not, look at last year's billable hours divided by available hours and use that, even if the number is uncomfortable. A rate assumed rather than measured is the single most common reason a capacity calculation is wrong.

Move three: subtract the work that is not sold at the marginal price. Scheduled or included work consumes capacity and produces no incremental revenue. In a service contract this is the preventive schedule; in a product business it is rework, setup and returns.

Move four: multiply by the price. What remains, at your price, is the revenue ceiling.

DO THE MATH Capacity, computed: the ceiling the forecast must respect The figures below belong to the fictional running case. Available hours per technician. Paid hours are 52 weeks × 40 hours = 2,080. Subtract two weeks of vacation (80 hours), eight statutory holidays at eight hours (64 hours), and an allowance of 40 hours for illness and training. Available: 2,080 − 80 − 64 − 40 = 1,896 hours. The owner's field time. Sophie spends about 30% of her time in the field: 0.30 × 1,896 = 569 hours. Total available field hours today. Two technicians plus the owner's share: (2 × 1,896) + 569 = 4,361 hours. Billable hours at the measured utilization. Last year's billable hours divided by available hours gave 68%: 4,361 × 0.68 = 2,965 billable hours. Subtract the preventive schedule. Thirty-four buildings at 1.5 hours a month: 34 × 1.5 × 12 = 612 hours. Available for billable interventions: 2,965 − 612 = 2,353 hours. The revenue ceiling today. Contract revenue is 34 × $8,220 = $279,480. Intervention labor is 2,353 × $96 = $225,888. Parts resale at the historical ratio is about $107,000. Ceiling: $612,368. The finding. Last year's actual revenue was $612,100. The business is operating at 99.96% of its own capacity ceiling — which is not a compliment, it is the whole argument of the plan. No sales effort of any kind can raise revenue by a dollar next year without additional capacity. The ceiling after the investment. With four technicians plus the owner's share: (4 × 1,896) + 569 = 8,153 hours, at 68% = 5,544 billable hours. The preventive schedule at 52 buildings is 52 × 1.5 × 12 = 936 hours, leaving 4,608 hours for interventions. Contracts 52 × $8,220 = $427,440; interventions 4,608 × $96 = $442,368; parts about $185,000. New ceiling: $1,054,808. The check that matters. The year-three forecast is $933,800, which is 933,800 ÷ 1,054,808 = 88.5% of the new ceiling. A forecast that sits below its own ceiling with room to spare is a forecast a reader can accept; one that sits above it is a forecast that has not been checked.

Figure 6.1 draws that calculation as a descending staircase, from paid hours at the top to the revenue ceiling at the bottom. Each step shows what is subtracted and what remains, so the compression from 2,080 paid hours per person to the hours that are actually sold is visible in one view. The two panels at the bottom compare the ceiling before and after the investment this plan requests, with the forecast plotted against each — which is the comparison a reader performs mentally and rarely finds drawn.

Figure 6.1 — Original figure. From paid hours to the revenue ceiling: what each subtraction removes, and how the forecast compares with the ceiling before and after the investment.

6.3.2 Capacity for businesses that do not sell hours

The same logic applies with a different unit. A retailer's capacity is a function of floor space, inventory turns and staffed hours. A manufacturer's is machine hours times yield. A software business's capacity constraint is usually support and onboarding rather than the software itself.

In every case the method is identical: start from a countable resource, subtract what is unavailable, apply a realistic utilization, subtract what is consumed without producing incremental revenue, and multiply by price.

THE RULE A sales forecast may never exceed the capacity the plan describes If the forecast requires more output than the operations section can produce, one of the two must change before the plan is delivered — and the honest change is almost always to the forecast, because capacity is a physical fact and the forecast is an assumption. Where the plan is asking for financing precisely in order to raise capacity, the sequence must be visible: capacity arrives in a stated month, and the revenue that depends on it begins after that month, not before. This is the single check that catches the most common contradiction in small-business plans, it takes fifteen minutes, and no reader who performs it will fail to notice if you did not.

6.4 Suppliers, Dependencies and Single Points of Failure

6.4.1 What to list

List the dependencies whose failure would stop or materially damage the business. For most small businesses there are between three and eight, and they fall into four kinds: suppliers of goods or materials, subcontractors who perform part of the work, platforms the business runs on, and people who hold knowledge or relationships that nobody else holds.

For each, state what it provides, what proportion of the business depends on it, what the switching cost and lead time would be, and whether an alternative exists today.

6.4.2 Concentration is a dependency

A customer who represents a large share of revenue is a dependency in the same sense as a supplier. State the concentration explicitly — the largest customer as a percentage of revenue, and the top three combined — because a reader will compute it from your numbers anyway and it is better coming from you with the mitigation attached.

The threshold at which concentration becomes a live concern varies, but a single customer above roughly a quarter of revenue changes how a lender views the whole file, and above a third it usually becomes a condition of the financing.

6.4.3 The person nobody has replaced

The most under-declared dependency in small business plans is a person: the one who knows the pricing, holds the customer relationships, or is the only one who can do a particular kind of work. In a business of four people it is usually the owner.

Naming it costs nothing and buys credibility, because every reader already assumes it. What earns more is stating what is being done about it — a documented procedure, a second person trained, a written handover, an insurance policy. → see Book J2 (Processes, SOPs and the Definition of Done).

CASE IN POINT Nordec Entretien: the capacity ceiling two years before the forecast This case is fictional; the company, the people and the figures are invented. The first version of Sophie Nadeau's plan forecast revenue of $1.1 million in year two. The operations section described the business as it existed: two technicians, one vehicle, and the owner in the field about a third of the time. Nobody had multiplied the two together. When she did the arithmetic in §6.3, the result was uncomfortable in a useful way. The current configuration produced a ceiling of about $612,000, and the business had billed $612,100 the previous year. There was no room at all. The $1.1 million forecast was not ambitious; it was arithmetically impossible without a second vehicle and two more technicians, which the first plan mentioned in a single line under "future growth." Three things changed. The forecast came down to $933,800 in year three, which is 88.5% of the ceiling after the investment — leaving room for the utilization rate to be a few points worse than assumed. The financing request moved from a vague growth argument to a specific one: the vehicle and the hires are what the revenue line depends on, and the revenue begins after they arrive rather than alongside them. And the hiring date entered the cash flow as month four, which is what produced the month-five trough that chapter 8 uses to derive the ask. Her account manager's comment, when the second version arrived, was that it was the first plan she had read that year in which the revenue and the payroll appeared to have been written by the same person.

6.5 Location, Equipment and What It Costs to Start

6.5.1 Premises

State what the business occupies, on what terms, and for how long. Owned or leased, the square footage, the monthly cost, the expiry date and the renewal options. If a move or an expansion is contemplated within the plan's horizon, say when and what it would cost.

A lease expiry inside the forecast period is a material fact and belongs in the risk register of chapter 9 as well as here. → see Book J9 (Business Premises: Lease, Buy and Manage Your Space).

6.5.2 Equipment and the start-up cost list

List the equipment the business needs to operate at the forecast volume, separating what it already owns from what the plan will acquire. For anything being acquired, the plan needs a quote from a named supplier, dated, in an appendix — because the total of this list becomes a capital cost in chapter 7 and a line in the use of funds in chapter 8.

For the running case, the acquisition list is short and each line is quoted: a service vehicle with its upfit at $78,000 and technician tooling at $14,000, totalling $92,000, which is exactly the term loan portion of the financing request. A reader can trace the ask to two invoices, which is the standard to aim for.

6.5.3 What people forget

Three costs are omitted from start-up lists with near-perfect regularity. Installation and commissioning, which can be a significant fraction of an equipment price. The deposit and the first months of everything: lease, insurance, utilities, subscriptions, all payable before revenue. And the working capital, which is not equipment at all but is the reason businesses with fully funded equipment still fail in month five — the subject of chapter 7.

6.5.4 The eight categories of a start-up cost list

A start-up cost list that is organized by category is harder to leave holes in, and it maps directly onto the use-of-funds table in chapter 8. The eight below cover nearly every small business; work through them in order and write a zero where a category does not apply, because a written zero is a decision and a missing line is an oversight.

• Premises: deposit, first month, fit-out, signage, keys and access systems.
• Equipment and vehicles: purchase or lease, delivery, upfit, installation, commissioning, and the licenses and registrations that go with a vehicle.
• Tools and small equipment: the items individually too small to capitalize that collectively are not, which for a trade is frequently five figures.
• Technology: computers, phones, network, software subscriptions for the first year, and the setup labor that nobody quotes for.
• Professional fees: incorporation, contracts, the accountant's setup, the insurance broker, permits and licenses.
• Insurance: the first premium, which is payable before you begin rather than after.
• Initial inventory and consumables, where applicable, plus the first supplier deposit if terms are not yet established.
• Working capital: enough cash to cover the gap between paying for work and being paid for it, computed in chapter 7 rather than guessed here.

Two rules make the list usable rather than decorative. Every line above a threshold you set — a thousand dollars is a reasonable one — carries a dated quote from a named supplier in an appendix. And the total of this list is the same number that appears in the use of funds in chapter 8; if the two differ, the plan has two start-up budgets and a reader will find the difference.

6.6 The Team Section: Roles Before Names

6.6.1 Write the roles first

Describe the roles the business requires, then say who fills each one. This ordering matters because it exposes the roles that are unfilled or that one person is filling twice, which is the information a reader is looking for.

For each role: what it does, what it requires, who holds it today, and what it costs. Where one person holds three roles, say so plainly. Where a role is unfilled, say when it will be filled and out of what.

6.6.2 The gap is the point

Every small business has a gap. A plan that shows none is not more impressive; it is less credible, and it also forfeits the opportunity to explain what the financing is for.

The running case names its gap directly: there is no dispatcher, the owner is carrying that role, and it is the first role the plan funds. That sentence connects the team section, the capacity arithmetic and the financing request in a way that three paragraphs of biography could not.

6.6.3 Biographies, and how long

Two to four sentences per person, relevant to this business. Years in the trade, scope of prior responsibility, credentials that matter. Full résumés go to an appendix. → see Book I2 (Organizational Design, Roles, Delegation and Accountability) for designing the roles themselves, and Book I3 (Recruiting, Hiring and Onboarding) for filling them.

6.6.4 What a role actually costs

The team section states costs, and a plan that states wages rather than loaded costs will understate payroll by a fifth or more. The loaded cost of an employee is the wage plus everything the employer pays on top of it, and the components are not optional.

Four categories make up the difference. Statutory employer contributions, which in Canada and Québec include the public pension plan, employment insurance, parental insurance, the health services fund and workers' compensation, each at its own rate on its own base — the rates change annually and must be taken from the current year's official tables rather than from memory. Paid time not worked: vacation, statutory holidays and any sick allowance, which is why §6.3 subtracts those hours from availability and why they must also be added to the cost. Direct costs of employment: tools, a vehicle allowance or seat, uniforms, a phone, equipment, training. And the supervision the role consumes, which is real even when nobody invoices for it.

The practical instruction for a plan is to build one loaded-cost line per role and use it consistently everywhere: in the capacity calculation, in the cost of sales, and in the acquisition cost of chapter 5. A plan that prices the owner's selling time at a bare hourly wage in one section and at a loaded rate in another has two payroll models, and a reader who spots it will ask which one the forecast used.

Two further habits are worth adopting. Take the contribution rates from the current year's official tables, at the tax authority's own page, and record the date read — the rates are among the most volatile facts in a plan. And state which costs are in your cost of sales and which are in overhead, because the gross margin in chapter 7 depends entirely on where that line is drawn, and moving it silently between years is the easiest way to produce a margin trend that does not exist. → see Book G5 (Business Taxes in Québec and Canada) for the contributions themselves and Book I4 (Compensation, Benefits and Performance Management) for what to pay.

6.7 The Founder's Own Time, Honestly Budgeted

6.7.1 The hours that are not available

The most consistently optimistic number in small-business plans is the founder's availability. A founder who sells, delivers, dispatches, quotes, invoices and manages has less time for each of those than any single line of the plan assumes.

Write the founder's week as a budget: hours on delivery, hours on selling, hours on administration, hours on management, hours on everything else. Total it. If it exceeds fifty-five or sixty hours, the plan contains an assumption that will fail, and the honest response is to change the plan rather than to hope.

6.7.2 What this does to capacity

A founder's field hours belong in the capacity calculation, and they should be discounted for the interruptions that only the founder receives. In the running case, the owner's 30% field time is already the discounted figure; the raw availability was higher and was reduced deliberately because dispatch calls arrive during field work.

6.7.3 The succession sentence

Every plan involving a business with fewer than about ten people should contain one sentence about what happens if the principal is unavailable for six weeks. Not a succession plan — a sentence. It is the question every reader has and almost no plan answers.

WHO DOES THIS Who does the work when the founder is the bottleneck In a solo business, the honest answer is that nothing happens, and the plan should say so and then say what reduces the exposure: a written procedure that lets somebody else answer the phone, a colleague with a reciprocal arrangement, or a deliberate decision to accept the risk with insurance behind it. In a small team, the constraint is usually a specific task rather than the person: quoting, or dispatch, or the one technical skill nobody else has. Identify the task, not the person, and the fix becomes obvious — document it, train a second person, or buy it from outside. In an established business, the risk migrates to relationships rather than skills. The sales relationship that lives in one person's head is the exposure, and the mitigation is a customer record that somebody else could act on, which is an operational discipline rather than a document. In every case, the sentence that belongs in the plan is the same shape: if X is unavailable for six weeks, Y happens, and Z is what we have done about it. Readers do not expect the answer to be reassuring. They expect it to exist.

6.8 Advisors, Board and Governance in a Small Company

6.8.1 Proportionate governance

Governance means: who decides what, and what is written down. In a three-person company that is a short section, and it should be short. Claiming a board that meets quarterly when three people have lunch occasionally is a false statement in a document you are asking someone to rely on.

The paragraph below introduces Table 6.1, which describes governance at three sizes. Its columns give who decides, what is written down, what meets and how often, and the failure mode characteristic of that size. Find your row and write two sentences from it; anything more is aspiration.

Table 6.1 — Governance at three sizes: who decides, what is written, what meets

6.8.2 The advisory board

An informal advisory board — three or four people who know something the business does not — is genuinely useful and costs little. If one exists, name the members and what each contributes. If one is planned, say who you are approaching and what for.

The distinction from a statutory board of directors matters and should be stated: an advisory board has no legal authority and no fiduciary duty; a board of directors does. Do not blur the two in a document a lender will read. → see Book H6 (Corporate Governance, Shareholder Agreements and Minute Books).

6.8.3 Professional advisors

Name the accountant, the lawyer or notary, the insurance broker and the banker, with the firm. It signals that the business has the relationships a lender expects it to have, and it takes four lines. Chapter 12 covers what each of them actually delivers.

6.9 Milestones: The Section That Becomes the Execution Plan

6.9.1 The five columns

A milestone entry has five parts, and it is worth being strict about all five: the milestone, the date, the owner (a person, not a department), the deliverable (a thing that exists afterwards), and the verification criterion (an observable fact that settles whether it happened).

The fifth column is the one that changes the document from a wish list into a plan, and it is the one most often left out. Project-management practice treats a milestone as a significant point or event in a project rather than as a piece of work (Project Management Institute, 2021), which is why a milestone carries a date and a deliverable while the effort behind it belongs to the tasks underneath.

THE RULE A milestone without a date and a verification criterion is an intention Both are required, and the verification criterion must be observable by somebody other than the person who did the work. "Marketing plan complete" is not a criterion; "the campaign calendar exists and the first three sends have gone out" is. This is the same discipline that chapter 10 applies to tasks, and it is the same discipline the product described in In La Machine enforces at the moment that matters most: a task offered without a verification criterion is refused at creation, and the assistant refuses to close one whose criterion is empty and offers to drop it instead. Writing milestones this way in chapter 6 means chapter 10 is transcription rather than invention.

The paragraph that follows introduces Table 6.2, which gives the milestones of the running case. Its columns are the five required parts, and the rows are ordered by date. Read the verification column on its own: every entry there is something a third party could confirm without asking anybody's opinion.

Table 6.2 — The milestones table: milestone, date, owner, deliverable, verification criterion

METHOD Writing the milestones table in six moves First, list the events that must happen for the forecast to be achievable, not the events you would like. The test is whether the revenue line depends on it. Second, put a date on each one, and make the dates months rather than quarters, because a quarter is a place for a milestone to hide. Third, assign one named person to each, never two and never a department. Shared ownership of a milestone is unowned. Fourth, write the deliverable as a noun — a thing that will exist. If you cannot name a thing, the milestone is an activity rather than a milestone. Fifth, write the verification criterion as an observable fact, checkable by someone else. Read it aloud and ask whether two people could disagree about whether it had happened. Sixth, check the dependencies between rows. If milestone D requires milestone B, D's date must be after B's, and a reader will check. The deliverable is a table of six to twelve rows. The verification criterion for the exercise itself is that every row could be pasted into a task list tomorrow without further editing.

6.9.2 How many milestones

Six to twelve for a three-year plan. Fewer than six and the plan is not specific enough to execute; more than twelve and the list contains activities rather than milestones, and nobody will track it.

6.10 Operations, Team and Governance in Three Situations

6.10.1 Solo

Your capacity calculation has one person in it, and the honest version subtracts far more than a plan usually does: administration, selling, quoting, invoicing, travel and the interruptions that only you receive. A utilization rate above 60% for a solo service business is unusual and should be defended if you claim it.

The governance section is three sentences, and the succession sentence of §6.7.3 is the one that matters.

6.10.2 Small team

The interesting question at this size is which single role, if added, would raise the ceiling most — and it is frequently not the obvious one. Adding a person who does the work raises capacity by their hours. Adding a person who removes administrative load from three others can raise capacity by more, and costs less. Compute both before deciding, and show the computation in the plan.

6.10.3 Established business

Capacity is measured by department and the bottleneck moves, which is worth stating: last year's constraint may not be this year's. Governance here is real and should be described accurately, including delegation of authority thresholds, because a lender will want to know who can commit the company to what.

TOOLS OF THE TRADE What an operations section needs from your tools Four capabilities produce the numbers in this chapter, independent of product. You need time recorded against work, at least at the level of billable versus non-billable, because the utilization rate is the input the whole calculation turns on and it cannot be guessed. You need a record of what was delivered and when, so that lead time and response commitments can be measured rather than asserted. You need a supplier and dependency list that is maintained, since the value of a dependency list is that it is current at the moment something fails. And you need milestones held somewhere they are seen weekly, because a milestone table that lives only in the plan document is a milestone table nobody checks. Chapter 13 names the products; chapter 10 describes the cadence.

WHAT CHANGED Utilization expectations have not moved, but measurement has The realistic utilization rates for field service, trades and professional services have been broadly stable for years, and they remain well below what owners assume. What has changed is how cheaply a small business can measure its own rate: work now leaves a timestamped trail through scheduling, dispatch and invoicing systems that a business of four people can afford, which means a rate can be computed from records rather than estimated. A plan written now has no good reason to assume a utilization rate, and a reader increasingly expects the number to be measured. As of September 2026.

Costly Mistakes

1. Writing a forecast that exceeds the described capacity. It is the most common internal contradiction in small-business plans, it is found in fifteen minutes by any careful reader, and it discredits every other number in the document.
2. Assuming a utilization rate instead of measuring it. A rate that is five points too high inflates the revenue ceiling by roughly the same proportion, which is enough to turn a workable plan into an impossible one.
3. Forgetting that included work consumes capacity. A preventive schedule, a warranty, a support commitment — all consume hours and produce no incremental revenue, and leaving them out of the calculation overstates what can be sold.
4. Hiring capacity in the same month the revenue starts. Payroll begins immediately and revenue arrives after the sales cycle; the gap between them is the cash trough, and a plan that does not show it will be surprised by it.
5. Omitting the person who is a single point of failure. Every reader assumes there is one. Naming it costs nothing and what you have done about it earns credibility.
6. Claiming a board that does not meet. It is a false statement in a document a lender relies on, and it is trivially disproved by asking when it last met.
7. Writing milestones without verification criteria. They become a list of intentions that nobody can settle, and chapter 10 then has to invent them under time pressure.

Key Takeaways

• The operations section exists to prove that the revenue in the financial section is deliverable by this specific company.
• Capacity is computed in four moves: available hours, measured utilization, subtraction of non-incremental work, and multiplication by price.
• Statutory holidays and annual leave are set by law and belong in the available-hours calculation; in Québec there are eight holidays in total, seven from the labour standards Act and the National Holiday.
• A business operating at its capacity ceiling cannot grow through sales effort, and discovering this is often the most valuable finding of the planning exercise.
• The forecast must sit below the capacity ceiling with room to spare, and where financing raises the ceiling, the revenue must begin after the capacity arrives.
• Dependencies include suppliers, subcontractors, platforms and people, and customer concentration is a dependency in exactly the same sense.
• Roles are written before names, and the gap that every small business has should be named rather than hidden, because it is usually what the financing is for.
• A milestone has five parts, and the verification criterion — observable by someone other than the doer — is what turns the table into an execution plan.

Your Action Plan

1. Compute your capacity ceiling this week. Deliverable: the four-move calculation on your own numbers, ending in a revenue ceiling. Timing: ninety minutes. Verification criterion: the utilization rate used is measured from last year's records, not assumed.
2. Compare the ceiling with your forecast and reconcile the two. Deliverable: a one-line statement of the forecast as a percentage of the ceiling, for each year. Timing: fifteen minutes. Verification criterion: no forecast year exceeds 90% of the ceiling that year's resources support.
3. Write the delivery model as steps and name the bottleneck. Deliverable: six to ten steps with who, how long and what it depends on, plus one sentence naming the constraint. Timing: two hours. Verification criterion: someone who does the work agrees with the step list.
4. List your dependencies and compute your customer concentration. Deliverable: a table of three to eight dependencies with switching cost and alternative, plus the largest customer and top three as a percentage of revenue. Timing: one hour. Verification criterion: the concentration figures are computed from the actual revenue ledger.
5. Write the roles before the names, and name the gap. Deliverable: a role list with holder, cost and one sentence per gap saying when and how it is filled. Timing: one hour. Verification criterion: at least one gap is named, and it connects to something in the financing request.
6. Write the milestones table with all five columns. Deliverable: six to twelve rows, dated by month, each with a named owner, a deliverable and an observable criterion. Timing: two hours. Verification criterion: every row could be pasted into a task list tomorrow without editing.

In La Machine

Every route named in this section was verified in the product's route inventory on 2026-09-09 (La Machine, 2026a), and what each page holds is described in the product's user guide (La Machine, 2026b).

• Milestones become projects and tasks. /projets holds projects — each one either a chain or a one-off — and /taches holds the tasks, each carrying an owner, a due date and a verification criterion. No page creates a plan task or a plan project: both arrive through the programming interface, a script or the assistant, and a task offered without a criterion is refused at that moment. The drawer, afterwards, will let a criterion be saved empty, and the Mark as done button does not check that one exists, so the discipline §6.9.1 asks of a milestone is enforced at the door rather than at the finish line.
• The six statuses of a task. A task is To do, In progress, Active (recurring), Completed, Later or Dropped, and two of the six are earned rather than chosen: Active (recurring) and Completed are set by the product when a run is recorded. A project is active, on hold, done or archived; the Projects page does not display that status, and a project does not become done by calculation — it is a status change somebody makes.
• Dependencies between tasks. A task blocked by a dependency will not queue itself until the task it depends on is Completed; a pivotal task that is dropped, or one that is recurring and therefore never completed, leaves its dependants blocked. No page creates or removes a dependency either, which is worth knowing before you build a chain.
• The equipment and tools inventory. /outils holds an inventory in which each tool carries one of six statuses, from To evaluate to Active, and in which the version and the folder of a locally installed tool are measured on the machine rather than declared; its summary also counts the tools no task uses. productivity/tools is a different thing: a calculator, a stopwatch, timers and a timesheet.
• Where the daily operating picture lives. /aujourdhui gathers what needs a decision now — runs awaiting approval, overdue tasks, postponed tasks whose date has arrived, failures and blocked tasks — and it gives the tasks due today as a count without a list. /executions holds the runs, two hundred at a time. /tableau-de-bord aggregates the plan by status, wave, owner, automation and project, with no filter of any kind and no export, and it counts a recurring task as done once it has run once.
• What it does not do. La Machine does not compute capacity, does not model utilization, and holds no scheduling or dispatch board. It does not hold a supplier list as such, and it does not produce an organization chart. The capacity arithmetic is done in the tools of chapter 13; what La Machine holds is the milestone table, once it has become projects and tasks.

Self-Check

1. Give the four moves of the capacity calculation, and name the one where optimism most often enters. Say what it should be replaced with.
2. A service business computes 3,100 billable hours and forecasts revenue that requires 3,600. Name the two possible fixes and say which one is honest, and why.
3. Your largest customer is 31% of revenue. State what you would write in the plan about it, in two sentences, including the mitigation.
4. Turn this milestone into a compliant entry: "improve our customer reporting." Supply a date, an owner, a deliverable and an observable verification criterion.
5. Explain, in one sentence each, why included work must be subtracted from capacity and why hiring in the same month revenue starts creates a cash problem.

Chapter References

Act respecting labour standards, CQLR c N-1.1, ss. 60, 66–69 (Québec). LégisQuébec, consolidated version updated to April 7, 2026. Read September 9, 2026.

Commission des normes, de l'équité, de la santé et de la sécurité du travail. (2026). Statutory general holidays and annual leave. Retrieved September 9, 2026.

International Organization for Standardization. (2015). ISO 9001:2015 — Quality management systems — Requirements, clauses 8.1 and 8.5 (operational planning and control; production and service provision). ISO.

La Machine. (2026a). Documentation — route reference inventory (211 routes: 191 Core, 20 Pilotage paths). Internal product documentation, read September 9, 2026.

La Machine. (2026b). Guide utilisateur [User guide] (FAQ: projects and task chains; tasks and their statuses; the tools inventory). Internal product documentation, read September 9, 2026.

National Holiday Act, CQLR c F-1.1 (Québec). LégisQuébec. Read September 9, 2026.

Project Management Institute. (2021). A guide to the project management body of knowledge (PMBOK guide) (7th ed.). PMI.

Chapter 7 — Financial Projections

This is the longest chapter in the book and the one a reader turns to first when they have already decided the business makes sense. It is also where more plans disqualify themselves than anywhere else — not through bad arithmetic, which is rare, but through three specific structural failures: a revenue line built from a growth percentage instead of from drivers, a cash flow that is really an income statement with a different title, and a set of statements that do not tie to one another. This chapter builds all of it, on one continuous case, with every number written out.

What You'll Be Able to Do

• Write an assumption log in which every input carries a value, a source, a confidence level and a consequence if wrong.
• Build a sales forecast from drivers — volume, price, capacity — rather than by applying a growth rate to a previous line.
• Produce an income statement, a cash flow forecast and an opening balance sheet that tie to each other.
• Separate fixed, variable and step costs, and say in which month each step is taken.
• Compute break-even three ways — in dollars, in units of production and in months — and read the margin of safety.
• Run a sensitivity analysis and three scenarios, each attached to a single decision rather than to a mood.

Why It Matters

There is a specific failure this chapter exists to prevent, and it has a shape that anybody who has watched small businesses will recognize. The business is profitable on paper in month seven. It runs out of money in month five. Both statements are true simultaneously, because profit and cash are different things measured at different times, and a plan that models only one of them will not see the collision coming.

The second reason is about credibility rather than survival. A lender reading a financial section is performing three checks that take about eight minutes: does the revenue trace to something countable, do the three statements tie, and does the cash ever go below zero. A plan that passes those three checks is in a different category from one that does not, and passing them is a matter of construction rather than of optimism.

The third reason is that the assumption log built here is what makes the plan reviewable a year from now. A forecast that is wrong tells you nothing. A forecast that is wrong, beside a list of the assumptions that produced it, tells you exactly which belief about the world needs revising — and that is the mechanism by which a business gets better at forecasting.

In This Chapter

• 7.1 What the Financial Section Must Contain
• 7.2 Assumptions: The Only Part That Is Really Yours
• 7.3 The Sales Forecast, Built From Drivers
• 7.4 The Income Statement, Line by Line
• 7.5 Cost Structure: Fixed, Variable and the Ones That Move Late
• 7.6 The Cash Flow Forecast, and Why It Is Not the Income Statement
• 7.7 Working Capital: Receivables, Payables and Inventory
• 7.8 The Opening Balance Sheet
• 7.9 Break-Even, Three Ways
• 7.10 Sensitivity: What Happens If You Are Wrong
• 7.11 Presenting Numbers So They Are Believed
• 7.12 Financial Projections in Three Situations

7.1 What the Financial Section Must Contain

Seven things, in this order.

The assumptions, as a table, before any statement. The sales forecast, built from drivers and reconciled to capacity. The income statement, by year for three years and by month for year one. The cash flow forecast, by month for year one and by quarter or year afterwards. The opening balance sheet, and a projected one at each year end. Break-even, with the margin of safety. Sensitivity and scenarios, with the decision each scenario triggers.

The monthly detail may live in an appendix, and several lenders and programs require it there — Futurpreneur, for example, requires financial forecasts covering two years with a month-by-month breakdown before an application is reviewed (Futurpreneur Canada, 2026). Whether it sits in the body or in an appendix, it must exist, because a yearly forecast cannot show a cash trough and the trough is the point. → see Book A5 (Financing a Business) for the public programs themselves — Futurpreneur among them — and for what each one asks of an applicant.

DEFINITION Revenue, cash and profit: three different things Revenue is what you have earned, recognized when the work is done or the goods are delivered, whether or not you have been paid. Cash is what is in the account, which moves when money actually arrives or leaves. Profit is revenue minus expenses over a period, which is an accounting result rather than a bank balance. A business can be profitable and out of cash, which is the most common way a growing small business fails: growth consumes cash before it produces profit, because you pay for the work before the customer pays you. A plan that models only profit has modeled the half that does not kill you.

7.1.1 The three checks a reader performs

Knowing what a reader checks lets you check it first.

Does the revenue trace to something countable? They will look for a volume and a price, and if they find a growth percentage instead, they will discount the whole forecast.

Do the statements tie? Net income must flow into the balance sheet's retained earnings; the closing cash on the cash flow must equal the cash on the balance sheet; the amortization in the income statement must match the change in accumulated amortization. Three checks, four minutes.

Does the cash ever go below zero? If it does and there is no financing shown, the plan has an unfunded hole and the conversation is over. If it does and the financing is shown arriving in the right month, the plan has done its job.

Figure 7.1 shows how one set of assumptions becomes three statements. The first column holds the assumptions — the driver, the price, the utilization, the collection period, the hiring month, the capital cost. The second turns them into an income statement: what was earned. The third turns the same assumptions into a cash flow: what actually moved, and when, including the two items that never touch profit at all — the capital purchase and the loan principal. The fourth is the balance sheet. The band across the bottom names the three ties a reader checks, which is the fastest way to find out whether a model was built or assembled.

Figure 7.1 — Original figure. From one assumption log to three statements: what each column adds, and the three ties that must hold between them.

7.2 Assumptions: The Only Part That Is Really Yours

7.2.1 Why the log comes first

Every number in the financial section is derived from a small set of inputs. Twelve to twenty assumptions typically drive an entire small-business model, and everything else is arithmetic.

Writing them down before building anything has three effects. It makes each one discussable separately, which is what allows a disagreement to be settled. It makes the model updatable, because changing an input changes everything downstream. And it makes the plan reviewable later, which is the value that outlives the financing.

THE RULE Every number in the financials traces to an assumption with a name and a source Follow the chain backwards from any figure in any statement and it must arrive at a row in the assumption log. If it does not, one of two things is true: the number was typed directly into a statement, which means it will not update when the world changes, or it was derived from something that is itself untraceable. Both are defects and both are found the same way — by picking three numbers at random and following them back. Do this to your own model before a reader does it to you. The rule has a corollary that saves more time than the rule itself: no hard-coded numbers inside formulas. A formula containing 0.68 rather than a reference to the utilization cell is an assumption hiding where nobody will find it.

7.2.2 The five columns

The assumption, named as a noun phrase. The value, with its unit. The source, which may legitimately be "our own estimate" as long as it says so. The confidence, as high, medium or low. What changes if it is wrong, in one clause — which is the column that tells you what to test in §7.10.

The paragraph that follows introduces Table 7.1, the assumption log of the running case. Its columns are the five above, and the rows are ordered from most to least consequential. Read the confidence column beside the last column: the assumptions that are both low-confidence and high-consequence are exactly the ones the sensitivity analysis must cover, and in this case there are three of them.

Table 7.1 — The assumption log: value, source, confidence, and what changes if it is wrong

One row of that table deserves a caution the table itself cannot carry. Where the loan will be made under a public guarantee program rather than on the lender's own terms, it is the program's published guidelines, and not the lender alone, that set what the loan may finance and what it may cost, so the rate row should be checked against them before it is written down (Innovation, Science and Economic Development Canada, 2026a). → see Book A5 (Financing a Business), which reads that program's parameters, including the ceilings it places on the interest rate of a guaranteed term loan and of a guaranteed line of credit.

7.2.3 What makes an assumption defensible

An assumption is defensible when a reader can see where it came from and how it would be tested. Three forms qualify: a measurement from your own records, with the period stated; an external figure with its source and date; and an explicit estimate, labeled, with the method in one clause.

The fourth form — an industry benchmark quoted from an article — is the one to avoid, for a reason that is practical rather than snobbish: you cannot update it. When reality differs, you have no way to tell whether the benchmark was wrong or your business is different.

7.3 The Sales Forecast, Built From Drivers

7.3.1 The driver method

A driver is the countable thing that produces revenue. Find it, forecast it, multiply by price.

For a service business the driver is usually customers or billable hours. For a retailer it is transactions and average basket. For a subscription business it is subscribers, churn and average revenue per subscriber. For a project business it is projects won and average project value.

METHOD Building a sales forecast from drivers, in seven steps Step one: name the driver. One countable thing. If you need two — a recurring line and a transactional line, as in the running case — build two forecasts and add them; do not average them into one. Step two: state the starting level. How many of the driver exist today, from a record rather than from memory. Step three: state the additions and the losses separately. New customers per month and lost customers per month are two different assumptions with two different causes; netting them hides both. Step four: apply the timing. A customer signed in month three at your cycle length starts producing revenue in month three or later — never earlier — and a monthly forecast is where that becomes visible. Step five: multiply by price, and state whether price changes during the horizon. Step six: reconcile against capacity. Take the resulting volume to the capacity ceiling of chapter 6. If it exceeds it, the forecast is wrong or the capacity plan is incomplete, and the fix belongs here rather than in a footnote. Step seven: sanity-check the growth rate you did not choose. Compute the implied year-over-year growth. If it is far above your history and nothing structural has changed, one of the driver assumptions is optimistic; find which. The deliverable is a monthly forecast for year one and an annual one for years two and three, with the driver visible on its own row. The verification criterion is that changing the driver by one unit changes the revenue by exactly the price.

7.3.2 The running case, built

The figures below are fictional and belong to the running case.

The business has two revenue lines and therefore two drivers.

Driver one: buildings under contract. The register starts the year at 34. Eleven are signed and four are lost, ending at 41. Because the signings are spread across the year, the average number under contract during year one is 37, and it is the average that produces revenue. Contract revenue: 37 × $685 × 12 = $304,140.

Driver two: billable intervention hours. The capacity model of chapter 6 produces 2,900 sellable intervention hours in year one, once the two technicians arrive in month four and the preventive schedule is subtracted. Intervention labor revenue: 2,900 × $96 = $278,400.

The dependent line: parts. Parts revenue has historically run at about 42% of intervention labor: 278,400 × 0.42 = $116,928, which the model rounds to $118,000 on the strength of a small assumed price increase.

Total year-one revenue: $304,140 + $278,400 + $118,000 = $700,540, which the plan states as $700,500.

Years two and three repeat the exercise with averages of 44 and 50 buildings and 3,450 and 3,800 intervention hours, giving $832,900 and $933,800.

The implied growth rates are 14.4% and 12.1%, against a historical rate of about 9%. That difference is not an accident and the plan says so explicitly: it is the second vehicle, arriving in month one of year one, and the two technicians arriving in month four. Growth above history requires a named cause, and here the cause is on the balance sheet.

7.3.3 Seasonality, and why an even twelfth is usually wrong

A yearly revenue figure divided by twelve is an assumption, and for most businesses it is a false one. Seasonality shows up in the cash flow rather than in the annual result, which is exactly why a plan that models only annual figures never sees it.

Three patterns are worth distinguishing. Demand seasonality is the customer's cycle: retail concentrated in the final quarter, construction in the warm months, professional services flat but interrupted by two slow weeks in summer and two at year end. Operating seasonality is yours: the months when your own people are away, when equipment is serviced, when the shop closes. And collection seasonality is the customer's payment behavior, which in many sectors slows in the same months that demand peaks, compounding the cash effect rather than offsetting it.

Model seasonality with a simple monthly index — twelve factors averaging 1.0 — applied to the annual driver rather than to revenue directly, so that volume and price remain separable. Where you have history, compute the index from your own last two or three years. Where you do not, state the pattern you are assuming and its basis in one clause.

For the running case the pattern is mild and it is real: intervention hours run about 15% above the yearly average in January and February, when heating systems fail, and about 20% below it in July, when two technicians take vacation and much of the client base is quiet. That variation moves no annual figure at all, and it moves the month-five trough by roughly $4,000 — enough to matter when the trough is $31,400 and the facility covering it is $40,000.

7.3.4 Reconciling to capacity, in one line

The forecast must be shown against the ceiling. Year three at $933,800 is 88.5% of the post-investment ceiling of $1,054,808 computed in chapter 6. Putting that single sentence in the plan pre-empts the most common lender question about a growth forecast, and it demonstrates that the two sections were written by the same person.

7.4 The Income Statement, Line by Line

7.4.1 The structure

An income statement for a small business has five blocks: revenue, cost of sales, gross profit, operating expenses, and the result after interest and tax. Keep the line count low — fifteen to twenty-five lines total — and put the detail in a schedule.

DEFINITION Gross margin, contribution margin, EBITDA Gross margin is revenue minus cost of sales, expressed in dollars or as a percentage of revenue; cost of sales includes everything that varies with production, and where you draw that line determines the percentage, which is why it must be drawn once and held. Contribution margin is revenue minus variable costs only, and it is the figure break-even uses — it differs from gross margin whenever some cost of sales is fixed in the short run, which for a service business with salaried technicians is most of it. EBITDA is earnings before interest, taxes, depreciation and amortization; it approximates operating cash generation and is the figure a lender uses to size debt capacity, which is why it is worth showing even in a small plan.

7.4.2 The running case, year one

Fictional figures; the structure and the arithmetic are the point.

The table below gives the year-one income statement line by line. Its two columns are the line and its amount, and its rows follow the five blocks in order: revenue, cost of sales, gross profit, operating expenses, and the result after interest and tax. Read the two bold subtotals first — gross profit at 39.0% of revenue, and net income of $34,300 — then read upward to see what produced each of them.

Table 7.2 — The income statement, year one, line by line

7.4.3 The three-year view

The body of a plan shows three years side by side, because a single year says nothing about direction. The paragraph that follows introduces Table 7.3, which gives the running case's income statement across the historical year and the three forecast years. Its columns are the periods; its rows are the five blocks in summary form. Read the two percentage rows: gross margin and net margin are what a reader compares across the columns, and a movement in either needs a cause stated in the same table.

Table 7.3 — Income statement summary: the year just ended and three forecast years

Three features of that table are worth imitating. The historical column sits beside the forecast, in the same units, so that the reader can judge the forecast against something real rather than against nothing. The gross margin improves gradually rather than in a jump, from 38.3% to 41.0%, and the plan attributes it to a single named cause: a rising share of contract revenue, which carries a better margin than parts resale. And interest rises in year one and falls afterwards, which is what a term loan does — a detail that costs nothing to model correctly and that a reader notices when it is missing.

7.4.4 The tax line, treated honestly

Tax in a projection is an estimate and should say so. Two facts anchor it. The federal net tax rate on active business income eligible for the small business deduction is 9%, against a general federal rate of 15% (Canada Revenue Agency, 2025). Provincial rates are additional and vary by province, with their own eligibility conditions, so the combined rate must be taken from the current tables for your own province rather than assumed.

Two further points belong in the plan rather than in a footnote. Sales taxes are not revenue and not an expense: they are collected and remitted, and they matter to the cash flow rather than to the income statement — the threshold at which registration becomes mandatory for most businesses is $30,000 of taxable supplies over four consecutive calendar quarters (Canada Revenue Agency, 2026b), and a business crossing it mid-year has a cash timing event, not a profit event. And the owner's compensation must appear, whether as salary or as a stated draw; a forecast in which the owner works for nothing shows a profit that does not exist. → see Book G5 (Business Taxes in Québec and Canada) and Book G10 (Owner Compensation, Holdings and the Entrepreneur's Personal Finances).

7.5 Cost Structure: Fixed, Variable and the Ones That Move Late

7.5.1 Three kinds, not two

Most plans separate fixed from variable and stop. The third kind is where the surprises live.

Variable costs move with each unit sold: parts, materials, shipping, transaction fees, subcontracted labor paid by the job.

Fixed costs do not move with volume within the relevant range: rent, insurance, salaried administration, software, the owner's salary.

Step costs are fixed until a threshold, then jump: an additional technician, a second vehicle, a larger unit, the person you must hire at a certain volume. They are the reason a business can be more profitable at 80% of capacity than at 105%, and they belong in the plan with the month the step is taken.

The paragraph below introduces Table 7.4, which sorts the running case's costs into the three kinds and — for the step costs — names the month the step occurs. Read the third column: every step cost has a trigger, and knowing the trigger is what turns a cost structure into a decision rule.

Table 7.4 — Fixed, variable and step costs, with the month each step is taken

7.5.2 Which costs go where

The boundary between cost of sales and operating expenses is a choice, and the only rule that matters is that it is made once and applied consistently across all years. Moving a cost between the two changes the gross margin without changing the net result, and a margin that improves because a cost was reclassified is the kind of thing a reader notices.

State your convention in one sentence beneath the income statement: "cost of sales includes technician payroll, parts at cost, vehicle operating costs and shop supplies; all other costs are operating expenses."

7.6 The Cash Flow Forecast, and Why It Is Not the Income Statement

7.6.1 The four differences

A cash flow forecast differs from an income statement in four specific ways, and each one is a place where a business gets surprised.

Timing of receipts. Revenue is recognized when earned; cash arrives when the customer pays. A business selling on 30-day terms whose customers actually pay in 47 days has a month and a half of revenue sitting outside its bank account at all times.

Timing of payments. Some costs are paid before the period they cover — insurance, subscriptions, deposits — and some after.

Non-cash expenses. Amortization reduces profit and moves no money. In the running case it is $21,000 a year, which means cash generation is $21,000 better than the profit line suggests.

Items that never touch the income statement. Capital purchases, loan principal repayments, drawings, and sales tax collected and remitted all move cash without appearing in profit. The second of those is the one most often forgotten: a term loan payment is part interest — an expense — and part principal, which is not.

KEY CONCEPT Profit is an opinion; cash is a fact The phrase is old and it earns its place because it is literally true. Profit depends on accounting choices: when revenue is recognized, how assets are amortized, how costs are classified. Cash depends on nothing except whether money moved. This is why a lender reads the cash flow before the income statement, and why a plan whose cash flow is a lightly edited income statement will be recognized as such in about thirty seconds. The test is simple: if your cash flow has no line for capital purchases, no line for loan principal, and receipts equal to revenue in the same month, you have written the income statement twice.

7.6.2 Building it monthly

Year one is monthly. Anything less cannot show a trough, and the trough is the reason the forecast exists.

The structure has three blocks. Operating: receipts from customers, payments to suppliers and staff, sales tax remitted, interest and tax paid. Investing: capital purchases and disposals. Financing: loan advances, principal repayments, capital injections and drawings. Each month ends with an opening cash balance, the net movement, and a closing balance that becomes the next month's opening.

DO THE MATH The month the business is profitable and out of cash Fictional figures from the running case; the mechanism is the point. The year opens with $18,600 in the account. Month 1. The second vehicle and its upfit are paid: −$92,000. The term loan of $92,000 is drawn the same month, so the two offset. Operating receipts reflect December's billing at 47 days: roughly $52,000 in, $49,300 out. Closing cash: 18,600 + 2,700 = $21,300. Months 2 and 3. Nothing has changed operationally; the new vehicle is in service but no additional technicians exist yet. Monthly operating surplus is about $2,400, less the term loan payment of $1,798, of which about $1,304 is principal in the first month. Closing cash at month 3: approximately $22,900. Month 4. The two technicians start. Payroll rises by $14,300 a month. The revenue they generate has not arrived: the sales cycle is eleven weeks and the receivable adds another 47 days, so work sold in month 4 is collected in month 6 at the earliest. Closing cash: −$4,100 before the operating line. Month 5. A second full month of the higher payroll, still without the matching receipts. Net operating outflow of about $27,300. Closing cash: −$31,400. This is the trough. Month 6. The first receipts from the expanded capacity arrive. Net inflow of about $9,800. Closing cash: −$21,600. Months 7 to 12. The gap closes at roughly $11,000 a month. The account returns to positive during month 9 and closes the year at approximately $39,700. What this shows, and why it is the whole point. The business is profitable in every month from month six onward and its worst cash position occurs in month five, before it has earned a dollar of the new revenue. Nothing in the income statement predicts month five. The trough of $31,400 is what the operating line requested in chapter 8 exists to cover, and the request is sized at $40,000 rather than $31,400 because a facility sized exactly to the forecast trough has no room for the forecast to be wrong.

7.6.3 Six errors that produce a wrong cash flow

Each of the following is common, each is mechanical, and each changes the trough by enough to matter.

Booking receipts in the month of the sale. If your customers pay in 47 days, roughly half of a month's revenue is collected in the following month and the rest the month after. Model the lag explicitly, using your measured collection period rather than your invoice terms.

Forgetting the sales tax float. Tax collected on sales sits in the account until it is remitted, which flatters cash between collection and remittance and then removes it in one movement on the filing date. A business that treats the float as its own money discovers otherwise on the remittance date.

Modeling the loan payment as interest. A term loan payment is principal plus interest. Only the interest is an expense; the principal is a cash outflow that never appears in the income statement, and omitting it overstates cash by the principal portion every single month.

Omitting the capital purchase. It appears in the balance sheet and in amortization, and it moves cash on the day it is paid. A cash flow with no investing section has not modeled it.

Paying payroll monthly in a business that pays every two weeks. Twenty-six pay periods do not divide evenly into twelve months: two months of the year carry three pay runs, and in a payroll-heavy business those two months are frequently the trough.

Assuming supplier terms you do not have. New suppliers often require payment in advance or on delivery for the first several orders. If the plan assumes 30-day terms from day one, the trough is deeper than modeled by the amount of those first orders.

7.6.4 What to show in the body

Show the monthly closing cash balance as a small table or a single line chart, with the trough labeled and the financing shown. That one exhibit answers the third of the reader's three checks, and it is worth more than three pages of narrative about the strength of the market.

A trough and a capital purchase are financed by two different instruments, and matching the instrument to the need is a decision this book records rather than makes. → see Book A5 (Financing a Business) for the difference between a term loan, which buys a long-lived asset and is paid down on a schedule, and an operating line, which finances the cycle and is meant to return to zero within it.

7.7 Working Capital: Receivables, Payables and Inventory

7.7.1 The cycle

Working capital is the money tied up in operating the business: what customers owe you, plus what is sitting in inventory, minus what you owe suppliers. It grows with revenue, which is why growth consumes cash.

DEFINITION Working capital, DSO, burn rate, runway Working capital is current assets minus current liabilities; operationally, it is the cash tied up in receivables and inventory, less the credit your suppliers extend to you. Days sales outstanding (DSO) is the average number of days between invoicing and collection, computed as receivables divided by revenue, times the number of days in the period. Burn rate is the net cash consumed per month while the business is cash-negative. Runway is the number of months the current cash balance and available facilities can sustain the burn rate, and it is the single number that determines how much time a plan has to be right.

7.7.2 Computing the receivable drag

The arithmetic is short and it changes financing requests more than any other calculation in this chapter.

At 47 days of sales outstanding, the business has 47 ÷ 365 = 12.9% of a year's revenue sitting in receivables at any moment. On year-one revenue of $700,500, that is 700,500 × 0.129 = $90,400 permanently outside the bank account.

Now the growth effect. Revenue rises from $612,100 to $700,500, an increase of $88,400. At the same 47 days, the receivable balance rises by 88,400 × 0.129 = $11,400, which is cash the business must find purely because it grew. Nobody sends an invoice for that amount, which is precisely why it is missed.

The lever that matters. If collections improved from 47 days to 38 days, the receivable balance would fall from $90,400 to 700,500 × (38 ÷ 365) = $72,900 — releasing $17,500 of cash, which is more than half the entire trough. In the running case this is the finding that reduced the financing request in the second version of the plan: the first version asked for a larger operating line, and the second asked for less because the collections assumption was tightened and the tightening was made a milestone with an owner. → see Book G6 (Invoicing, Payments and Collections).

7.7.3 Payables and inventory

Supplier credit works in the opposite direction: every day of payment terms you receive is a day of financing your suppliers are extending. State your actual terms rather than the terms on the invoice, since the two often differ.

Inventory, where it exists, is cash on a shelf. State the turns and the value; a business holding 90 days of inventory has three months of purchases sitting still, and it is often the largest single lever on a small business's cash position. → see Book J5 (Supply Chain, Procurement and Inventory).

Where the receivable or the equipment is to be financed rather than collected faster or bought outright, that is a financing decision rather than a working-capital one. → see Book A5 (Financing a Business) for factoring and invoice financing, and for leasing equipment instead of purchasing it.

7.8 The Opening Balance Sheet

7.8.1 What it shows and why it is required

The balance sheet states what the business owns, what it owes, and what remains for the owners, at a moment. It is required for one reason above all others: it is the only statement that shows leverage, and leverage is what determines whether a lender can advance anything at all.

Three blocks: assets, liabilities and equity. Assets equal liabilities plus equity, always, which is also the arithmetic check that catches most modeling errors.

7.8.2 The running case, at the start of year one

Fictional figures.

The table that follows gives the opening balance sheet. Its two columns are the line and its amount, and its three blocks are assets, liabilities and equity, in that order. The line to check first is the last one: total liabilities and equity must equal total assets exactly, and that identity is the arithmetic check that catches most modeling errors.

Table 7.5 — The opening balance sheet at the start of year one

Two ratios follow immediately and a lender will compute both. The current ratio is current assets divided by current liabilities: (18,600 + 78,900 + 14,200) ÷ (41,700 + 9,800) = 111,700 ÷ 51,500 = 2.17. The debt-to-equity ratio is total liabilities divided by equity: 79,900 ÷ 93,200 = 0.86. After drawing the $92,000 term loan, debt-to-equity moves to (79,900 + 92,000) ÷ 93,200 = 1.84, which is the number the lender is actually deciding on. Putting the post-financing ratio in the plan yourself is a small act that signals you know what is being assessed. → see Book G4 (Financial Analysis, Ratios and KPIs), and Book A5 (Financing a Business) for what the lender attaches to the advance once it is made: the security it takes, the personal guarantee it asks of the owners and the levers for negotiating it, and the coverage covenant it tests each year.

7.8.3 The projected balance sheets

Show a projected balance sheet at each year end. The tie is what matters: closing cash from the cash flow, retained earnings equal to opening retained earnings plus net income less any dividends, and accumulated amortization increased by the year's amortization. If those three do not tie, the model has an error, and finding it now is much cheaper than having a reader find it.

7.9 Break-Even, Three Ways

7.9.1 Why three

Break-even answers "how much do we need to sell to cover our costs," and the answer is more useful in three units than in one, because each unit speaks to a different reader. Dollars speak to a lender. Units of production speak to whoever schedules the work. Months speak to everybody.

DO THE MATH Break-even in dollars, in hours and in months Fictional figures from the running case. Step one: separate variable from fixed. Truly variable costs are parts at cost ($85,000) and shop supplies ($8,400), plus the variable portion of vehicle costs, estimated at $14,000 of the $34,000 total. Variable costs: 85,000 + 8,400 + 14,000 = $107,400 — but note that technician payroll is not variable here: the technicians are salaried and paid whether or not the hours are sold, which is the correct treatment for a short-run break-even and is the point most plans get wrong. Step two: total the fixed costs. Technician payroll ($299,900), the fixed portion of vehicle costs ($20,000), operating expenses ($228,000), and interest ($5,800), less amortization ($21,000) if you want a cash break-even. For the accounting break-even, keep amortization in: 299,900 + 20,000 + 228,000 + 5,800 = $553,700. Step three: the contribution margin ratio. Revenue of $700,500 less variable costs of $107,400 gives a contribution of $593,100, which is 593,100 ÷ 700,500 = 84.7% of revenue. Break-even in dollars. 553,700 ÷ 0.847 = $653,700 of annual revenue. Break-even in billable hours. Contribution per billable intervention hour is 593,100 ÷ 2,900 = $204.52. Fixed costs divided by that: 553,700 ÷ 204.52 = 2,707 hours, against 2,900 forecast. Break-even in months. At an even revenue rate of 700,500 ÷ 12 = $58,375 a month, break-even is reached at 653,700 ÷ 58,375 = month 11.2. The margin of safety. (700,500 − 653,700) ÷ 700,500 = 6.7%. In hours: (2,900 − 2,707) ÷ 2,900 = 6.7%, which is the same figure expressed differently and is a useful check that the two calculations agree. How to read 6.7%. It means revenue can fall by two-thirds of one month before the year produces a loss. That is thin, and the plan says so rather than hiding it — which is exactly why the sensitivity analysis in §7.10 exists and why the scenario that matters is the low one.

7.9.2 Cash break-even and why it differs

Amortization is a real expense and not a cash payment. Removing it gives the cash break-even: 553,700 − 21,000 = 532,700, divided by 0.847 = $628,900, reached at month 10.8. A business can be below accounting break-even and still cash-positive, which is common in the first year after a capital purchase and is worth stating explicitly so a reader does not mistake a small accounting loss for a cash problem.

7.10 Sensitivity: What Happens If You Are Wrong

7.10.1 One variable at a time

Sensitivity analysis moves one assumption and holds the rest, which tells you which assumption the result is most exposed to. Do it for the three or four assumptions marked medium or low confidence in the log.

For the running case, moving each of four assumptions in the unfavourable direction gives the following effects on year-one net income of $34,300. Utilization falling from 68% to 64% removes 170 billable hours, which at $96 plus the associated parts costs about $23,200 of revenue and roughly $20,100 of net income after tax — leaving $14,200. The average contract price falling 5% removes $15,200 of revenue and about $13,200 of net income — leaving $21,100. Signing 8 contracts instead of 11 lowers the average building count from 37 to 35.5 and removes about $12,300 of revenue and $10,700 of net income — leaving $23,600. Collections slipping from 47 days to 60 days does not change net income at all and deepens the trough by about $25,000, which is the point: the assumption with the largest cash consequence has no profit consequence whatsoever.

That last line is the single most useful output of a sensitivity analysis, and it is why the exercise must be run against both statements rather than only against profit.

7.10.2 Scenarios, and the decision each one triggers

A scenario moves several assumptions together into a coherent picture. Three is the right number, and each must be attached to a decision — otherwise a scenario is a mood.

The paragraph that follows introduces Table 7.6, the three scenarios of the running case. Its columns give what moves in each scenario, the resulting revenue and net income, and — the operative column — the single decision that scenario triggers, with the observable signal that says the scenario is happening. Note that each trigger is something visible by a specific month, which is what makes the scenarios usable rather than decorative.

Table 7.6 — Three scenarios, the numbers they produce, and the one decision each triggers

7.10.3 Presenting scenarios without three plans

Present the base case in full and the other two as a summary table with the decision attached. A plan containing three complete sets of statements is three times as long and no more informative; the reader wants to know that you have thought about being wrong and what you would do, not to read the arithmetic three times.

CASE IN POINT Nordec Entretien: profitable in month six, out of cash in month five This case is fictional; the company, the people and the figures are invented. The first version of Sophie Nadeau's plan contained an annual income statement, an annual cash flow that was the income statement with amortization added back, and no balance sheet. It showed a profitable year one. It was declined. The second version was built monthly, and the monthly view produced the finding that changed everything: the business went cash-negative in month four and reached −$31,400 in month five, two months before it became reliably profitable. The cause was not the vehicle, which the term loan funded on the same day it was paid for. The cause was the hiring date. Two technicians started in month four; their work was sold on an eleven-week cycle and collected 47 days after invoicing, so the payroll preceded the matching receipts by roughly four months. Three changes followed. The plan added an operating line to cover the trough, sized at $40,000 rather than at the trough itself, because a facility sized exactly to the forecast leaves no room for the forecast to be wrong. Collections became a milestone with an owner and a verification criterion — days sales outstanding under 40 by month nine — because the arithmetic in §7.7.2 showed that nine days of improvement released $17,500, more than half the trough. And the hiring date became a decision point rather than a fixed date: the low scenario defers it by three months on an observable signal, which is fewer than four signings by month five. The lender's comment was that it was the first version in which the risk had a number attached to it.

7.11 Presenting Numbers So They Are Believed

Six habits make a financial section read as competent, and all six are mechanical.

State the units once, at the top, and never mix them. Whole dollars or thousands, chosen once. A statement that mixes them is the fastest way to lose a reader's trust in the arithmetic.

Round consistently and do not over-round. Whole dollars for a business this size; a forecast to the cent implies a precision that does not exist, and a forecast rounded to the nearest ten thousand hides the arithmetic.

Make the statements tie, and say that they do. A short note — "closing cash on the cash flow agrees with cash on the balance sheet at each year end" — tells the reader you checked.

Label every figure with its period and its basis. "Year 1" is ambiguous; "Year ending December 31, 2027, forecast" is not.

Show the comparison. Historical actuals beside the forecast, in the same table, is worth more than any narrative about credibility.

Explain every discontinuity. A line that jumps must carry a note saying why, in the same table, in six words. A jump without an explanation invites the reader to supply one.

WATCH OUT The forecast that grows 40% a year because the previous line did Applying a growth percentage to a revenue line is the single most recognizable defect in a financial section, and readers spot it in seconds because the resulting numbers have a signature: they are smooth, they have no relationship to any operational quantity, and they continue rising past the point where the business would have run out of people or space. The corrective is not to lower the percentage — it is to delete it and build the line from a driver. If, having built it from drivers, the implied growth rate is high, that is fine and it is now defensible, because you can name the cause and point at the capacity that supports it.

QUÉBEC & CANADA NOTE The tax and threshold facts a projection must get right Four facts anchor the tax lines of a Canadian projection, and all four change, so all four carry a date. The federal net tax rate on active business income eligible for the small business deduction is 9%, against a general federal rate of 15%; provincial rates are additional and vary, each with its own eligibility conditions (Canada Revenue Agency, 2025). The small supplier threshold for GST/HST registration is $30,000 of taxable supplies over four consecutive calendar quarters for most businesses, with registration required no later than the day of the supply that exceeds it, and a business has 29 days from its effective registration date to register (Canada Revenue Agency, 2026b). Québec's QST operates on a parallel threshold and must be verified with Revenu Québec. And the Bank of Canada's policy interest rate was raised to 2.50% on October 28, 2026, with the next scheduled announcement on December 9, 2026 (Bank of Canada, 2026) — which matters because a floating-rate loan assumption should state the base rate, the spread and the date, rather than a single number that will be stale before the plan is read. → see Book G5 (Business Taxes in Québec and Canada).

WHAT CHANGED Rate assumptions need a base and a date, not a number Interest rate assumptions written as a single figure age badly and are difficult to update. Write them as a base plus a spread, with the date of the base: "prime plus 2.0%, prime at the rate posted by our lender on the date of this plan." The plan then remains readable when the base moves, and a reader can substitute the current base themselves. The same practice applies to any assumption tied to a published rate. With the policy rate raised to 2.50% on October 28, 2026 and the next announcement scheduled for December 9, 2026, a plan written today should state which side of that date it was written on. As of October 29, 2026.

7.12 Financial Projections in Three Situations

7.12.1 Solo

Build the model in a spreadsheet with an assumptions tab, and keep it to about a dozen assumptions. A free financial plan template published by a development bank is a reasonable place to start, because it arrives with the statements already laid out and spares you the work of building the skeleton before you can test anything (Business Development Bank of Canada, 2026c). The two things a solo model most often gets wrong are the owner's compensation, which must appear even if it is not being drawn, and the receivable timing, which for a solo business is frequently the entire difference between a workable year and a stressful one.

Run the monthly cash flow even if nobody asks for it. It takes an afternoon and it is the only thing that will show you your own trough.

7.12.2 Small team

The model should be built by whoever will maintain it, which is not always the person who is best at spreadsheets. A model nobody but the owner understands is a model that stops being updated the first busy month.

Have the model reviewed by somebody who reads statements for a living before the plan is delivered. Chapter 12 describes what to ask for; the short version is a review of the model's structure and its ties, which is a much smaller engagement than a review of the plan.

7.12.3 Established business

Reconcile the forecast to actuals explicitly, in a table showing three historical years beside three forecast years. Any discontinuity between the last actual and the first forecast year needs a named cause in the same table.

At this size the risk is a model that has grown by accretion over several years, with assumptions embedded in formulas that nobody remembers making. The assumption log is the remedy, and rebuilding a legacy model around one is usually a two-day job that pays for itself the first time somebody asks where a number came from.

WHO DOES THIS Who builds the numbers, and what a CPA adds The owner owns the assumptions, always, because they are business judgments rather than accounting ones. Nobody else can decide how many contracts you will sign. A bookkeeper or accountant owns the structure: that the statements tie, that the classifications are consistent, that the tax and payroll lines are treated correctly. This is a small, well-defined engagement and it is the highest-value hour a small business plan can buy. A CPA adds three specific things beyond structure: an opinion on whether the assumptions are within a defensible range for your sector, the correct treatment of anything unusual — a capital purchase, a lease, a related-party transaction — and, where required, an assurance engagement on the historical statements, which is a different service at a different level and is described in chapter 12. The reporting framework those historical statements are prepared under, and the levels of assurance available on them, are set by the profession's published standards for private enterprises rather than by the accountant's preference (CPA Canada, 2026b). Nobody should own the model who cannot explain it. If the person who built it cannot walk a lender through the cash flow line by line, the plan will not survive the meeting, regardless of how correct it is.

TOOLS OF THE TRADE What a forecasting tool must do Independently of any product, a financial model for a plan must do six things. It must keep assumptions in one place, separate from the statements, so that changing one input changes everything downstream. It must show its formulas, since a model whose logic is hidden cannot be reviewed by an accountant. It must produce monthly output for at least the first year, because that is where the trough lives. It must tie the three statements automatically, so that an error announces itself instead of hiding. It must support scenarios without duplicating the model, ideally by switching an assumption set. And it must be sendable, since a lender will ask for it. A spreadsheet does all six, which is why it remains the default; chapter 13 names the alternatives and their pricing models, dated.

Costly Mistakes

1. Building revenue from a growth percentage. The result has a recognizable signature — smooth, unrelated to any operational quantity, and continuing past the point where the business runs out of people — and it discredits the section in seconds.
2. Writing a cash flow that is the income statement with amortization added back. It has no capital purchases, no loan principal and receipts equal to revenue, and it cannot show a trough, which is the only reason it exists.
3. Omitting the owner's compensation. A forecast in which the owner works for nothing shows a profit that does not exist and will not survive the first question.
4. Treating salaried production staff as a variable cost in break-even. They are paid whether or not the hours are sold, and treating them as variable understates break-even substantially.
5. Ignoring the receivable growth that comes with revenue growth. Growing revenue by $88,400 at 47 days ties up an extra $11,400 that nobody invoices you for, and it is a frequent cause of an otherwise well-planned trough.
6. Sizing an operating line exactly to the forecast trough. A facility sized to the forecast has no room for the forecast to be wrong, which is the one thing forecasts reliably are.
7. Running sensitivity only on profit. The assumption with the largest cash consequence — collection days — often has no profit consequence at all, and a profit-only sensitivity will miss it entirely.

Key Takeaways

• The financial section contains seven things in order, and the monthly detail must exist somewhere even when it lives in an appendix.
• Twelve to twenty assumptions drive an entire small-business model; the log comes before any statement and every figure must trace back to it.
• A sales forecast is built from drivers — a countable thing times a price — with additions and losses stated separately and timing applied.
• Profit and cash differ in four ways: receipt timing, payment timing, non-cash expenses, and items that never touch the income statement at all.
• Step costs are the third kind of cost and the one that produces surprises; every step belongs in the plan with the month it is taken.
• Break-even in dollars, in units of production and in months are three expressions of one fact, and the margin of safety is what tells you how thin the year is.
• Working capital grows with revenue: at 47 days, every dollar of new annual revenue ties up about thirteen cents that nobody sends you an invoice for.
• Sensitivity must be run against both statements, because the assumption with the largest cash effect may have no profit effect whatsoever.

Your Action Plan

1. Write the assumption log before opening a statement. Deliverable: a five-column table with twelve to twenty rows, ordered by consequence. Timing: two hours. Verification criterion: no row has an empty source cell, and at least three rows are marked medium or low confidence.
2. Build the revenue line from drivers, monthly, for year one. Deliverable: a monthly forecast with the driver on its own row, additions and losses separate. Timing: three hours. Verification criterion: changing the driver by one unit changes revenue by exactly the price.
3. Reconcile the forecast to the capacity ceiling of chapter 6. Deliverable: one sentence per forecast year stating the forecast as a percentage of that year's ceiling. Timing: twenty minutes. Verification criterion: no year exceeds 90%.
4. Build the monthly cash flow with all three blocks. Deliverable: twelve months with operating, investing and financing, ending in a closing balance each month. Timing: four hours. Verification criterion: the cash flow contains capital purchases and loan principal, and receipts lag revenue by your actual collection period.
5. Compute break-even three ways and the margin of safety. Deliverable: break-even in dollars, in your unit of production, and in months, plus the margin of safety as a percentage. Timing: one hour. Verification criterion: the margin of safety computed in dollars and in units gives the same percentage.
6. Run sensitivity on your three lowest-confidence assumptions, against both statements. Deliverable: the effect of each on net income and on the cash trough. Timing: ninety minutes. Verification criterion: at least one assumption shows a large cash effect and a small profit effect, or you can explain why none does.
7. Write the three scenarios with a decision and a signal each. Deliverable: a table with what moves, the result, the decision and the observable signal with its month. Timing: one hour. Verification criterion: each signal is something you could check on a specific date.

In La Machine

Every route named in this section was verified in the product's route inventory on 2026-09-09 (La Machine, 2026a), and what each page holds is described in the product's user guide (La Machine, 2026b).

• The productivity tools, and what they are not. productivity/tools holds a calculator, a stopwatch, timers and a timesheet. It holds no financial calculator of any kind: there is no return on investment, no break-even, no net present value, no payback and no amortization tool in the product, and the arithmetic of this chapter is done elsewhere. /outils is the separate inventory of what the business runs on, in which each tool carries one of six statuses, from To evaluate to Active, and in which the version and the folder of a locally installed tool are measured on the machine rather than declared.
• What the business bills and is billed — and what is not there. account/billing and account/transaction-detail hold what you owe La Machine for using it, not the invoices you send your own customers. The product has no commercial quotation page at all: business/quotes holds motivational quotations, which is a different sense of the word, and there is no invoice, no proposal, no order, no cart and no price list anywhere in the route inventory. business/subscription-packages defines the subscription packages you sell, with a price, a cost and a billing period that is monthly or annual; business/subscriptions is the revenues-and-expenses view that reads the sales recorded against them.
• Where the operating numbers are read. /tableau-de-bord aggregates the plan — tasks by status, wave, owner, automation and project — with no filter of any kind, no period and no export, so it is not a place to read a financial figure. /aujourdhui gathers what needs a decision now. The measures that matter to a plan — contracts signed, contracts lost, hours delivered — are read from the operating screens rather than computed by the product.
• Assumptions that become decisions. An assumption that turns out to be wrong produces a decision, and arbitrages is where a decision that cannot be settled in the moment is recorded rather than lost; the page sits with the administration pages rather than in a domain group of the menu.
• What it does not do. La Machine does not build financial projections. It does not produce an income statement, a cash flow forecast or a balance sheet; it does not consolidate accounting data; it holds no model, no scenarios and no break-even calculation. The model is built in the tools of chapter 13 and reviewed by the professionals of chapter 12. What La Machine holds is the operating record from which next year's assumptions will be measured rather than guessed.

Self-Check

1. A plan shows revenue rising 22% a year for three years. Name the two things you would ask for before accepting the line, and say what a driver-built version would look like instead.
2. Explain, in two sentences, how a business can be profitable in month six and out of cash in month five. Name the two timing effects that produce it.
3. Your technicians are salaried. Are they a fixed or a variable cost for a one-year break-even calculation? Justify the answer and say what happens to break-even if you get it wrong.
4. Revenue grows from $612,100 to $700,500 at 47 days sales outstanding. Compute the additional cash tied up in receivables, and say who invoices you for it.
5. Give the three checks a lender performs on a financial section, in order, and say roughly how long they take.

Chapter References

Bank of Canada. (2026). Policy interest rate (target for the overnight rate raised to 2.50% on October 28, 2026; next scheduled announcement December 9, 2026). Retrieved October 29, 2026.

Business Development Bank of Canada. (2026c). Free financial plan template for entrepreneurs. Retrieved September 9, 2026.

Canada Revenue Agency. (2025). Corporation tax rates (net federal rate of 9% on active business income eligible for the small business deduction; general net federal rate of 15%; page modified May 30, 2025). Retrieved September 9, 2026.

Canada Revenue Agency. (2026b). When to register for and start charging the GST/HST (small supplier threshold of $30,000 in taxable supplies over four consecutive calendar quarters; registration within 29 days of the effective date; page modified June 16, 2026). Retrieved September 9, 2026.

CPA Canada. (2026b). Financial reporting and assurance standards for private enterprises. Retrieved September 9, 2026.

Futurpreneur Canada. (2026). Frequently asked questions (two-year financial forecasts with a month-by-month breakdown required before an application is reviewed). Retrieved September 9, 2026.

Innovation, Science and Economic Development Canada. (2026a). Canada Small Business Financing Program — Program guidelines. Retrieved September 9, 2026.

La Machine. (2026a). Documentation — route reference inventory (211 routes: 191 Core, 20 Pilotage paths). Internal product documentation, read September 9, 2026.

La Machine. (2026b). Guide utilisateur [User guide] (FAQ: dashboard and reports; the tools inventory). Internal product documentation, read September 9, 2026.

Chapter 8 — Financing Request and Use of Funds

There is one sentence that determines how the rest of this section is read, and it is the sentence stating the amount. If the amount is a round number, the reader concludes that it was chosen and the plan fitted to it. If the amount is an odd number that falls out of a cash curve, the reader concludes the opposite — and everything that follows is read as arithmetic rather than as advocacy. This chapter is about producing the second kind of sentence.

What You'll Be Able to Do

• Derive the financing requirement from the cash curve, rather than choosing an amount and justifying it afterwards.
• Build a use of funds table in which every dollar is assigned, timed, and attached to what it buys.
• Name what each source of money wants in return, and choose a mix rather than a lender.
• Read a term sheet's essential terms — rate, term, amortization, security, covenants, guarantees — and say what each one commits you to.
• Cite public programs accurately, with their published parameters and the date you read them.
• Answer the exit question honestly, including when the honest answer is that there is no exit.

Why It Matters

Two failures dominate this section, and both are expensive in ways that are not obvious.

The first is asking for too little. A business that asks for the trough and gets it has no margin for the forecast to be wrong, and forecasts are wrong. Returning to a lender in month seven to ask for more is a far worse conversation than asking for the right amount in month zero, because the second request arrives with evidence that the first plan was inaccurate.

The second is signing terms that were never read. A personal guarantee survives the company; a covenant can put a loan in default because a ratio moved for a reason unrelated to your ability to pay. Neither is unusual and neither is unreasonable — but a business owner who discovers a guarantee after signing has learned something at the worst possible moment.

Beneath both is a structural point. The financing request is the section where the whole plan gets tested for internal consistency, because the amount must be traceable back through the cash flow, the operations plan and the sales forecast. A reader who can follow that chain will trust the document; one who cannot will suspect that the sections were written independently, which is usually true when the chain does not hold.

In This Chapter

• 8.1 Asking for a Number You Can Defend
• 8.2 How Much: Building the Requirement From the Cash Curve
• 8.3 What For: The Use of Funds Table
• 8.4 The Financing Mix, and What Each Source Wants
• 8.5 What a Lender Reads, in Order
• 8.6 What an Investor Reads, and Why It Differs
• 8.7 Terms, Covenants and Personal Guarantees
• 8.8 Government Programs and Guarantees, With Their Dates
• 8.9 The Exit Question, Answered Honestly
• 8.10 The Financing Request in Three Situations

8.1 Asking for a Number You Can Defend

8.1.1 The direction of the derivation

There are two directions in which a financing request can be produced, and only one of them survives contact with a reader.

The wrong direction starts with an amount — a figure you have heard is obtainable, a round number, what a competitor borrowed — and works backwards to a justification. The signature of this direction is a request that is round, an amount that appears in the plan before the cash flow does, and a use of funds table whose lines have been sized to total the pre-chosen figure.

The right direction starts with the monthly cash flow, finds the point at which cash is lowest, adds what is being purchased and a buffer, and arrives at an amount. The signature of this direction is an odd number.

KEY CONCEPT The amount is an output of the plan, never an input Everything else in this chapter follows from that sentence. The financing request is the last thing computed and the last thing written, because it depends on the capital cost from chapter 6 and the cash trough from chapter 7, both of which depend on the forecast, which depends on the market and go-to-market work. A request written before those exist is a guess wearing a decimal point. The practical test: can you state, in one sentence, the calculation that produced your number? If the sentence is "the equipment costs X, the cash trough is Y, and we have added Z as a buffer for a stated reason," the request is defensible. If the sentence begins "we thought," it is not.

8.1.2 Where the request appears in the document

Twice. In full, as its own section after the financials, and in one sentence in the executive summary. The two must match exactly, including the split between instruments, because a discrepancy between the summary and the section is the kind of error that makes a reader check everything else.

8.2 How Much: Building the Requirement From the Cash Curve

8.2.1 The three components

A financing requirement has three components and they are financed differently, which is why they must be separated rather than totalled into one number.

The capital component is what you are buying: equipment, vehicles, leasehold improvements, systems. It is financed by a term loan or a lease, over a period matched to the asset's useful life.

The working capital component is the trough: the cash consumed by the gap between paying for work and being paid for it. It is financed by an operating line, because it is temporary and recurring rather than permanent.

The buffer is the amount by which you exceed the forecast requirement, so that the facility still works if the forecast is wrong. It is not padding, and it should be stated as a deliberate percentage with a reason.

Matching the instrument to the component matters more than most owners realize. Financing equipment on an operating line consumes the facility that working capital needs; financing working capital with a term loan means paying principal on money you needed only for six months.

DO THE MATH The financing requirement, derived from the cash curve Fictional figures from the running case. Component one — capital. Two dated supplier quotes: a service vehicle with its upfit at $78,000 and two technician tool sets at $14,000. Capital requirement: $92,000. Component two — the trough. The monthly cash flow of chapter 7 reaches its lowest point in month five at −$31,400, computed with the term loan already drawn and applied to the vehicle. That is the working capital requirement. Component three — the buffer. The plan adds $8,600, taking the operating line to $40,000. The reason is stated rather than implied: the sensitivity analysis showed that collections slipping from 47 days to 60 would deepen the trough by about $25,000, and the low scenario shows a further deterioration. A buffer of 27% over the forecast trough covers roughly a third of that deterioration, and the plan says so plainly rather than claiming the buffer makes the business safe. The request. $92,000 as a five-year term loan, plus a $40,000 operating line: $132,000 in total. The coverage check a lender will perform. Year-one EBITDA is $66,200. The term loan at the assumed rate — prime plus 2.0%, with prime assumed at 4.45% as posted by major Canadian lenders in early September 2026, which is an assumption to confirm rather than a published figure — amortized over sixty months, gives a monthly payment of $1,798, or $21,576 a year. Adding roughly $1,400 of interest on the operating line gives total debt service of about $22,976. Debt service coverage: 66,200 ÷ 22,976 = 2.88. Lenders commonly look for at least 1.25, so this file has room — and stating the ratio yourself, with the rate assumption named, saves the reader the arithmetic and shows you know what is being assessed.

Figure 8.1 draws the derivation. The upper panel is the monthly cash curve across year one, with the trough at month five marked and the zero line visible. The three bands beneath it show, in proportion, the three components of the request: the capital, the trough, and the buffer, adding to the total. The point of putting them in one view is that a reader can see the request and the curve that produced it without turning a page, which is what makes the number feel derived rather than chosen.

Figure 8.1 — Original figure. From the cash curve to the ask: the month-five trough, the three components of the request, and the instrument matched to each.

8.2.2 Sizing the buffer

There is no rule, and there are two defensible methods.

The scenario method sizes the buffer at the difference between the base-case trough and the low-scenario trough, which is what a stress test would produce. It is the more rigorous method and it usually produces a larger number.

The percentage method adds a stated proportion — twenty to thirty percent is common — and states the reason. It is less rigorous and it is transparent, which is why the running case uses it and says what it does and does not cover.

What is not defensible is a buffer that appears without explanation, because a reader cannot distinguish it from padding.

8.2.3 What to do when the requirement exceeds what is obtainable

Sometimes the honest requirement is larger than the business can borrow. Three responses are legitimate, and all three belong in the plan rather than being discovered later.

Reduce the requirement by changing the plan: delay a hire, buy used equipment, lease rather than purchase, tighten collections. The running case's second version did exactly this — the collections milestone released $17,500, which is more than half the trough.

Stage the request, taking part now against a milestone that unlocks the rest. Lenders are frequently more comfortable with a staged facility than with a single large advance, because the milestone gives them information.

Change the source, which usually means accepting different terms rather than finding a more generous lender.

8.2.4 Four errors in the derivation

The arithmetic is short, and four errors in it recur.

Double-counting the term loan. The trough must be computed with the term loan already drawn and the equipment already paid for, because that is the world you are asking about. A trough computed before the term loan is a different and much larger number, and adding a term loan to it produces a request roughly twice the real requirement. The running case is explicit about this: the −$31,400 is measured with the $92,000 drawn and spent.

Taking the trough from an annual figure. A year that ends with cash of $39,700 tells you nothing about month five. Only the monthly view produces a trough, which is why chapter 7 insists on it.

Forgetting that the line must also be repaid. An operating line is not a permanent addition to working capital; a lender expects it to return to zero at some point in the cycle, and many require it explicitly. The plan should show the month it returns to zero — month twelve, in the running case — because a line that never clears is really a term loan with the wrong label.

Ignoring the timing of the advance. Funds are available after the conditions precedent are met, not on the day of approval. If your model has the vehicle paid in month one, check that the approval process realistically permits it; if it does not, the trough moves and so does the request.

8.3 What For: The Use of Funds Table

8.3.1 The five columns

A use of funds table has five columns: the category, the amount, the timing, what it buys stated as a capability rather than as an object, and what will prove it was spent well. The last column is unusual and it is the one that most improves the section, because it converts spending into a commitment.

The paragraph below introduces Table 8.1, the use of funds for the running case. Read the fourth and fifth columns together: each line says what capability the money produces and what observable fact would demonstrate that it did. Note that the total is an odd number and that it equals the request exactly — a use of funds table that does not sum to the request is the most easily avoided error in this chapter.

Table 8.1 — Use of funds: amount, timing, what it buys, and what proves it was spent well

8.3.2 What not to put in a use of funds table

Three categories weaken the table and should be handled differently.

"General working capital" with no explanation is the least persuasive line a request can contain, because it is where padding is expected to hide. If working capital is genuinely needed, say what consumes it — payroll during a cycle, inventory build, a receivable gap — and for how long.

"Contingency" as a line item invites the reader to remove it. The buffer belongs inside the operating line with its reason stated, not as a separate row.

Owner's salary, unless it is a genuine part of a start-up budget in a business with no revenue. In an operating business, the owner's salary is an operating expense funded by revenue, and putting it in a use of funds table reads as financing the owner rather than the business.

8.3.3 Own contribution

Every lender will ask what the business or the owner is contributing. State it explicitly: cash injected, equipment already owned and dedicated to the project, retained earnings being applied, work already done and paid for.

The running case contributes the existing vehicle and tooling — carried at $61,400 net of accumulated amortization — and $93,100 of retained earnings already in the business. Naming both means the reader does not have to compute the equity contribution from the balance sheet, and it makes the leverage discussion of §8.5 shorter.

8.4 The Financing Mix, and What Each Source Wants

8.4.1 Sources want different things

Every source of money wants something, and the something is different in each case. Matching your situation to what a source wants is more productive than approaching sources in order of preference.

One source is easy to overlook because it is neither a chartered bank nor a public program. The Business Development Bank of Canada, the federal development bank, lends to small and medium-sized businesses and provides advisory services alongside the financing (Business Development Bank of Canada, 2026b). → see Book A5 (Financing a Business) for its statutory mandate to complement the commercial institutions rather than compete with them, for the criteria it publishes for its own small-business loans, and for its role as the co-lender behind the startup route of §8.8.2.

The paragraph that follows introduces Table 8.2, which compares the sources available to a small business. Its columns give what each source wants, what it costs, what it takes from you beyond the cost, and what it is typically used for. Read the third column first: it is where the real difference between sources lies, and it is the column most owners consider last.

Table 8.2 — Financing sources: what each wants, what it costs, what it takes

8.4.2 The cheapest money is usually not from a lender

Two sources cost little and are systematically underused by small businesses.

Supplier terms are financing. Moving from payment on delivery to net 30 with a supplier who represents a meaningful share of purchases funds part of the working capital requirement at no interest. It is a negotiation rather than an application.

Customer terms work the same way in reverse. A deposit, a monthly billing cycle instead of on-completion billing, or an annual prepayment at a small discount can each remove a large part of a trough. In the running case, moving preventive contracts from arrears billing to billing on the first of the month would shift roughly $23,000 of receipts earlier in the cycle — which is a change of billing practice rather than a loan.

Both belong in the plan, because a reader who sees that you have exhausted the cheap options before asking for the expensive one reads the request differently. → see Book A5 (Financing a Business) for the financing ladder that ranks every one of these sources by the control it takes, and Book G6 (Invoicing, Payments and Collections) for the collection practice.

8.5 What a Lender Reads, in Order

8.5.1 The five things

A commercial lender assessing a small business file looks at five things, and knowing them lets you put each one where it will be found.

Cash flow coverage. Can the business service the debt from operations, with room? The measure is usually a debt service coverage ratio, and the threshold is commonly around 1.25, though it varies by institution and by sector.

Security. What is the lender's recourse if the business cannot pay? Equipment, receivables, inventory, real property, and — very often for a small business — a personal guarantee.

The operator. Track record, credit history, and the impression given by the document itself. This is less quantifiable and it is real.

Leverage. Debt relative to equity, before and after the advance. A business asking to move from 0.86 to 1.84, as the running case does, is in ordinary territory; one asking to move to 4 is having a different conversation.

Purpose and repayment source. What the money buys and what will repay it. A loan for equipment that increases capacity has a visible repayment source; a loan to cover losses does not.

8.5.2 The ratios worth computing yourself

Three ratios are computed by every lender and can be computed by you in ten minutes. Debt service coverage: EBITDA divided by total annual debt service, which for the running case is 66,200 ÷ 22,976 = 2.88. Debt to equity: total liabilities divided by equity, post-advance, which is 1.84. Current ratio: current assets divided by current liabilities, which is 2.17.

Put all three in the plan with their arithmetic visible. It takes four lines and it changes the tone of the first meeting, because the conversation starts from agreed numbers.

8.5.3 What weakens a file, and what to do about it

Four things weaken a file predictably. Thin equity, addressed by an owner contribution or by staging the request. Customer concentration, addressed by naming it with the mitigation. A recent loss year, addressed by explaining the cause and showing what changed. Personal credit problems, which cannot be argued away and should be disclosed early rather than discovered.

The general principle: a weakness disclosed with its mitigation is a manageable fact; the same weakness discovered by the lender is a credibility problem on top of a financial one.

8.5.4 What actually happens to your document

Understanding the process changes what you send and when.

The first meeting is a conversation, not an assessment. The account manager is establishing what the business is and whether the request fits their institution's appetite. Bring the one-page summary and the request; do not bring forty pages.

The package follows: the plan, the statements, the model, the quotes, the register extract, personal financial information for the guarantors, and whatever the institution's checklist adds. Nothing moves until it is complete, which is why §8.3's checklist matters more than its individual items.

The analysis is where your document is read by somebody you will not meet. They are reconstructing your cash flow, testing your ratios, and looking for the internal contradictions this book has spent seven chapters preventing. This stage produces questions, and the speed of your answers is itself information about you.

The decision comes with conditions, and the conditions frequently matter more than the amount: security to be registered, statements to be delivered on a schedule, covenants to be met, guarantees to be signed, insurance to be assigned. Read the conditions before celebrating the approval.

The advance happens after the conditions precedent are satisfied, which routinely takes longer than owners expect — two to six weeks is ordinary for a small commercial file, and longer where a public program's registration is involved. A plan whose month-one capital purchase assumes funds available on day one has usually mis-timed its own cash flow by a month.

METHOD Assembling a financing package in eight documents One: the plan itself, in its delivered version, dated and numbered. Two: the one-page summary, which is what circulates internally. Three: the financial model, sendable, with an assumptions tab and visible formulas. Four: historical financial statements, two or three years, as your accountant produced them. Five: interim statements, if the year end is more than a few months back. Six: the supporting quotes for every capital item above your stated threshold, dated and from named suppliers. Seven: the corporate documents — register extract, incorporation, shareholder agreement if one exists, and any lease the plan depends on. Eight: personal financial information for each guarantor, in the institution's own format. The deliverable is one folder with a checklist showing what is present and what is outstanding. The verification criterion is that a person who has never seen your business could tell, from the checklist alone, whether the package is ready to send.

8.6 What an Investor Reads, and Why It Differs

Most readers of this book will not seek equity, and this section is short for that reason. → see Book A5 (Financing a Business) for who should, for what dilution actually costs in ownership, control and preference, and for the cap table arithmetic behind a round.

An equity investor is buying a share of a future outcome, which produces a different reading order and different priorities. The size of the opportunity comes first, because an investor's return depends on the business becoming much larger. The team comes second, because at early stages the plan will change and the team is what remains. The model and its unit economics come third: does each customer make money, and does that improve with scale? The ask and the dilution come fourth. The exit comes fifth, and it is a real question rather than a formality: an investor needs a path to converting ownership into money.

The consequence for a plan is that a business with a solid, profitable, geographically-bounded operation — which describes most readers of this book — is often an excellent lending candidate and a poor equity candidate, and there is nothing wrong with that. Presenting such a business to an equity investor wastes months. Recognizing which category you are in is worth an hour of thought before writing this section.

8.7 Terms, Covenants and Personal Guarantees

8.7.1 The seven terms to read

Every credit agreement has seven terms that determine what you have actually agreed to.

The rate, and whether it is fixed or floating. A floating rate is expressed as a base plus a spread; know both, and know what happens if the base moves.

The term, meaning when the loan comes due, which is not always the same as the amortization period. A five-year term on a fifteen-year amortization means a large balance is due for renewal in year five, at whatever rate exists then.

The amortization, which determines the payment.

The security, meaning what the lender can seize. Read the description of the collateral, since general security agreements typically cover all present and future assets.

The guarantees, which are the subject of §8.7.3.

The covenants, which are the subject of §8.7.2.

The fees and conditions precedent — application fees, standby fees on undrawn amounts, and the documents that must be delivered before funds are advanced. The last of these determines your actual timeline and is routinely underestimated.

DEFINITION Term loan, line of credit, guarantee, covenant A term loan is a fixed amount advanced once and repaid on a schedule, normally secured against the asset it financed. A line of credit is a revolving facility that can be drawn and repaid repeatedly up to a limit, with interest on the drawn balance only; it finances the operating cycle rather than assets. A guarantee is a promise by someone other than the borrower — usually the owner personally — to pay if the borrower does not; it makes personal assets available to the lender and it survives the company. A covenant is a promise about how the business will be run while the loan is outstanding: financial covenants require ratios to stay within limits, and positive or negative covenants require or prohibit specific actions such as taking on further debt, selling assets or paying dividends.

8.7.2 Covenants, and how they actually bite

A covenant breach is a default, and a default entitles the lender to demand repayment even when every payment has been made on time. This is the part that surprises people.

Two habits reduce the risk. Model the covenants before signing: take the ratios the covenant requires, compute them on your own forecast, and see how much room exists. A covenant that your base case satisfies by a hair will be breached by the low scenario. And know the reporting dates, because a covenant is tested on financial statements delivered on a schedule, and the delivery obligation is itself a covenant that businesses breach by simply being late. → see Book A5 (Financing a Business) for the procedure that tests a covenant against your own forecast before signature, and for the reason the same ratio can be a strict test or a loose one depending on how the agreement defines it.

8.7.3 Personal guarantees

For a small business, a personal guarantee is usual rather than exceptional, and the useful posture is neither to refuse it reflexively nor to sign it unread.

Three things are worth knowing. A guarantee is a separate obligation: the lender can pursue the guarantor without first exhausting the company's assets, unless the document says otherwise. It survives the loan's purpose: it typically continues until it is explicitly released, which can be long after the equipment it financed is gone. And it is sometimes negotiable in scope: limited to an amount, limited to the facility being signed, tied to a milestone, or shared among several owners severally rather than jointly. → see Book A5 (Financing a Business) for those four levers in detail, including the caution that a borrower who asks for all four at once usually obtains none.

THE RULE A personal guarantee survives the company; know what you signed Before signing any guarantee, obtain and read the document itself rather than the summary in the term sheet, and be able to answer four questions. What is the maximum amount? — an unlimited guarantee and a guarantee limited to $92,000 are very different instruments. When does it end? — on repayment, on a date, or only on written release. What does it cover? — this facility only, or all present and future obligations. Who else has signed? — and whether the guarantors are jointly and severally liable, which means the lender may pursue any one of them for the entire amount. If your spouse is being asked to sign, they should have their own advice; this is a genuine legal matter and not a formality. → see Book H1 (Contracts and Commercial Agreements).

8.8 Government Programs and Guarantees, With Their Dates

Public programs change, which is why this section states its sources and its dates precisely and why you must verify before quoting anything to a lender.

8.8.1 The federal loan guarantee

The Canada Small Business Financing Program, administered by Innovation, Science and Economic Development Canada and delivered through chartered lenders, shares the risk of a loan with the lender rather than lending directly. Its published parameters, read on September 9, 2026, are: a term loan of up to $1,000,000, of which at most $500,000 may finance assets other than real property, and of that, at most $150,000 may finance intangible assets and working capital; a line of credit of up to $150,000; eligibility restricted to businesses whose estimated gross annual revenues do not exceed $10 million in the fiscal year the loan is approved, excluding farming; a registration fee of 2% of the loan or authorized amount; and interest capped at the lender's prime rate plus 3% for a floating-rate term loan, including a 1.25% administration fee, at the lender's posted residential mortgage rate plus 3% for a fixed-rate term loan, and at prime plus 5% for a line of credit (Innovation, Science and Economic Development Canada, 2026a).

Two practical points follow for a plan. The program is applied for through your lender, not directly, which means the plan you write for the bank is the plan the program sees. And the sublimits matter: a request that is mostly working capital does not fit the term loan structure, which is why the running case separates its $92,000 of equipment from its $40,000 operating line. → see Book A5 (Financing a Business) for the program's statutory basis, the maximum outstanding a single borrower and its related persons may hold, the maximum length of each facility, the security the lender must take, and the uses the program will not finance.

8.8.2 Startup lending

Futurpreneur Canada lends to entrepreneurs aged 18 to 39 who are Canadian citizens or permanent residents residing in Canada. Its startup program provides up to $25,000 from Futurpreneur with up to $50,000 more from a co-lending partner, for a maximum of $75,000, repayable over five years with interest-only payments in the first year and no prepayment penalty. Three documents must be uploaded before an application is reviewed: a completed business plan, financial forecasts covering two years with a month-by-month breakdown, and government identification. Each borrower is matched with a mentor for up to two years (Futurpreneur Canada, 2026).

That month-by-month requirement is worth noting for anyone writing under age forty: the monthly cash flow of chapter 7 is not optional for this route, it is an admission requirement. → see Book A5 (Financing a Business) for the rest of the eligibility, which this chapter does not cover: how long the business may already have been operating, how much of it the applicant must own and control, and the credit and tax conditions the lender checks.

8.8.3 Provincial and regional

Investissement Québec offers loans, loan guarantees, equity investment and the delivery of government programs, in the three families its own pages describe; specific product parameters and minimum amounts are not published on those pages and change, so obtain them from an advisor rather than quoting a figure of unknown vintage (Investissement Québec, 2026). Regional development bodies, local investment funds and sector-specific programs exist across Canada with parameters that vary by region and by year. → see Book A5 (Financing a Business) for the three families read one by one, and for the local and regional funds that sit beside them with their own published ceilings.

QUÉBEC & CANADA NOTE What to verify before naming a program in your plan Naming a program in a plan is a claim about eligibility, and it should be verified rather than assumed. Check four things at the program's own page and record the date you read it. The eligibility conditions — revenue ceiling, sector exclusions, age or status conditions, region. The current parameters — maximum amounts, sublimits, rates, fees — which change and which no third-party summary keeps current. The application route — directly, or through a financial institution, which changes your timeline substantially. And the documents required, since several programs require the month-by-month forecast that chapter 7 produces. Where a parameter is not published, say so and give the mechanism without a number; a plan that states "the program's maximum is set by the lender within the program's published ceiling" is accurate, and one that invents a figure is not. → see Book G7 (Grants, Subsidies and Public Programs) and Book A5 (Financing a Business).

WHAT CHANGED Rate-linked caps move with the base rate Several public programs cap interest as a spread over a lender's prime rate rather than as a fixed number, which means the effective ceiling moves whenever the base does. The Bank of Canada raised its policy interest rate from 2.25% to 2.50% on October 28, 2026, with the next scheduled announcement on December 9, 2026 (Bank of Canada, 2026). A plan written before such an announcement and read after it will carry a stale rate assumption unless it was written as a base plus a spread with a date, which is the practice recommended in chapter 7. State the base, the spread and the date; do not state a single number. As of October 29, 2026.

8.9 The Exit Question, Answered Honestly

8.9.1 When the question applies

The exit question is an equity question. A lender does not need an exit; they need repayment, and repayment is described by the amortization schedule.

If you are not raising equity, this section is two sentences: the debt is repaid from operations on the schedule shown, and the owner has no current intention to sell. That is a complete answer and it is better than a speculative paragraph about future acquirers.

8.9.2 When it does apply

An investor needs a path from ownership to money, and there are only four: a sale of the business to a strategic buyer, a sale to a financial buyer, a sale of the investor's stake to another investor or to the company, or distributions from profits over time. The fourth is legitimate and is systematically undersold in plans that assume every investor wants an acquisition.

State which path is contemplated, what would have to be true for it to be available, and over what horizon. → see Book M1 (Preparing Your Business for Sale) and Book M2 (Business Valuation).

8.9.3 Succession is not an exit

A family transfer or an employee buyout is a succession rather than an exit, and it has different mechanics, different tax consequences and a much longer timeline. If that is the intention, say so; it changes what a lender thinks about the loan's term and what an investor thinks about the opportunity. → see Book M5 (Succession, Family Transfer and Employee Ownership).

CASE IN POINT Nordec Entretien: the request that got smaller This case is fictional; the company, the people and the figures are invented. The first version of Sophie Nadeau's plan asked for $180,000. The figure came from a conversation with another owner who had borrowed that amount, and the use of funds table had been built to reach it: the vehicle, the tooling, "working capital" at $60,000 with no explanation, and a $28,000 line described as contingency. The second version asked for $132,000, and every dollar of the reduction came from arithmetic rather than from courage. The working capital line was computed instead of guessed. The monthly cash flow put the trough at $31,400 in month five. The $60,000 had no basis and the honest figure was half of it. The collections assumption was tightened and made a milestone. Improving days sales outstanding from 47 to 38 releases $17,500 — more than half the trough — and the plan made it a milestone with an owner and an observable criterion rather than an aspiration. The contingency line disappeared into the operating line, as an $8,600 buffer with its reason stated: a partial cover against the collections slippage identified in the sensitivity analysis. The account manager's reaction was the interesting part. She said that a smaller request derived from a cash curve was easier to approve than a larger one derived from a conversation, and that the collections milestone was the line that made the difference — because it showed the borrower had found a way to need less money before asking for any.

8.10 The Financing Request in Three Situations

8.10.1 Solo

Your personal and business finances will be assessed together, and a personal guarantee is close to certain. Two things are worth doing before applying: obtain your own credit report and read it, and separate your business and personal accounts if you have not, because a lender assessing commingled accounts is assessing a harder file.

The realistic sources for a solo business are a term loan for equipment, a modest operating line, supplier terms, and — if you are under forty — the startup route in §8.8.2.

8.10.2 Small team

Where there are several shareholders, settle the guarantee question among yourselves before the lender raises it. Guarantees are frequently joint and several, meaning the lender may pursue any one guarantor for the whole amount, and shareholders who have not discussed this in advance discover it in a difficult moment.

8.10.3 Established business

You have covenants already, and the first thing to check is whether the new facility conflicts with an existing one — most credit agreements restrict additional indebtedness without consent. Model the existing covenants against the post-advance forecast before applying.

At this size the request should also state what alternative uses of the capital were considered and rejected, because a reader at this level expects capital allocation to be a decision rather than a reflex.

WHO DOES THIS Who negotiates, and what a commercial banker actually decides The owner negotiates, and cannot delegate it, because the questions are about judgment rather than about numbers and because the person signing the guarantee is the person who should have heard the answers. The accountant prepares, meaning the statements, the ratios and the covenant modeling. Ask for the covenant model specifically; it is a small piece of work and it is the one most likely to prevent an unpleasant year two. The lawyer or notary reads the security and the guarantee before signing, not after. This is a bounded engagement, not an open-ended one. What the account manager actually decides is worth understanding: in most institutions they assess and recommend, and a credit adjudicator approves. This has a practical consequence — your document has to work for somebody you will never meet, who will read it without you in the room. That person is the real audience for the financing section.

TOOLS OF THE TRADE What a financing package needs from your document tools A financing package is eight to twelve documents and the failure mode is assembly rather than authorship. You need one folder with a checklist, because the last missing document is why packages sit unsent. You need the model in a sendable form, since it will be requested. You need historical statements as your accountant produced them, not retyped. You need the quotes and the register extract as originals, dated. You need a version scheme, since the package will be revised at least once. And you need a record of what was sent, to whom, and when, because in a process involving several institutions this is the thing people lose track of first. Chapter 13 names the products; a folder and a checklist are sufficient.

Costly Mistakes

1. Choosing the amount before building the cash flow. The request comes out round, the use of funds is fitted to it, and a reader who notices concludes the plan was written backwards.
2. Asking for exactly the trough. A facility sized to the forecast has no room for the forecast to be wrong, and returning for more in month seven is a much harder conversation than asking correctly in month zero.
3. Financing working capital with a term loan, or equipment with an operating line. Both work in the short term and both create problems: one pays principal on money needed for six months, the other consumes the facility that the operating cycle needs.
4. Writing "general working capital" in the use of funds table. It is where padding is expected to hide, and it invites the reader to reduce the request.
5. Signing a guarantee without reading the guarantee. The four questions in §8.7.3 take ten minutes and the obligation can outlive both the loan and the company.
6. Failing to model the covenants. A covenant satisfied by a hair in the base case will be breached in the low case, and a breach is a default even when every payment has been made.
7. Quoting a public program's parameters from a summary of unknown vintage. They change, the lender knows the current ones, and being wrong about an eligibility condition is a poor way to open a conversation.

Key Takeaways

• The financing amount is an output of the plan, derived from the capital cost and the cash trough, never an input chosen in advance.
• A request has three components — capital, working capital and a stated buffer — and each is financed by a different instrument.
• The buffer must have a reason attached; the running case adds 27% over the trough and says exactly what that does and does not cover.
• A use of funds table has five columns, and the fifth — what will prove the money was well spent — is what converts spending into a commitment.
• Lenders assess five things in order: coverage, security, the operator, leverage, and the purpose with its repayment source; compute the three ratios yourself.
• Supplier and customer terms are financing, they cost almost nothing, and a plan that shows you used them first is read differently.
• Public program parameters are volatile: cite them with the date read, verify eligibility at the source, and give the mechanism rather than a number where none is published.
• A personal guarantee is a separate obligation that survives the company; four questions answered before signing prevent most of the unpleasant surprises.

Your Action Plan

1. Derive the requirement from your cash curve, in three components. Deliverable: capital, trough and buffer, each with its arithmetic. Timing: one hour, once chapter 7 is built. Verification criterion: you can state the calculation that produced the total in one sentence.
2. Build the use of funds table with all five columns. Deliverable: a table that sums exactly to the request, with a capability and a proof for each line. Timing: forty-five minutes. Verification criterion: no line says "general working capital" or "contingency."
3. Compute the three ratios a lender will compute. Deliverable: debt service coverage, debt to equity post-advance, and current ratio, with the arithmetic shown. Timing: thirty minutes. Verification criterion: each ratio is computed from your own statements and can be reproduced by the reader.
4. Exhaust the cheap sources before the expensive one. Deliverable: a short note on supplier terms and customer billing terms, with what each would release and whether you have asked. Timing: one hour, plus the conversations. Verification criterion: at least one has been asked for, with the answer recorded.
5. Model your prospective covenants against the low scenario. Deliverable: each covenant ratio computed under base and low cases, with the headroom stated. Timing: one hour with your accountant. Verification criterion: you know which covenant breaks first and in which month.
6. Verify every public program parameter at the source and record the date. Deliverable: a short list of programs with their parameters, eligibility, application route and the date each page was read. Timing: ninety minutes. Verification criterion: no parameter in your plan comes from a third-party summary.
7. Obtain and read any guarantee document before the signing meeting. Deliverable: written answers to the four questions of §8.7.3. Timing: thirty minutes, plus advice where a spouse is involved. Verification criterion: you can state the maximum amount and the release condition from memory.

In La Machine

Every route named in this section was verified in the product's route inventory on 2026-09-09 (La Machine, 2026a), and what each page holds is described in the product's user guide (La Machine, 2026b).

• What the product holds about money, and for whom. Two of these pages are about the product rather than about your customers: account/billing holds what you owe La Machine for the product itself, and account/transaction-detail holds one of those charges. Your own recurring offers live elsewhere: business/subscription-packages holds each one with its price, its cost and a monthly or annual billing period, and business/subscriptions is the revenue-and-expense view that reads subscription, package and product sales. A lender assessing revenue durability reads the recurring offers and their history, and reads them from your accounting records rather than from this view.
• The financing package as work. The assembly of eight to twelve documents is a project: a one-off at Control center → Projects (chains) (/projets), whose identifier takes the form P-001, with one task per document at Control center → Plan tasks (/taches), each carrying the document itself as its verification criterion. Two things about those tasks are worth knowing before you start. No page creates a plan task, so the tasks arrive through the programming interface, through a script or through the assistant, and every one of those paths refuses a task that carries no verification criterion. And Control center → Today (/aujourdhui) surfaces what is overdue — which in a financing process is usually the document waiting on somebody else — while showing the tasks due today as a count with no list.
• The deck that accompanies the request. business/assets/presentations holds decks, with business/assets/presentations/:id/edit for one of them.
• Decisions taken along the way. Whether to accept a covenant, whether to stage the request, whether to give a guarantee — each is a decision worth recording rather than remembering, and arbitrages is where it goes. That page is one of the administration screens that sit outside the domain groups rather than a Control center entry, and most of those are reserved to the administrator.
• What it does not do. La Machine does not model debt, does not compute a debt service coverage ratio, and holds no amortization schedule. It does not track covenants or their reporting dates, and it does not hold your credit agreements. It also holds no invoice and no customer-receivable ledger of your own, since neither an invoice nor a commercial quotation is among the product's 211 routes, so the receivable and collection figures this chapter uses are read from your accounting system. The modeling is done in the tools of chapter 13; what La Machine holds is the work of assembling and delivering the package.

Self-Check

1. State the direction in which a financing request must be derived, and give the one-sentence test that tells you whether yours was.
2. Your cash trough is $22,000 and your equipment costs $46,000. Describe the request you would make, including the instrument for each component and how you would size and justify a buffer.
3. Name the five things a commercial lender assesses, and compute a debt service coverage ratio for a business with EBITDA of $88,000 and annual debt service of $52,000. Say whether that would ordinarily be sufficient.
4. Give the four questions to answer before signing a personal guarantee. Which one most often surprises people?
5. A plan states that a public program will provide $250,000 for working capital. Name two things you would verify before that sentence goes into a document, and say where you would verify them.

Chapter References

Bank of Canada. (2026). Policy interest rate (target for the overnight rate raised to 2.50% on October 28, 2026; next scheduled announcement December 9, 2026). Retrieved October 29, 2026.

Business Development Bank of Canada. (2026b). Financing solutions for small and medium-sized businesses; advisory services. Retrieved September 9, 2026.

Futurpreneur Canada. (2026). Frequently asked questions (startup program: ages 18 to 39; up to $25,000 from Futurpreneur with up to $50,000 in co-lending, for a maximum of $75,000; five-year term with interest-only payments in year one and no prepayment penalty; business plan, two-year month-by-month forecasts and identification required; mentoring for up to two years). Retrieved September 9, 2026.

Innovation, Science and Economic Development Canada. (2026a). Canada Small Business Financing Program — Program guidelines (term loans up to $1,000,000 with a $500,000 sublimit for assets other than real property and a $150,000 sublimit within it for intangible assets and working capital; lines of credit up to $150,000; eligibility limited to businesses with estimated gross annual revenues of $10 million or less; registration fee of 2%; floating-rate term loans capped at the lender's prime rate plus 3% including a 1.25% administration fee; lines of credit capped at prime plus 5%). Retrieved September 9, 2026.

Investissement Québec. (2026). Financial products: loans, loan guarantees, equity investment and government programs. Retrieved September 9, 2026.

La Machine. (2026a). Documentation — route reference inventory (211 routes: 191 Core, 20 Pilotage paths). Internal product documentation, read September 9, 2026.

La Machine. (2026b). Guide utilisateur [User guide] (FAQ: projects and task chains; dashboard and reports). Internal product documentation, read September 9, 2026.

Chapter 9 — Risk Analysis and Mitigation

There is a widespread and mistaken belief that naming risks in a plan weakens it. The opposite is true, and the reason is structural: every reader already knows the business has risks, so the only thing a plan can demonstrate here is whether the people running it have identified the right ones. A section listing "competition" and "economic conditions" tells the reader that nobody looked. A section naming the one customer relationship that carries 46% of contract revenue, with a probability, a dollar figure and an owner, tells them something entirely different.

What You'll Be Able to Do

• Find the risks you would otherwise miss, using four systematic passes rather than a brainstorm.
• Build a risk register in which every entry carries a probability, a dollar impact, an owner and a trigger.
• Quantify a risk in dollars and compare its expected loss with the cost of doing something about it.
• Choose deliberately between mitigating, transferring, accepting and avoiding, and say why.
• Write a contingency plan with a trigger stated in advance, in observable terms.
• Name key-person risk honestly, including your own, and state what reduces it.

Why It Matters

The practical value of this section is not that it prevents the risks. It is that it converts a set of vague worries into a small number of decisions, each of which either gets made or gets deliberately deferred. A business owner who has done this work sleeps differently, not because the risks are gone but because they are enumerated and priced.

For a reader, the section performs a specific diagnostic. A plan that names the risks a professional in that sector would name, in roughly the right order of importance, signals that the people writing it understand their own business. A plan whose risk section could be pasted into any other plan signals the reverse, and it colors how the rest of the document is read.

There is also a purely financial reason. Two of the four responses to a risk — transferring it and mitigating it — cost money, and that money belongs in the forecast. A risk section written after the financial section, which is what chapter 2's construction order prescribes, is the last opportunity to catch a cost that has not been budgeted.

In This Chapter

• 9.1 Why the Risk Section Makes a Plan More Credible, Not Less
• 9.2 Finding Risks: Four Systematic Passes
• 9.3 The Risk Register: Probability, Impact and Owner
• 9.4 Quantifying a Risk in Dollars
• 9.5 Mitigation, Transfer, Acceptance and Avoidance
• 9.6 Scenarios: Three, Not Thirty
• 9.7 The Contingency Plan and Its Trigger
• 9.8 Insurance, Contracts and What They Actually Cover
• 9.9 Key-Person Risk and the Bus Factor
• 9.10 Writing the Section in Three Pages
• 9.11 Risk in Three Situations

9.1 Why the Risk Section Makes a Plan More Credible, Not Less

9.1.1 What the reader is testing

A reader uses this section to answer one question: do these people know what could go wrong in their own business? They are not looking for reassurance and they are not scoring the risks. They are checking whether the list matches what they would expect from somebody who has operated in that sector.

This has a counterintuitive consequence. Omitting a well-known sector risk is worse than naming it badly, because omission suggests you have not encountered it, and an operator who has not encountered the obvious risk of their own industry is either inexperienced or not being candid.

DEFINITION Risk, issue, mitigation, contingency, trigger A risk is something that has not happened and might, with a probability and an impact. An issue is something that has already happened, which is a different management problem and does not belong in this section. Mitigation is action taken now to lower the probability or the impact of a risk. A contingency is a plan for what you will do if it happens anyway, prepared in advance and not executed unless it does. A trigger is the observable event that says the contingency should now be executed, defined ahead of time precisely so that the decision is not made under pressure by somebody arguing about whether things are bad enough yet.

9.1.2 Risk in a plan is not risk management

This section is a small piece of a discipline that has an international standard behind it: ISO 31000:2018 sets out the principles, framework and process for managing risk (International Organization for Standardization, 2018), and its process — establishing the context, identifying, analyzing, evaluating and treating risks, with monitoring and communication throughout — is the structure this chapter compresses into three pages.

What the plan needs is the output of that process for the most consequential risks, not the process itself. Six to ten risks, treated properly, is the target.

9.2 Finding Risks: Four Systematic Passes

A brainstorm produces the risks you are already worried about, which are by definition the ones you are already managing. The four passes below produce the others.

METHOD The four passes that find the risks you would otherwise miss Pass one: walk the plan. Take each section of the plan in order and ask, for each of its key claims, what would have to be true for it to fail. The market section assumes a reachable count; what if the count is wrong? The operations section assumes a utilization rate; what if it is four points lower? This pass finds the risks that are specific to your plan rather than to your industry, and it is the pass most people skip. Pass two: walk the money. Follow the cash: where it comes from, who owes it, who you owe, what would delay each. This pass finds concentration, collection and covenant risks. Pass three: walk the dependencies. Take the dependency list from chapter 6 — suppliers, subcontractors, platforms, people — and ask what happens when each is unavailable for a month. This pass finds single points of failure. Pass four: walk the calendar. Look at the next thirty-six months for expiries and renewals: leases, contracts, licenses, certifications, insurance, loan terms, key employees' likely tenure. This pass finds the risks with dates attached, which are the easiest to manage and the most embarrassing to be surprised by. The deliverable is a raw list, usually of fifteen to thirty items, which §9.3 then reduces. The verification criterion is that each of the four passes produced at least two risks that were not on your list before you started.

9.2.1 Sorting the raw list

Reduce the raw list to six to ten entries by two rules. Merge risks with the same cause: three symptoms of one supplier failing are one risk. Remove risks with negligible impact or negligible probability, but record the removal, because "we considered X and judged it immaterial because Y" is itself a credible sentence.

What survives goes into the register.

9.2.2 A starting catalog, by category

The four passes are the method; the catalog below is a prompt for each pass, so that you are recognizing rather than inventing. Nine categories cover most small businesses, and the point is not to use all of them — it is to read each one and be able to say either "yes, that applies" or "no, and here is why."

Demand risks. Fewer customers than forecast; a segment that turns out not to buy; a competitor's price move; a substitute becoming acceptable; a trend that was a fashion.

Delivery risks. A capacity shortfall; a quality failure that costs a customer; a subcontractor who does not perform; a lead time that lengthens; a step in the process that depends on one person.

Financial risks. Collections slipping; a covenant breach; a rate rising on a floating facility; a large unexpected cost; a tax or filing deadline missed with the penalty that follows.

Concentration risks. One customer, one decision-maker, one supplier, one channel, one platform — each measured at the level where the decision is actually made, which §9.9 and the running case show is not always the obvious level.

People risks. A key person unavailable; a resignation at a bad moment; a hire who does not work out; a skill nobody else holds; a founder whose own capacity is the constraint.

Legal and compliance risks. A contract term that allocates more risk to you than you realized; a license or certification lapsing; a regulatory change with a compliance cost; an employment matter handled without regard to the notice the law requires.

Technology and data risks. A platform outage; a data loss; a security incident affecting customer records; a system nobody can administer but one person.

Physical risks. Premises unavailable; equipment out of service; a vehicle off the road; weather affecting a season's work.

Reputational risks. A public failure; a review that circulates; the loss of a reference customer whose name opened doors.

Two habits make the catalog useful rather than a checklist. Write one sentence per category, even where the answer is that nothing material applies, because the sentence is the evidence that you looked. And date the pass, because a risk catalog reviewed once a year and dated is a governance artifact, while an undated one is a document.

9.3 The Risk Register: Probability, Impact and Owner

9.3.1 The six columns

The risk, stated as a specific event rather than a topic. Not "competition" but "a competitor offering the same coverage at a lower price to our three largest portfolio clients."

The probability, over a stated period. Use percentages if you have any basis for them, and bands — low, medium, high, with the bands defined — if you do not. What matters is that the period is stated: a 10% risk over a year and over five years are different things.

The impact in dollars, computed rather than described. §9.4 shows how.

The response, chosen from the four in §9.5.

The owner, a named person who is responsible for watching it.

The trigger, the observable event that says the contingency should be executed.

KEY CONCEPT A risk with no owner and no trigger is a worry, not a risk The two columns most often left blank are the two that make the register operational. Without an owner, nobody watches the risk, and the register becomes a document rather than an instrument. Without a trigger, the contingency is executed when somebody finally says "this is bad enough," which is always later than it should be, because the person who must say it is usually the person who chose the plan that is failing. Defining the trigger in advance moves the judgment to a calm moment, and it is the single most valuable thing a risk section does for the business rather than for the reader.

The paragraph that follows introduces Table 9.1, the risk register of the running case. Its columns are the six above, and the rows are ordered by expected loss — probability times impact — rather than by probability alone, which is the ordering a reader expects and the one that puts effort where it belongs. Read the trigger column: each entry is something that could be checked on a specific date by somebody who is not the owner of the risk.

Table 9.1 — The risk register: probability, dollar impact, response, owner, trigger

9.4 Quantifying a Risk in Dollars

9.4.1 The two numbers

Every risk in the register carries two numbers: the impact, which is what it costs if it happens, and the expected loss, which is the impact multiplied by the probability. The first tells you how bad it would be; the second tells you how much it is worth spending to prevent.

Computing the impact means following the consequence through the model rather than guessing. A lost customer costs the revenue, less the variable costs no longer incurred, plus any costs that do not go away — which for a business with salaried staff is most of them, and which is why a lost customer in a service business hurts more than a lost sale in a resale business.

DO THE MATH Expected loss and the cost of mitigation, compared Fictional figures from the running case. The risk. Three property management firms control 19 of the 41 buildings the plan reaches by year end — 46% of contract revenue. The concentration is invisible at the building level and obvious at the decision-maker level, which is exactly why the dependency pass of chapter 6 matters. The largest of the three holds eight buildings. The impact, computed through the model. Eight buildings produce 8 × $8,220 = $65,760 of contract revenue, plus about 8 × 24 hours × $96 = $18,432 of intervention labor, plus roughly $7,700 of parts: $91,892, which the register states as $91,900. At the contribution margin of 84.7% established in chapter 7 — the technicians remain on payroll whether or not those buildings exist — the contribution lost is 91,900 × 0.847 = $77,800. The probability. The business lost 4 of 34 buildings last year, but buildings are not the right unit here: the decision is made at the firm level. Of nine firm relationships over three years, one ended: about 11% a year, which the register rounds to 12%. The expected loss. 0.12 × 77,800 = $9,336 a year. The cost of mitigation. A standardized monthly report costs about fourteen hours a year of the administrator's time at a loaded $32 an hour, or $448, plus a quarterly review with each of the three firms at three hours each, twelve times a year, of the owner's time at $58: $2,088. Total: $2,536 a year. The comparison. 9,336 ÷ 2,536 = 3.7. The mitigation costs about a quarter of what the risk costs in expectation, which makes it obviously worth doing — and, more usefully, gives you the sentence to put in the plan: "we spend $2,536 a year to address a risk whose expected annual cost is $9,336." The honest caveat. Expected loss is an average over many repetitions, and this business will experience this risk once or not at all. The calculation is a way of allocating attention, not a prediction. Where an impact would be fatal rather than merely painful, expected loss is the wrong tool: you mitigate regardless of probability, which is why the key-person risk in the register is addressed at 8% while a larger expected loss is merely accepted.

9.4.2 When not to compute

Two categories resist quantification and should be handled differently rather than forced into a number.

Existential risks — the ones that end the business — are addressed regardless of expected loss, as the caveat above says. State the impact as "the business does not continue" and move to the response.

Reputational risks have real costs that arrive over years and cannot be computed in advance. State the mechanism instead: what would be damaged, with whom, and what recovering would require.

9.4.3 Estimating a probability when you have no data

Most small businesses have no statistical basis for a probability, and inventing a precise-looking percentage is worse than admitting it. Four methods produce a defensible estimate, in descending order of strength.

Your own frequency. Count how often the event has occurred in your own history and over how long. The running case does exactly this: one firm relationship ended out of nine over three years, which is about 11% a year. The sample is tiny and it is yours, which makes it more useful than a better sample from a different business.

A related frequency. Where the event itself has never happened, count something adjacent and adjust. A business that has never lost a technician can count how many stayed beyond two years across the trade, or use its own turnover in any role.

A structural bound. Some probabilities can be bounded by the number of opportunities for the event to occur. A lease that expires once in five years produces a renewal risk in exactly one of those years, not in each of them; a contract with three renewal dates a year offers three occasions for a price refusal.

A calibrated band. Where none of the three applies, use bands rather than percentages, and define what each band means in the plan: low is once in ten years or less, medium is once in three to ten years, high is more often than once in three years. A band with a stated definition is honest and comparable across rows; an undefined "medium" is neither.

Whichever method you use, write it in the register or in a note beneath it. A probability whose basis is stated can be argued with and improved; one that appears without a basis will simply be disbelieved, along with the arithmetic that depends on it.

9.5 Mitigation, Transfer, Acceptance and Avoidance

Four responses exist, and naming which one you have chosen is more informative than describing what you will do. Project-management practice groups the same choices under its uncertainty performance domain, where responding to a threat is presented as a deliberate selection among avoiding it, transferring it, reducing it and accepting it (Project Management Institute, 2021).

Mitigate: act now to reduce probability or impact. Costs money or time, and the cost belongs in the forecast.

Transfer: move the financial consequence to somebody else — an insurer, a supplier through a contract term, a customer through a deposit. Costs a premium, and never transfers the operational consequence.

Accept: decide deliberately to carry the risk. This is a legitimate and underused response, and it becomes credible when it is stated with its reasoning: "we accept this because the expected loss is $2,100 a year and the mitigation would cost $6,000."

Avoid: do not do the thing that carries the risk. The strongest response and the most expensive, because it usually means forgoing the opportunity along with the risk. The running case uses it once: the institutional segment of chapter 4 is declined rather than pursued, because winning one requires a bid capability and a bonding capacity the business does not have.

9.5.1 The response must be affordable and scheduled

A response with no cost and no date is a wish. Each mitigation in the register should have a cost that appears in the forecast, an owner, and a date by which it is in place — which means most of them become milestones in the table of chapter 6 and tasks in chapter 10.

Figure 9.1 draws the four responses onto a probability-and-impact grid. The horizontal axis is impact, the vertical axis is probability, and each quadrant carries the response that ordinarily belongs to it: accept in the low-low corner, mitigate where probability is high, transfer where impact is high, and avoid where both are. The seven risks of Table 9.1 are plotted as numbered points, named in the legend beneath with the response actually chosen for each — which is not always the response the quadrant suggests, and the first risk is the example: it sits where a transfer would be expected and is mitigated instead, because no insurer will sell you a policy against a customer changing their mind. The dashed line across the upper right marks the region where expected loss stops being the right tool, because an impact in that zone is not something a business absorbs.

Figure 9.1 — Original figure. The four responses on a probability and impact grid, with the register's seven risks placed and the region where expected loss stops applying.

9.6 Scenarios: Three, Not Thirty

Scenarios were built in chapter 7 and they belong here as well, in summary, because a reader arriving at the risk section wants to see the financial consequence of the risks just named.

Three scenarios, each moving several assumptions coherently, each attached to one decision and one observable signal. The paragraph below introduces Table 9.2, which restates them with the risk each one embodies. Read the last column: a scenario whose signal cannot be observed by a specific month is a scenario nobody will act on.

Table 9.2 — Three scenarios, the risks they embody, and the decision each triggers

The point of the low scenario is not that it is pessimistic; it is that it has a decision attached which can be taken calmly in month five rather than urgently in month eight.

9.7 The Contingency Plan and Its Trigger

9.7.1 What a contingency plan contains

Four things: the trigger, the actions in order, who does each, and what it costs. Half a page for each of the two or three risks that warrant one.

Most risks do not warrant a contingency plan. Reserve them for the risks whose impact is large enough that improvising would be expensive — in the running case, the loss of a portfolio firm and the extended unavailability of the owner.

THE RULE A contingency plan states its trigger in advance, in observable terms The trigger must be a fact rather than a judgment: "days sales outstanding above 52 in any month," not "if collections deteriorate." The reason is behavioral rather than administrative. When a risk begins to materialize, the person best placed to notice is usually the person whose plan is failing, and they will interpret ambiguous evidence charitably — not from dishonesty but because that is what people do. A trigger defined in a calm month removes the interpretation. The corollary is that somebody other than the plan's author should be able to check the trigger, which is why the register has an owner column and why the triggers in Table 9.1 are all things that appear on an operating screen or a monthly statement.

9.7.2 Writing one

For the running case's largest risk, the contingency reads roughly like this. Trigger: a portfolio firm gives notice, or two consecutive quarterly reviews are missed. Actions: within 48 hours, the owner requests a meeting with the firm's decision-maker; within one week, the affected buildings' service history and response times are compiled; within two weeks, either a retention proposal is made or the buildings are formally transitioned; simultaneously, the acquisition effort of chapter 5 is redirected to the two segments not affected. Who: the owner leads, the administrator compiles. Cost: about twenty hours of the owner's time, plus whatever the retention proposal concedes, which is capped in advance at 8% of the contract value.

That last clause — the cap set in advance — is what prevents a retention negotiation conducted in a bad week from conceding more than the relationship is worth.

9.8 Insurance, Contracts and What They Actually Cover

9.8.1 Insurance is transfer, not elimination

Insurance moves the financial consequence of a risk to an insurer. It does not move the operational consequence: an insured vehicle fire still stops the work, and the plan should say what happens to the customers during the claim.

List the policies the business carries with their limits, and — more usefully — state what is not covered that a reader might assume is. The exclusions and the deductibles are where the surprises live, and a plan that names two of them reads as written by somebody who has read the policy.

9.8.2 Contracts as risk transfer

Contract terms transfer risk as effectively as insurance and cost nothing but negotiation. Limitation of liability clauses, exclusions of consequential damages, payment terms, termination notice periods, and the definition of scope all allocate risk between the parties, and a small business that has never read its own standard terms is carrying risks it did not choose. → see Book H1 (Contracts and Commercial Agreements).

9.8.3 The coverages a small business typically carries

A plan does not need an insurance treatise; it needs a list a reader can check against the risks named. The coverages below are the ones most commonly held by a small commercial business, with the question to ask about each. Availability, terms and names vary by insurer and by jurisdiction, so verify each with your broker rather than assuming that a category name means the same thing everywhere.

• Commercial general liability, covering third-party bodily injury and property damage arising from operations. Ask what the limit is per occurrence and in the aggregate, and whether the work you actually do is excluded.
• Property, covering premises, equipment and contents. Ask whether the valuation is replacement cost or actual cash value, since the difference on a ten-year-old vehicle is substantial.
• Business interruption, covering lost income while operations are stopped by an insured event. Ask what the waiting period is and what evidence of loss will be required, because this is the coverage most often bought and least often understood.
• Commercial automobile, where vehicles are used. Ask specifically whether a replacement vehicle is covered during a repair, which in the running case is the transfer that addresses a $9,400 risk.
• Errors and omissions or professional liability, where advice or professional services are provided. Ask whether coverage is on a claims-made basis, which determines what happens when the policy ends.
• Cyber, covering incident response, notification obligations and certain losses. Ask what it requires of you as a condition of coverage, since several policies impose control requirements.
• Key-person life or disability, funding the transition rather than performing it.
• Employment-related coverages, whose availability and scope vary widely.

Two habits belong in the plan itself. Name two exclusions you know your policies carry, because doing so demonstrates that the policy has been read. And state the deductibles, since a $5,000 deductible converts several small risks back into risks you are carrying yourself, which is a legitimate choice only once it has been noticed. → see Book G8 (Business Insurance and Risk Financing).

QUÉBEC & CANADA NOTE Three regimes a Québec business must price into its risk section Workplace injury. In Québec, coverage for employment injuries and occupational diseases is administered by the Commission des normes, de l'équité, de la santé et de la sécurité du travail (CNESST), which sets the employer's contribution rate by activity and manages claims (Commission des normes, de l'équité, de la santé et de la sécurité du travail, 2026). The contribution is a cost, and the prevention obligations are an operational requirement rather than an option; both belong in the forecast and in the operations section. → see Book J10 (Health, Safety and Sustainable Operations). Insurance itself. Insurers and brokers operating in Québec are regulated by the Autorité des marchés financiers, which is also where a business can verify that a broker or firm holds a current registration — a two-minute check worth doing before placing a policy (Autorité des marchés financiers, 2026). Employment obligations. The Act respecting labour standards, CQLR c N-1.1 sets the minimum terms governing hours, leave, notice and termination; a plan that assumes an employee can be released without notice is assuming something the Act does not permit, and the notice cost belongs in a contingency (consolidation updated to April 7, 2026, read September 9, 2026). → see Book H7 (Employment Law Essentials (Québec and Canada)). For cyber and data risks specifically, the Canadian Centre for Cyber Security publishes baseline controls aimed at small and medium organizations, which is the right starting point for a plan that needs to say something concrete rather than something reassuring (Canadian Centre for Cyber Security, 2026). → see Book K3 (Cybersecurity, Backups and Business Continuity) and Book H2 (Privacy and Personal Information).

9.9 Key-Person Risk and the Bus Factor

9.9.1 The measure

DEFINITION Expected loss, risk appetite, bus factor Expected loss is probability multiplied by impact, used to allocate attention across risks rather than to predict any single outcome. Risk appetite is the amount and kind of risk a business is willing to carry in pursuit of its objectives, and stating it — even in one sentence — makes the accept decisions in the register coherent rather than arbitrary. The bus factor is the number of people who would have to become unavailable before the business could not continue; a bus factor of one means a single person holds something nobody else can do, and in most small businesses that person is the owner.

The running case has a bus factor of one, and the plan says so. The owner holds pricing, dispatch and every portfolio relationship. The mitigation is specific rather than aspirational: the dispatch procedure is documented and transferred to a named person by month four — which is a milestone in chapter 6 — the pricing rules are written down, and a second signatory is added to the bank account.

9.9.2 What actually reduces it

Four things reduce key-person risk, in ascending order of effort and effectiveness. Documentation of what the person does, which is the cheapest and the least sufficient. A second person trained, which is the real answer and costs capacity for a period. Externalizing the function — a bookkeeper, a dispatch service, a fractional role — which trades cost for resilience. And structural change, meaning the business is redesigned so the function is no longer critical, which is rare and worth naming when it is available.

Key-person insurance is a transfer of the financial consequence, not a reduction of the operational one; it funds the transition rather than performing it.

CASE IN POINT Nordec Entretien: the concentration nobody saw This case is fictional; the company, the people and the figures are invented. Sophie Nadeau's first plan stated, correctly, that no single customer exceeded 4% of revenue. With thirty-four buildings and an average contract of $8,220, that was arithmetically true and it was the wrong unit. The dependency pass of chapter 6 asked a different question: not who pays the invoice, but who decides. Three property management firms controlled nineteen of the forty-one buildings the plan reached by year end — 46% of contract revenue — and the largest held eight. A decision taken in one office by one portfolio manager could remove $91,900 of revenue, which is more than the entire increase the plan was asking to finance. Three things followed. The risk entered the register with a probability derived from firm-level history rather than building-level history, which is a different and much smaller denominator: one relationship ended out of nine over three years. The mitigation — the standardized monthly report and the quarterly review, costing $2,536 a year against an expected loss of $9,336 — became a milestone with an owner. And the segment strategy of chapter 4 changed: the plan deliberately targeted owner-operator buildings alongside portfolio buildings, accepting a higher acquisition cost per building in exchange for a decision-maker base that could not be lost in a single conversation. The account manager's note on the file, which Sophie saw later, was three words: "knows their concentration."

9.10 Writing the Section in Three Pages

Page one: the register. The table itself, six to ten rows, ordered by expected loss, with all six columns. This page is the section; the other two support it.

Page two: the two or three risks that warrant more. A short paragraph each for the largest risks: what it is, what it would cost, what you are doing, and what the contingency is with its trigger.

Page three: scenarios and the general posture. The three scenarios in summary with their decisions, a sentence on risk appetite, and the list of insurance policies with two named exclusions.

WATCH OUT Writing "competition" as a risk and stopping there Four entries appear in weak risk sections with such regularity that a reader treats them as a signal: competition, economic conditions, key person risk, and regulatory change — each stated as a topic, without a probability, a dollar figure, an owner or a trigger. The problem is not that these are wrong; three of the four are usually real. The problem is that stated this way they carry no information, and a reader who sees all four in a row concludes that the section was completed rather than done. The fix takes an hour: for each of the four, ask what specific event you actually mean, what it would cost, how likely it is over the next year, who would notice first, and what they would see. "Competition" becomes "a competitor offers our three largest portfolio clients the same coverage at 12% less, at their renewal dates in March, June and September."

9.11 Risk in Three Situations

9.11.1 Solo

Your bus factor is one and there is no arrangement that makes it two. What is available is a specific set of reductions: a documented procedure that lets somebody else answer and triage, a reciprocal arrangement with a colleague in the same trade, disability insurance to fund the gap, and a written note telling whoever steps in where things are.

Concentration risk is also higher for a solo business, because one person can only serve so many customers and the largest is therefore a larger share. Compute it, state it, and treat it as a constraint on how quickly you can afford to lose anyone.

9.11.2 Small team

The characteristic risk at this size is the second key person — not the owner, but the one employee who holds a skill or a relationship nobody else has. Identify them, and be aware that the mitigation is a training investment that costs capacity now against a risk that may never materialize, which is exactly the kind of decision the expected loss calculation exists to inform.

9.11.3 Established business

The register should be maintained rather than written once, reviewed quarterly with the plan, and owned by named people at the level where the risk actually lives. At this size the failure is the opposite of the small-business failure: a register with forty entries that nobody reads. Keep the plan's version to the ten that would change a decision.

WHO DOES THIS Who owns risk when everybody is busy The owner owns the register itself — that it exists, that it is reviewed, that each entry has an owner. This cannot be delegated, because the act of maintaining it is the act of managing risk. Each entry has a named individual, who is the person who would notice first rather than the person most senior. The technician who is in the buildings notices a deteriorating relationship before the owner does. The accountant owns the financial risks: covenant headroom, collection deterioration, tax filing dates. Ask specifically for the covenant model and the filing calendar; both are small pieces of work. The insurance broker owns the coverage review, annually, and should be asked one question in writing: what has changed in our business this year that the policy does not reflect? Nobody owns "risk" as a function in a business of this size, and creating such a role produces a register rather than a reduction.

TOOLS OF THE TRADE What a risk register must record Whatever holds it, the register must record six things and support two behaviors. The six are the risk, the probability with its period, the impact in dollars, the response, the owner and the trigger. The two behaviors matter more: it must be reviewed on a schedule rather than when somebody remembers, which means it needs a recurring date attached to it; and it must record what changed, because the useful information in a register is the movement — a probability that has risen, a mitigation that is now in place, a trigger that fired. A spreadsheet with a date column does all of this. What no tool can supply is the discipline of opening it quarterly, which is why chapter 10 puts the review in the cadence rather than in the register.

WHAT CHANGED Two risk categories now belong in every small-business plan Two entries that were once specialist concerns are now ordinary. Cyber and data risk applies to any business holding customer records, and the practical instruction for a plan is to name a baseline of controls and say which are in place rather than to assert that security is taken seriously; the Canadian Centre for Cyber Security's baseline controls for small and medium organizations are the appropriate reference point. A business that needs a formal structure rather than a baseline will find one in ISO/IEC 27001, which requires an information security management system in which the risks are identified, treated and reviewed on a schedule (International Organization for Standardization, 2022). Third-party platform risk applies to any business whose operations depend on a service it does not control — scheduling, payments, communications, the site itself — and the question a plan should answer is what happens during a multi-day outage, which is an operational question rather than a technical one. Both belong in the register with a probability and an impact like anything else. As of September 2026.

Costly Mistakes

1. Naming topics instead of events. "Competition" carries no information; "a competitor offers our three largest clients 12% less at their March, June and September renewals" is a risk somebody can watch for.
2. Leaving the owner and trigger columns blank. They are what make the register operational, and without them the document is a list of worries that nobody checks.
3. Measuring concentration in the wrong unit. No customer above 4% of revenue can coexist with three decision-makers controlling 46% of it, and only the second number matters.
4. Using expected loss on an existential risk. A risk that ends the business is mitigated regardless of probability; multiplying a fatal impact by a small percentage produces a number that invites the wrong decision.
5. Deciding the trigger when the risk is materializing. The person best placed to notice is the person whose plan is failing, and they will read ambiguous evidence charitably. Set the trigger in a calm month.
6. Treating insurance as elimination. It transfers the financial consequence and leaves the operational one entirely with you; the plan should say what happens to customers during the claim.
7. Writing a mitigation with no cost and no date. It never appears in the forecast, it never becomes a task, and by the second quarter nobody remembers it was decided.

Key Takeaways

• A risk section makes a plan more credible, because the only thing it can demonstrate is whether the people running the business have identified the right risks.
• Four systematic passes — walking the plan, the money, the dependencies and the calendar — find the risks a brainstorm misses.
• A register entry has six columns, and the two most often left blank, owner and trigger, are the two that make it operational.
• Impact is computed through the model; expected loss is impact times probability, and it allocates attention rather than predicting outcomes.
• Four responses exist — mitigate, transfer, accept, avoid — and naming which one you chose is more informative than describing what you will do.
• A contingency plan has a trigger stated in observable terms, defined in a calm month, checkable by somebody other than the person whose plan is failing.
• Insurance transfers the financial consequence and never the operational one; contract terms transfer risk for the cost of a negotiation.
• Concentration must be measured at the level where the decision is made, which is frequently not the level at which the invoice is paid.

Your Action Plan

1. Run the four passes and produce a raw list. Deliverable: fifteen to thirty candidate risks, grouped by which pass found them. Timing: two hours. Verification criterion: each pass produced at least two items that were not on your list beforehand.
2. Reduce the list to six to ten and build the register. Deliverable: a table with all six columns, ordered by expected loss. Timing: two hours. Verification criterion: no row has an empty owner or trigger cell.
3. Compute the impact and expected loss for your three largest risks. Deliverable: the arithmetic for each, traced through your own model. Timing: ninety minutes. Verification criterion: each impact figure can be reconciled to a line in your forecast.
4. Compare each mitigation's cost against its expected loss. Deliverable: a one-line comparison per mitigated risk, with the ratio. Timing: forty-five minutes. Verification criterion: every mitigation you keep has a cost that now appears in the forecast.
5. Measure your concentration at the decision-maker level. Deliverable: the share of revenue controlled by your three largest decision-makers, not your three largest invoices. Timing: thirty minutes. Verification criterion: the two numbers differ, or you can explain why they do not.
6. Write two contingency plans with triggers. Deliverable: half a page each — trigger, actions in order, who, and cost with any concession capped in advance. Timing: one hour. Verification criterion: somebody other than you could check whether the trigger has fired.
7. Ask your broker the one written question. Deliverable: a written answer to "what has changed in our business this year that the policy does not reflect?" Timing: fifteen minutes to ask. Verification criterion: the answer names at least one thing, or explicitly says nothing has changed.

In La Machine

Every route named in this section was verified in the product's route inventory on 2026-09-09 (La Machine, 2026a), and what each page holds is described in the product's user guide (La Machine, 2026b).

• The register of automations, and what it teaches about registers. Control center → Workflows (/registre) holds the workflow registry, where each automation carries one of five states — To build, Under construction, In production, Paused and Removed — and being listed there does not mean something is running. The page's subtitle makes the distinction visible by giving two numbers: how many entries exist, and separately how many are in production. A risk register works the same way: presence is not treatment.
• The trace. Control center → Audit log (/journal) lists the events, newest first, each with its date, its actor, the entity concerned, the action and a payload that can be shown or hidden, filtered by actor, by entity, by action and by date range. Its total is computed with the same conditions as the list beneath it, so a filtered count is a true count. Two limits belong in a continuity plan: notes added to a task and documents attached to one write no event, and nothing exports. Where a risk materializes, the log is what lets you reconstruct the sequence rather than the recollection.
• What ran and what failed. Control center → Runs (/executions) holds the runs in six states: Queued, Running, Awaiting approval, Done, Failed and Skipped. A pattern of failures is a trigger in the sense this chapter means: observable, checkable by somebody else, and available on a date. Read the count with care, though — the page lists the two hundred most recent runs by default, so the number in its heading is the number it received rather than the number that exist.
• Decisions that follow a risk. arbitrages holds an unsettled decision. A risk whose response is "accept for now, revisit at the quarterly review" belongs there rather than in memory. The page is one of the administration screens that sit outside the domain groups rather than a Control center entry, and most of those are reserved to the administrator.
• Backups and versions. backups holds backups and versions the version registry; migration-registry holds migrations. A backup runs once a day without anybody asking, another is taken before any operation that could substantially change the data, and a failed backup never overwrites a valid earlier one. Note the honest limit: restoring a backup is not a user action, because a restore brings back the system's data as a whole rather than only yours, so a continuity plan that assumes self-service restoration should say who performs it and how it is requested.
• What it does not do. La Machine holds no risk register and computes no expected loss. It does not track insurance, covenants or contract expiry dates, and it will not alert you that a lease is ending. The register is a document; what the product supplies is the operating evidence from which triggers are read.

Self-Check

1. Name the four passes of §9.2 and say which one finds risks specific to your plan rather than to your industry.
2. A risk has a 15% annual probability and would cost $60,000. Compute the expected loss, and say what you would do if the same impact would instead end the business.
3. Give the six columns of a register entry. Which two are most often blank, and what does each one make possible?
4. Turn this into a compliant register entry: "supplier problems." Supply an event, a probability with a period, an impact, a response, an owner and a trigger.
5. Your largest customer is 4% of revenue. Explain, in two sentences, why that figure may still conceal a serious concentration, and say what you would measure instead.

Chapter References

Act respecting labour standards, CQLR c N-1.1 (Québec). LégisQuébec, consolidated version updated to April 7, 2026. Read September 9, 2026.

Autorité des marchés financiers. (2026). Register of firms and representatives; insurance sector oversight. Retrieved September 9, 2026.

Canadian Centre for Cyber Security. (2026). Baseline cyber security controls for small and medium organizations. Retrieved September 9, 2026.

Commission des normes, de l'équité, de la santé et de la sécurité du travail. (2026). Employer obligations, contribution rates and prevention. Retrieved September 9, 2026.

International Organization for Standardization. (2018). ISO 31000:2018 — Risk management — Guidelines. ISO.

International Organization for Standardization. (2022). ISO/IEC 27001:2022 — Information security, cybersecurity and privacy protection — Information security management systems — Requirements. ISO.

La Machine. (2026a). Documentation — route reference inventory (211 routes: 191 Core, 20 Pilotage paths). Internal product documentation, read September 9, 2026.

La Machine. (2026b). Guide utilisateur [User guide] (FAQ: the workflow registry; executions and recurrences; backups and data; versions and migrations). Internal product documentation, read September 9, 2026.

Project Management Institute. (2021). A guide to the project management body of knowledge (PMBOK guide) (7th ed.), uncertainty performance domain. PMI.

Chapter 10 — From Plan to Execution

Almost every book about business plans ends at the printer. This one does not, because the failure that costs the most is not a badly written plan — it is a well-written plan that nobody opens again. The conversion from document to work is mechanical, it takes about four hours, and it is what separates a plan that changed something from a plan that recorded an intention. This chapter performs that conversion.

What You'll Be Able to Do

• Reduce a plan to three objectives, and say why twelve is a symptom rather than an ambition.
• Cascade an objective into projects and tasks, each with an owner and a date.
• Write a verification criterion that somebody other than the doer can check.
• Hold a ninety-minute quarterly review that ends in decisions rather than in a status update.
• Choose five numbers to watch, each attached to a decision it would trigger.
• Amend a plan without abandoning it, and keep a decision record that survives the year.

Why It Matters

There is an asymmetry worth naming. Writing a plan takes thirty to fifty hours; converting it into work takes four. Yet the four hours are skipped far more often than the fifty, because the fifty end with a visible artifact and the four end with a task list that looks like ordinary work. The result is a business that has paid the full cost of planning and collected none of the benefit.

The second reason is that execution is where a plan's assumptions are tested. A forecast is a set of beliefs about the world; the only way to find out which ones were wrong is to act on them and watch. A plan converted into dated, owned tasks produces that evidence continuously. A plan that stayed a document produces it once a year, in aggregate, too late to act on.

The third reason is more human. A plan with three objectives and a quarterly rhythm gives a small team something that is otherwise very hard to obtain: a shared answer to the question what is this quarter for? That answer, more than any efficiency, is what makes a small business feel like it is going somewhere.

In This Chapter

• 10.1 Why Most Plans Stop at the Printer
• 10.2 From Strategy to Objectives: Writing Three, Not Twelve
• 10.3 Objectives and Key Results, Used Honestly
• 10.4 From Objective to Project, From Project to Task
• 10.5 The Verification Criterion: The Sentence That Makes a Task Real
• 10.6 Owners, Dates and What "Done" Means
• 10.7 The Quarterly Review, in Ninety Minutes
• 10.8 Measuring: Five Numbers, Not Fifty
• 10.9 When the Plan Is Wrong: Amending Without Abandoning
• 10.10 The Decision Record
• 10.11 Execution in Three Situations

10.1 Why Most Plans Stop at the Printer

10.1.1 Three causes, all fixable

The plan has no next action. It ends with a conclusion, which is a literary convention rather than an operational one. A document whose last page is a summary of itself gives the reader nothing to do on Monday.

The objectives are not decomposed. "Grow revenue by 14%" is not something anybody can do. Somebody can contact forty property managers, and forty contacts eventually produce 14%, but the distance between the two is exactly the work this chapter describes.

Nothing recurs. The plan is reviewed when somebody remembers, which is when something has already gone wrong. A cadence set in advance — quarterly, on dates — is what makes the plan an instrument rather than a memory.

10.1.2 What conversion actually costs

Four hours, once, plus ninety minutes a quarter. That is the entire ongoing cost of a plan that stays alive, and it is worth stating because the perceived cost is much higher.

The four hours break down as follows: one hour to reduce the plan to three objectives, one hour to decompose them into projects, ninety minutes to write the tasks with their criteria, and thirty minutes to schedule the reviews and choose the five numbers.

10.2 From Strategy to Objectives: Writing Three, Not Twelve

10.2.1 Why three

A small business can pursue three significant objectives at once. It can list twelve, and it will make progress on the two that were urgent, which are usually not the two that were important.

The number is not arbitrary and it is not about ambition. It follows from the observation that every objective requires attention from the same small number of people, and attention is the binding constraint in a business of under about twenty people. Three objectives is what fits.

If your plan produces twelve candidate objectives, that is useful information rather than a problem: it means the plan contains twelve things worth doing, and the exercise is to choose which three come first. The other nine are not canceled; they are queued, with a note saying when they will be reconsidered.

10.2.2 What makes an objective usable

Four properties. It must be falsifiable: at the end of the period, it is either met or not, and two reasonable people would agree which. It must be owned by one person. It must have a date. And it must be connected to the plan, meaning you can point at the section of the plan it comes from.

THE RULE An objective that cannot be falsified is not an objective Apply this test before anything else: at the end of the period, could two people who both want the business to succeed disagree about whether this was achieved? If yes, the objective is a direction rather than an objective, and it needs a number or an observable event attached. "Improve customer satisfaction" fails. "Hold median response time under ninety minutes on contract calls, measured across at least four hundred calls" passes, and it passes because the measurement method is stated as well as the target — a target without a stated measurement method is falsifiable in principle and arguable in practice, which is the same thing three quarters later.

10.2.3 The running case's three

Fictional figures. The plan of chapters 4 through 8 produces three objectives for year one, and each traces to a specific section.

Objective one: reach 41 buildings under contract by month 12, generating $304,100 of contract revenue. It comes from the sales forecast of chapter 7 and it is the driver on which everything else depends.

Objective two: hold median response time on contract calls under 90 minutes, measured across at least 400 calls. It comes from the offer of chapter 5 — the response commitment is the reason customers choose the contract — and from the risk register of chapter 9, where a service failure is the trigger for losing a portfolio firm.

Objective three: bring days sales outstanding under 40 by month 9 and hold it. It comes from the working capital arithmetic of chapter 7, where nine days of improvement releases $17,500, and from the financing request of chapter 8, where that release is what reduced the ask.

Notice what is not on the list. Profit is not an objective, because it is an outcome of the three; putting it on the list adds nothing that could be worked on directly. Growth is not an objective, for the same reason. And nothing appears that the plan did not produce.

10.3 Objectives and Key Results, Used Honestly

10.3.1 The structure

The objectives-and-key-results framework, popularized by Doerr (2018) and long used in technology companies, pairs a qualitative objective with two to four quantitative key results that would demonstrate it was achieved. The structure is genuinely useful and it is frequently misapplied in small businesses, so it is worth being precise about both.

Where it helps: it forces the distinction between the thing you want and the evidence that you got it, which is the same distinction as between an objective and its verification criterion. And it limits the count, because writing four key results for each of twelve objectives is self-evidently absurd in a way that writing twelve objectives is not.

Where it goes wrong in a small business: when key results become a performance measurement system. In a company of four people, tying compensation to key results produces conservative targets and gaming, and it does so quickly. Keep the structure for clarity and keep it out of the compensation discussion.

10.3.2 Leading and lagging

The pairing of the two is older than the framework above. The balanced scorecard argued that a business measured only by its financial results is measured only by outcomes that have already occurred, and that every such outcome measure needs the driver that precedes it standing beside it (Kaplan & Norton, 1996).

DEFINITION Objective, key result, project, task, verification criterion An objective is what you are trying to achieve in a period, stated so it can be judged true or false. A key result is a measurable outcome that would demonstrate the objective was achieved — an outcome, not an activity. A project groups the work that serves one objective; in the product described in In La Machine it is either a chain, meaning a sequence that continues over time, or a one-off, which closes when its need is met. A task is a unit of work attached to a single project, carrying an owner, a date and a criterion. A verification criterion is the observable fact that settles whether the task is finished, and it is what §10.5 is entirely about.

DEFINITION Leading and lagging indicator A lagging indicator measures a result after it has occurred: revenue, buildings under contract, profit. It is accurate and it is too late to act on. A leading indicator measures an activity or a condition that precedes the result: contacts made, proposals issued, response times, utilization. It is less certain and it is actionable. A dashboard of only lagging indicators tells you how last month went; a dashboard of only leading indicators tells you how busy you are. Both are needed, and the practical rule is that each objective should have at least one of each.

10.4 From Objective to Project, From Project to Task

10.4.1 The cascade

An objective becomes one to three projects. A project becomes five to fifteen tasks. Below that is ordinary work, which does not need to be in the plan.

The paragraph that follows introduces Table 10.1, which shows the cascade for the running case's first objective. Its columns give the level, the item, its owner and its verification criterion, so you can see the criterion becoming more concrete as the level descends — which is the point of the exercise. Read the bottom rows: those are things somebody does on a Monday.

Table 10.1 — The cascade from objective to task, with the owner and criterion at each level

10.4.2 One-off and chain

The distinction matters operationally. A one-off project closes: the vehicle is in service, and the project is done. A chain continues: outreach does not finish, it recurs, and treating it as a one-off produces a project that is never done and slowly becomes invisible.

The same distinction applies to tasks. A task that recurs is not a task that is overdue; it is a task in a healthy repeating state, and confusing the two produces a task list that looks alarming and is not.

10.4.3 How many tasks, and the task that is too big

Five to fifteen tasks per project is the working range, and the number itself is less important than two properties of the individual tasks.

A task should fit inside one working period. For a small business that means a few hours to a few days. A task estimated at three weeks is a project wearing the wrong label: it has no observable midpoint, so nobody can tell whether it is on track until it is late.

A task should have one owner and one criterion. Where a task needs two people, it is usually two tasks with a dependency between them, and splitting it makes the handoff visible — which is where the time actually goes.

The test for a task that is too big is simple: can the owner tell you, today, whether it is on schedule? If the honest answer is "I think so," the task needs splitting. Split it at the point where something observable exists: not "build the report," but "agree the four figures the report will show," then "produce the first one," then "send it to three clients and ask whether it is useful."

DO THE MATH The arithmetic of a quarter that actually fits Fictional figures from the running case; the method is what transfers. A quarter has thirteen weeks. Sophie works about 46 hours a week, of which the plan already commits 6 hours to outreach and roughly 14 hours to field work at 30% of a technician's load. Administration, dispatch, quoting and customer conversations consume about 20 hours. What remains for the plan's projects is 46 − 6 − 14 − 20 = 6 hours a week, or 6 × 13 = 78 hours a quarter. Now the demand side. The three objectives produced eleven tasks owned by Sophie in the first quarter. Estimating each at its honest size — the two hiring tasks at 14 hours each, the financing tasks at 10, the vehicle tasks at 6, the rest between 2 and 5 — gives a total of 94 hours. 94 against 78. The quarter is over-committed by about 20% before anything goes wrong, and something always goes wrong. Three responses are available and all three are legitimate: move two tasks to the next quarter, which is what the running case did; reassign a task to the administrator or a supplier, which the tool-set preparation was; or reduce a task's scope, which the role description was, from a full hiring framework to a one-page description. The point of the calculation is not the precision — the estimates are rough by nature. It is that a quarter's capacity is finite and knowable, and that comparing the two takes fifteen minutes. Most over-committed quarters are not the result of ambition; they are the result of nobody adding up the hours.

10.4.4 Dependencies

Where one task cannot start until another finishes, say so. The dependency is what keeps a sequence in its intended order without anybody policing it.

One behavior is worth knowing before you build a chain: a dependency waits for the preceding task to be completed, not merely closed. If the task it depends on is dropped rather than completed, the dependent tasks stay blocked — which is correct, because a pivotal task that was abandoned should stop what came after it, but it surprises people. When you drop a pivotal task, check what was waiting on it. The same rule has a second consequence that surprises more people: a task that recurs is never completed, because completing it sets it back to its recurring state, so a dependency hung on a routine never releases. Order the work that follows a routine by date instead.

10.5 The Verification Criterion: The Sentence That Makes a Task Real

This is the most important paragraph in the chapter.

10.5.1 What it is

A verification criterion is the observable fact that settles whether a task is finished, and the test of a good one is that it is observable by somebody other than the person who did the work.

"The report returns a number" is a criterion. "It's done" is not. "The campaign is launched" is not, because two people can disagree about what launched means. "The first three sends have gone out and the bounce report has been read" is.

KEY CONCEPT A task without an observable criterion cannot be completed, only dropped This is a rule the product described in In La Machine applies at the moment a task is created, and it is worth adopting whatever tool you use, because it changes behavior rather than record-keeping. When the criterion is mandatory at creation, the moment of writing the task becomes the moment of deciding what finishing means — which is exactly when that decision is cheap. Without it, the decision happens at the end, under time pressure, by the person with the strongest interest in declaring victory. Note what that product does not do, because the same gap exists in most tools: it demands the criterion when the task is born, and it checks neither its presence nor its satisfaction when somebody marks the task done. The corollary is the part people resist and then come to value: a task you no longer want to do gets dropped, it does not get forced to done. Dropping is honest and it leaves a trace; forcing a task to done destroys the meaning of every other completed task in the list.

10.5.2 How to write one

Three patterns produce good criteria.

The artifact pattern: name a thing that will exist. "A one-page role description exists and has been used in a posting."

The measurement pattern: name a number and where it is read. "The contract register counts 41 on the last day of month 12."

The event pattern: name something that will have happened, with a date attached. "Both technicians have started and are billing within ten days."

What all three have in common is that a stranger could check them. That is the whole standard.

10.5.3 The criteria that fail

Four failures recur. Effort criteria — "spent six hours on outreach" — measure input rather than result, and a task can be fully satisfied while producing nothing. Judgment criteria — "the section is good enough" — cannot be settled. Compound criteria — three conditions joined by "and" — are really three tasks. And criteria that depend on someone else — "the client approves" — make the task uncompletable through no fault of the owner, and should be rewritten as "the proposal has been sent and a response requested by a stated date."

10.6 Owners, Dates and What "Done" Means

10.6.1 One owner

Every task has exactly one owner. Two owners is no owner; a department is no owner. The owner is not necessarily the person who does the work — they are the person who is accountable for it happening, which in a small business is frequently the same person and in a growing one increasingly is not.

10.6.2 Dates that mean something

A date is a commitment or it is decoration. Two habits keep dates meaningful. Move a date deliberately and record that you did, rather than letting it pass silently — an overdue task that everyone has stopped looking at is worse than a rescheduled one. And do not date everything for the end of the quarter, which is what happens when dates are assigned without thinking and produces a quarter that is empty for ten weeks and impossible for two.

10.6.3 The states a task passes through

A small vocabulary of states is worth adopting deliberately, and the one used by the product described in In La Machine is a good default because each state means something distinct.

A task is To do when it is defined but not started; In progress when work has begun; Completed when its verification criterion is met; Active (recurring) when it runs again, which is a healthy final state rather than an overdue one; Later when it is set aside for a written reason and remains available; and Dropped when it will not be done, which ends it.

The two that carry the most information are Later and Dropped, and they are the two most often collapsed into each other. "Later" keeps a reason on record and preserves the option. "Dropped" is a decision. A list where everything postponed is dropped loses the queue; a list where nothing is ever dropped becomes a graveyard nobody reads.

10.7 The Quarterly Review, in Ninety Minutes

10.7.1 What it is not

It is not a status meeting. A status meeting reports what happened; a review decides what changes. The difference shows up in the output: a status meeting produces notes, and a review produces a small number of decisions with owners and dates. Quality-management standards ask for the same thing under the name of management review, which is held at planned intervals and whose output is a set of decisions and actions rather than a record of what occurred (International Organization for Standardization, 2015).

METHOD The ninety-minute quarterly review, in six moves Move one, ten minutes: the five numbers. Read them aloud, with last quarter's values beside them. No discussion yet; the point is that everybody has seen the same figures before anybody interprets them. Move two, twenty minutes: the three objectives. For each, state whether it is on track, at risk, or off track, using the verification criterion rather than a feeling. "At risk" must come with the specific reason. Move three, fifteen minutes: what the assumptions did. Open the assumption log from chapter 7 and check the three lowest-confidence rows against reality. This is the move that makes the whole planning exercise compound, and it is the one most often skipped. Move four, twenty minutes: the decisions. What changes because of moves two and three? Each decision gets an owner and a date, or it is not a decision. Move five, fifteen minutes: the risk register. Any trigger fired? Any probability that has visibly moved? Any new risk from the quarter's experience? Move six, ten minutes: the next quarter's objectives. Usually the same three, sometimes one replaced. Confirm them explicitly rather than letting them carry over silently. The deliverable is a one-page record: the five numbers, the three objectives with their status, the decisions with owners and dates, and any change to the register. The verification criterion is that somebody who missed the meeting can read the page and act correctly.

10.7.2 Who attends

Everybody whose work is in the plan, which in a business of four people is four people and in a business of thirty is perhaps six. The test is whether the person owns an objective, a project, or one of the five numbers.

An external attendee is worth considering once a year: an advisor, a board member, or simply somebody who will ask the question nobody inside will ask. → see Book I7 (Communication, Meetings and Decision-Making).

10.7.3 What to do when an objective is off track

The review's second move produces a status, and "off track" needs a response rather than a conversation. Four responses exist and choosing among them explicitly is what keeps a review to ninety minutes.

Add resources. More time, more people, more money. Available rarely, and it should be checked against the quarter's arithmetic in §10.4.3 rather than asserted — adding a task to an already over-committed quarter moves the problem rather than solving it.

Reduce the target. Legitimate, and it must be recorded with a date and a reason, or the objective quietly becomes whatever was achieved. A target reduced with a written reason is a decision; one reduced silently is a failure with better presentation.

Extend the date. Also legitimate, and it carries a consequence to be traced: if the objective feeds the forecast — and objective one does — then moving its date moves revenue, and the model should be updated in the same session.

Drop it. The response nobody chooses and the correct one more often than people admit. Where the quarter has revealed that the objective was less important than a thing that was not on the list, dropping it and promoting the other is better than carrying both badly.

The one response that is not available is to note it and move on. An objective that has been off track for two consecutive reviews without one of the four responses being chosen has stopped being an objective, and the list is now decoration.

10.7.4 The monthly version

Between quarters, a twenty-minute monthly pass reads the five numbers and checks whether any risk trigger has fired. It produces no decisions except urgent ones and it exists to prevent a quarter from ending in a surprise.

10.8 Measuring: Five Numbers, Not Fifty

10.8.1 The selection rule

A number belongs on the list if you can complete this sentence: if this number moves past X, we will do Y. A number with no decision attached is information, and information without a decision is a cost. Project-management practice reaches the same conclusion in its measurement performance domain, which holds that a measure earns its cost only when somebody acts on it and that a proliferation of measures hides the few that would change a decision (Project Management Institute, 2021).

Five is the working limit for a small business. It is small enough that everybody knows all five and large enough to cover the objectives.

The paragraph that follows introduces Table 10.2, the five numbers of the running case. Its columns give each number, whether it leads or lags, how often it is read, and — the operative column — the decision it would trigger and at what threshold. Read that last column: every row commits somebody to an action, which is what distinguishes this from a dashboard.

Table 10.2 — Five numbers, and the decision each one triggers

10.8.2 Where the numbers come from

Each number needs a source that is read rather than estimated, and the source should be named beside it. Two of the five in the running case are counted from the task and project records described in In La Machine, since that product calculates no indicator of its own and offers no place to store a number you chose; two come from the accounting records; and one — new conversations — comes from a log the owner keeps, which is the weakest source and is named as such.

A number whose source is "roughly" is not a number. Either find a source or replace it with something you can actually read.

WATCH OUT The dashboard with fifty indicators and no decision attached to any of them Dashboards proliferate because adding an indicator is free and removing one feels like admitting it did not matter. The result is a screen that takes fifteen minutes to read and produces nothing, which is then read less and less until it is not read at all. The countermeasure is the selection rule of §10.8.1, applied strictly: any indicator that cannot complete the sentence if this moves past X we will do Y is removed. Applying it to an existing dashboard typically removes three quarters of it, and the remaining quarter gets read.

10.9 When the Plan Is Wrong: Amending Without Abandoning

10.9.1 Being wrong is expected

A plan is a set of dated assumptions. Some will be wrong, and the plan's value is that it tells you which ones. The failure mode is not being wrong; it is the two responses that follow — pretending the plan is still current, or throwing it away and operating without one.

10.9.2 The amendment procedure

When an assumption turns out to be wrong, four steps take about an hour.

Name the assumption and what it actually is. From the log: "we assumed 47 days; the last three months averaged 58."

Trace the consequence through the model. Which lines move, and by how much? This is why the model must have no hard-coded numbers.

Decide what changes. A target, a date, a resource, or nothing — "nothing" is a legitimate outcome when the deviation is inside the buffer.

Record the amendment with its date in the assumption log and in the decision record, and issue a new version number if the plan has an external reader.

10.9.3 When to rewrite rather than amend

Amend when the assumptions moved. Rewrite when the structure changed: a new line of business, a fundamental change of customer, an acquisition, the loss of a capability the plan depended on. Rewriting for anything less discards the record of what you believed, which is the asset.

The annual refresh described in chapter 11 is the natural moment to decide which of the two the next twelve months need.

CASE IN POINT Nordec Entretien: twelve objectives became three This case is fictional; the company, the people and the figures are invented. The first execution attempt after Sophie Nadeau's plan was approved produced a list of twelve objectives for the year. Reading it now, it is obviously a list of everything the plan mentioned: hire two technicians, put the vehicle in service, reach 41 buildings, improve collections, standardize the monthly report, hold response times, reduce non-billable travel, formalize pricing, document dispatch, review insurance, build a supplier list, and start a referral program. By the end of the first quarter, four had progressed, seven had not been started, and one — the vehicle — was complete because it had a supplier chasing it. The second attempt kept three: buildings under contract, response time, and days sales outstanding. The other nine were not canceled; they were written on a second page headed "queued, reconsidered at the July review," with a one-line reason each. Two of them turned out to be tasks inside the three objectives rather than objectives at all — documenting dispatch belonged to the response-time objective, and standardizing the monthly report belonged to the portfolio-retention mitigation from chapter 9. The difference at the end of the year was not that more got done in total. It was that the three that mattered were all met, and that Sophie could answer, in one sentence and without opening anything, the question of what the quarter was for.

10.10 The Decision Record

10.10.1 Why decisions need a record

A business of any size makes decisions that will be questioned later — usually by the person who made them, who no longer remembers the reasoning. A decision record answers that question in one line and costs almost nothing to keep.

Four fields per entry: the decision, the date, the reason in one sentence, and what would cause us to revisit it. That last field is what turns a record into an instrument, because it converts a closed decision into one with a defined reopening condition.

10.10.2 What belongs in it

Not every decision. The ones worth recording are those that cost something to reverse, those that somebody disagreed with, and those made under uncertainty that will resolve later — which is most of the decisions a plan produces.

The running case's record contains eleven entries at the end of year one. Three of them were reopened, and in each case the reopening condition written at the time was what triggered the reopening rather than somebody's memory. Four of the eleven, written out, show what the four fields actually look like in practice.

Defer the institutional segment. 2026-03-14. We are not pursuing school boards, municipalities or health institutions this year, because winning one requires a bid capability we do not have and a bonding capacity we cannot currently obtain. Revisit if: we reach 48 buildings before month 20, or a client introduces us to a procurement process with no bonding requirement.

Hire two technicians in month 4 rather than month 1. 2026-03-22. Hiring earlier deepens the cash trough by roughly $28,000 without producing revenue sooner, because the sales cycle is eleven weeks. Revisit if: more than seven contracts are signed by month 3, which would mean the work exists before the capacity does.

Bill preventive contracts on the first of the month rather than in arrears. 2026-04-02. Shifts about $23,000 of receipts earlier in the cycle at no cost beyond one conversation per client. Revisit if: more than three clients refuse, in which case we return to arrears for those and keep the change for the rest.

Accept the technician-resignation risk rather than mitigating it. 2026-04-02. Expected loss is about $3,850 a year; a retention arrangement that would meaningfully change it costs more than that. Revisit if: a second resignation occurs within eighteen months, which would suggest the probability is higher than assumed.

Four things are worth noticing about those entries. Each one is a sentence, not a memorandum — the record is only maintained if entries take two minutes. Each carries the reason rather than the conclusion, which is what makes it useful later. Each reopening condition is observable, in the same sense as a verification criterion. And one of the four is a decision not to act, which is exactly the kind of decision that vanishes without a record and is then re-argued every quarter.

10.10.3 The undecided ones

A decision that cannot be made in the moment should be recorded as open rather than deferred silently, with what is blocking it and who will decide. This is the difference between a decision that is waiting and one that has been lost, and it is the single most useful thing to keep between reviews.

10.11 Execution in Three Situations

10.11.1 Solo

Three objectives is still three, and the temptation to hold them in your head is strong and should be resisted — not because you will forget them but because writing the verification criteria is what makes them precise.

The quarterly review with nobody else in the room is worth holding anyway, with one adaptation: read the numbers aloud. It sounds absurd and it works, because reading aloud recruits a different kind of attention than scanning does, and it is the closest a solo operator gets to being asked a question.

10.11.2 Small team

The conversion from plan to tasks should be done with the team rather than for them, in one session. It takes longer and it produces two things a solo conversion cannot: verification criteria that the people doing the work believe are achievable, and the discovery of the dependency nobody had noticed.

Assign one of the five numbers to each person, so that every number has somebody who notices it.

10.11.3 Established business

The cascade goes one level deeper — objective, project, task, and often a sub-task owned within a department — and the risk is that the connection to the plan is lost by the third level. The countermeasure is that every project names the objective it serves, and every objective names the section of the plan it comes from.

At this size the quarterly review is usually already happening. The question is whether it decides anything; the test is to read the last four sets of minutes and count the decisions with owners and dates.

WHO DOES THIS Who runs the review, and who must be in the room The owner or general manager runs it, and the role is to keep it to ninety minutes and to force each objective to a status using its criterion rather than a discussion. Everybody who owns an objective, a project or one of the five numbers is in the room. Nobody else needs to be, and adding people who own nothing turns the review into a briefing. Somebody other than the person presenting reads the numbers. It is a small trick and it changes the tone, because the person who owns a number tends to explain it before it has been heard. One person writes the one-page record during the meeting, not afterwards. A record written afterwards is written by memory, and the decisions lose their dates. An outsider once a year — an advisor, a board member, a peer — asks the question that nobody inside will ask. It is the cheapest governance a small business can buy.

Figure 10.1 draws the whole chain in one view: the plan on the left, then three objectives, then the projects each one contains, then the tasks with their verification criteria, and finally the Monday on which somebody does something. The loop returning from the quarterly review to the objectives is the part that most plans lack: it is what turns a one-directional document into a cycle, and the labels on it name what the review actually changes.

Figure 10.1 — Original figure. From plan to Monday morning: the cascade from three objectives to owned tasks, and the quarterly loop that amends them.

QUÉBEC & CANADA NOTE Aligning the cadence with the dates that already exist A quarterly rhythm works better when it sits beside the deadlines a business already has rather than beside arbitrary quarter-ends. Three families of dates are worth putting on the same calendar. Corporate tax and filing dates follow the fiscal year end, and the review held shortly after the year end is the one where actuals and forecast can be compared properly. Sales tax remittance dates — monthly, quarterly or annual depending on the registration — are cash events that belong in the cash flow and in the monthly pass. Employment-related dates include the reference year for annual leave, which runs from 1 May to 30 April unless an agreement fixes a different start, under sections 66 and following of the Act respecting labour standards, CQLR c N-1.1 (consolidation updated to April 7, 2026, read September 9, 2026), and the statutory holidays that shape available capacity. Setting the four review dates once, beside these, means the review happens when the information exists. → see Book G5 (Business Taxes in Québec and Canada).

TOOLS OF THE TRADE What an execution environment must provide Six capabilities, independent of product. It must hold a task with an owner, a date and a verification criterion, and ideally refuse to complete one whose criterion is not met. It must distinguish recurring work from overdue work, since conflating them makes a list unreadable. It must support dependencies, so a sequence holds without policing. It must surface what is due and overdue today, because a list nobody opens is a list that does not exist. It must keep a trace of what changed, by whom and when, which is what makes a review possible rather than reconstructive. And it must hold the numbers you chose to watch somewhere they are seen without being sought. Chapter 13 names the products; the product described below does four of the six, applies the first only at the moment a task is created, and does not hold a number of your own at all.

WHAT CHANGED The annual plan has given way to a quarterly rhythm Small-business practice has shifted over the past decade from an annual planning cycle with an annual review to an annual plan with a quarterly review, and the reason is practical rather than fashionable: the assumptions that drive a small business now change faster than twelve months. The consequence for the document is that the plan must be built to be amended — an assumption log, a model with no hard-coded numbers, a version scheme — rather than built to be final. A plan produced in a form that cannot be updated quarterly is a plan that will be rewritten annually from scratch, at four times the cost. As of September 2026.

Costly Mistakes

1. Ending the plan with a conclusion instead of a task list. The document is complete and nothing changes, which is the most expensive possible outcome of thirty to fifty hours of work.
2. Carrying twelve objectives. Four will progress, seven will not start, and the one that completes will be the one with an outside party chasing it.
3. Writing tasks without verification criteria. The decision about what finishing means then happens at the end, under pressure, by the person most invested in declaring victory.
4. Forcing a task to done when it should be dropped. It destroys the meaning of every other completed task, and the list stops being usable within a quarter.
5. Treating recurring work as overdue. A list that shows healthy repeating tasks as failures becomes alarming, then ignored.
6. Holding a status meeting and calling it a review. The test is the number of decisions with owners and dates; if it is zero, no review occurred.
7. Rewriting the plan instead of amending it. It discards the record of what you believed, which is the asset the whole exercise was building.

Key Takeaways

• Converting a plan into work takes about four hours once and ninety minutes a quarter, against thirty to fifty hours to write it — and it is the part most often skipped.
• Three objectives is what a small business can pursue, because attention rather than ambition is the constraint; the rest are queued with a reconsideration date.
• An objective must be falsifiable, owned, dated and traceable to a section of the plan.
• The cascade runs objective to project to task, and the verification criterion becomes more concrete at each level.
• A verification criterion must be observable by somebody other than the doer; where it is mandatory, the decision about what "finished" means happens when it is cheap.
• Five numbers is the limit, and a number earns its place only if it completes the sentence "if this moves past X, we will do Y."
• The quarterly review has six moves and ends in decisions with owners and dates; the move most often skipped — checking the assumptions — is the one that makes planning compound.
• Being wrong is expected; amend with a dated record, and rewrite only when the structure of the business has changed.

Your Action Plan

1. Reduce your plan to three objectives, each falsifiable. Deliverable: three sentences, each with a number or an observable event, an owner and a date, plus a second page of what was queued. Timing: one hour. Verification criterion: two people reading each objective would agree at year end whether it was met.
2. Cascade each objective into one to three projects. Deliverable: a list of projects, each marked one-off or chain, each naming the objective it serves. Timing: one hour. Verification criterion: no project serves two objectives.
3. Write the tasks, with a criterion for every one. Deliverable: five to fifteen tasks per project, each with one owner, a date and an observable criterion. Timing: ninety minutes. Verification criterion: a stranger could check any criterion without asking you.
4. Choose your five numbers and name their sources. Deliverable: a five-row table with lead or lag, frequency, threshold, decision and source. Timing: thirty minutes. Verification criterion: every row completes the sentence "if this moves past X, we will do Y."
5. Put the four quarterly reviews in the calendar today, on dates. Deliverable: four ninety-minute blocks with the six moves as the agenda. Timing: ten minutes. Verification criterion: the dates sit shortly after the periods whose numbers they will read.
6. Start the decision record with the decisions this plan already made. Deliverable: a four-column table holding the decisions taken while writing the plan. Timing: forty minutes. Verification criterion: every entry has a reopening condition.

In La Machine

Every route named in this section was verified in the product's route inventory on 2026-09-09 (La Machine, 2026a), and what each page holds is described in the product's user guide (La Machine, 2026b).

• Objectives become projects. Control center → Projects (chains) (/projets) holds the projects of the plan, which are not the same objects as the personal ones at productivity/projects. A project is either a chain, carrying a short code such as C5, or a one-off, carrying an identifier of the form P-001. Its status is active, on hold, done or archived, and it becomes done through a status change somebody makes rather than by calculation from its tasks; the route that saves the change validates nothing, and the assistant's status action is the only place the value is checked. Each card shows a gauge of the tasks that are completed or recurring out of the total, a count of the tasks marked critical and a count of those set to run automatically, and the last two counts ignore the status of the tasks they count.
• Projects become tasks. Control center → Plan tasks (/taches) holds the tasks of the plan, distinct again from productivity/tasks and crm/tasks. Each plan task points to at most one project. Its status is one of six: To do, In progress, Active (recurring), Completed, Later and Dropped.
• The verification criterion is required at creation, and only there. Every path that creates a plan task demands one and refuses the task without it, with a message saying that a task without a verification criterion cannot be completed. After that the enforcement stops: the drawer lets the field be saved empty without a warning, and the Mark as done button does not check that a criterion exists, let alone that it is met. The product's own guidance for a good criterion matches §10.5 — observable by somebody other than the person who wrote it, such as "the report returns a number" rather than "it's done" — but the discipline of §10.5.3 is yours to keep, not the product's to impose.
• Dependencies behave as §10.4.4 describes, with one addition. A task that depends on another is treated as blocked — not queued if it is automated, listed on Today under Blocked by a dependency, and left out of the next work offered — for as long as the task it depends on is anything other than Completed. A dropped predecessor therefore blocks its dependants for good. So does a recurring one, since a task that recurs returns to Active (recurring) instead of Completed and is never completed at all, which is why a dependency should never be hung on a routine. No page creates or removes a dependency: they are written by scripts and by the initial seeding.
• The wave. A project or task carries a wave, a code from V0 to V4 that places it in a stage of the plan, and it is the field that connects daily work back to the horizon the plan set. Nothing is scheduled by it: a task whose due date has arrived is due, whatever its wave says.
• The daily view and the construction view. Control center → Today (/aujourdhui) gathers the runs waiting for approval, the overdue tasks, the postponed tasks whose date has arrived, the open work, the runs in progress, the recent failures and the blocked tasks. Two cautions come with it: the tasks due today have a count and no list anywhere on the page, and the completed tile compares days in universal time, so in Québec it reads zero in the evening after a full day of work. Control center → Dashboard (/tableau-de-bord) measures construction rather than execution — how much of the plan has been built, by wave, owner, status and chain — with no filter, no period and no export. Control center → Audit log (/journal) lists the events with their date, actor, entity, action and payload, filtered by actor, entity, action and dates, and its total is computed with the same conditions as its list; notes and attached documents write no event. Control center → Runs (/executions) holds the runs in six states — Queued, Running, Awaiting approval, Done, Failed and Skipped — and lists the two hundred most recent by default, so its heading count is a page rather than a total.
• Where the decisions go. arbitrages holds open decisions. It is one of the administration screens that sit outside the domain groups rather than a Control center entry, and most of those screens are reserved to the administrator.
• What it does not do. La Machine holds no objectives-and-key-results structure and no objective object as such: an objective becomes a project, which is the closest equivalent. No page creates a plan task at all — tasks arrive through the programming interface, through a script or through the assistant. It does not compute a cascade, does not schedule your reviews, does not hold your assumption log, and calculates no indicator of your own. Across the productivity module every capability the assistant offers is read-only — it lists, counts and reads a record — and its few writes sit on the plan's side instead, where it can create, postpone, complete or drop a plan task and change a plan project's status, each after a click.

Self-Check

1. Give the four properties of a usable objective, and apply the falsifiability test to "improve our customer service."
2. Write a verification criterion for the task "prepare the marketing calendar," using one of the three patterns of §10.5.2. Say which pattern you used.
3. Explain the difference between a task marked later and one marked dropped, and say what a list loses if the two are collapsed.
4. Your dashboard has nineteen indicators. Describe the rule you would apply to reduce it, and estimate what proportion would survive.
5. Name the six moves of the quarterly review, and say which one is most often skipped and what is lost by skipping it.

Chapter References

Act respecting labour standards, CQLR c N-1.1, ss. 66–69 (Québec). LégisQuébec, consolidated version updated to April 7, 2026. Read September 9, 2026.

Doerr, J. (2018). Measure what matters: How Google, Bono, and the Gates Foundation rock the world with OKRs. Portfolio.

International Organization for Standardization. (2015). ISO 9001:2015 — Quality management systems — Requirements, clause 9 (performance evaluation: monitoring, measurement, analysis, evaluation and management review). ISO.

Kaplan, R. S., & Norton, D. P. (1996). The balanced scorecard: Translating strategy into action. Harvard Business School Press.

La Machine. (2026a). Documentation — route reference inventory (211 routes: 191 Core, 20 Pilotage paths). Internal product documentation, read September 9, 2026.

La Machine. (2026b). Guide utilisateur [User guide] (FAQ: projects and task chains; pilotage, tasks and their statuses; dashboard and reports; cadences and targets). Internal product documentation, read September 9, 2026.

Project Management Institute. (2021). A guide to the project management body of knowledge (PMBOK guide) (7th ed.). PMI.

Chapter 11 — Typical Tasks, Cadence and Workflows

A plan is produced by a finite, repeatable set of tasks, and it stays useful through a finite, repeatable rhythm. This chapter names both. It gives the ten working sessions that produce a document from nothing, the task table in the format La Machine uses, the five cadences that keep the plan alive after delivery, and — the part most books omit entirely — an honest account of which of these tasks can be automated and which must not be.

What You'll Be Able to Do

• Plan ten writing sessions, each with a deliverable and a verification criterion, and know what the whole thing costs in hours.
• Read and fill a task table in La Machine format, with the trigger, the frequency, the automation level and the definition of done.
• Choose the right cadence for each activity, from daily to annual, and say why each sits where it does.
• Run the annual refresh without rewriting the plan from scratch.
• Produce a lender or board report on the date the agreement requires rather than when convenient.
• Separate what can be automated from what must not be, using one test rather than enthusiasm.

Why It Matters

Two things go wrong with plans after delivery, and both are cadence failures rather than content failures.

The first is that nothing recurs, so the plan is opened when something has already gone wrong. A business that reads five numbers weekly and holds a review quarterly finds a deviation while it is still small; one that opens the plan annually finds it at a size that requires a decision rather than an adjustment.

The second is that the refresh becomes a rewrite. A plan rebuilt from scratch every year costs the full thirty to fifty hours again and — worse — discards the record of what was assumed the previous year, which is the one thing that would have made the new assumptions better than the old ones.

There is also a practical reason to name the tasks explicitly. Writing a plan feels like an amorphous obligation, which is why it gets postponed; ten named sessions with a deliverable each feels like a schedule, which is why it gets done.

In This Chapter

• 11.1 The Writing Sprint: Producing a Plan in Ten Working Sessions
• 11.2 The Task Table, in La Machine Format
• 11.3 The Daily and Weekly Cadence While Writing
• 11.4 The Monthly Cadence After Delivery
• 11.5 The Quarterly Review Cycle
• 11.6 The Annual Refresh
• 11.7 Board and Lender Reporting
• 11.8 What Can Be Automated, and What Must Not Be
• 11.9 The Workflows Worth Building
• 11.10 Cadence in Three Situations

11.1 The Writing Sprint: Producing a Plan in Ten Working Sessions

DEFINITION Cadence, sprint, refresh, reporting pack A cadence is a rhythm of recurring work with a fixed period and a defined output: daily, weekly, monthly, quarterly, annual. What makes it a cadence rather than a habit is that the period is set in advance and the output is named. A sprint is a bounded set of working sessions producing one deliverable — here, a plan — with a defined end rather than a rolling effort. A refresh updates an existing plan against reality without rebuilding it, keeping its structure, its sources and its version history. A reporting pack is the fixed set of documents delivered to a lender, a board or a program on a schedule; its value comes from being identical every period, which is what makes producing it a routine rather than a project.

Figure 11.1 shows the year of a plan on one line. The left-hand block is the writing sprint — ten sessions across four to six weeks — ending in delivery. The four bands to its right are the cadences that follow, each with its period, its duration and the single output it produces: ten minutes daily, twenty minutes weekly, one hour monthly, ninety minutes quarterly. The block at the far right is the annual refresh, and the arrow returning from it to the start is the reason a plan is written to be amended: the refresh consumes the previous version rather than replacing it.

Figure 11.1 — Original figure. The year of a plan: the writing sprint, the four cadences with what each one produces, and the annual refresh that returns to the beginning.

11.1.1 Why sessions rather than a deadline

A plan written against a deadline is written in the last week, badly. A plan written in ten sessions with one deliverable each is written over four to six weeks, and each session ends with something that exists.

The rule that makes it work is one deliverable per session. Not "work on the market section" but "produce the five-line funnel with every source named." A session that ends without its deliverable is rescheduled rather than absorbed into the next one, because absorbing is how a ten-session plan becomes a four-session plan with six sessions of debt.

METHOD The ten writing sessions, each with its deliverable Session 1 — the assumption log. Deliverable: a fifteen-row table with a value, a source, a confidence and a consequence for each. Three hours. Nothing else is written until every row has something in the source column. Session 2 — the population and the funnel. Deliverable: the five-line bottom-up funnel of chapter 4, each line with its source. Three hours. Session 3 — segments, trends and competition. Deliverable: pages two to five of the market section. Three hours. Session 4 — offer and pricing. Deliverable: each offer in the five elements, with exclusions and price structure. Two and a half hours. Session 5 — go-to-market. Deliverable: the conversion table with its evidence, the acquisition cost calculation, the ninety-day sequence. Three hours. Session 6 — operations, capacity and team. Deliverable: the capacity ceiling, the role list with the gap named, the milestones table. Three hours. Session 7 — the model, first half. Deliverable: revenue built from drivers, the cost structure, the income statement for three years. Four hours. Session 8 — the model, second half. Deliverable: monthly cash flow, opening balance sheet, break-even, sensitivity, three scenarios. Four hours. Session 9 — request, risks and company. Deliverable: the financing request derived from the trough, the risk register, the facts of record. Three hours. Session 10 — summary and reconciliation. Deliverable: the one-page executive summary and the reconciliation pass on paper. Three hours. The deliverable of the sprint is a complete draft. The verification criterion is that each session's own deliverable exists and is dated, and that session 10's reconciliation pass found fewer than three inconsistencies.

The paragraph that follows introduces Table 11.1, which gives the ten sessions with their deliverable, their hours and the verification criterion for each. Read the hours column and total it: the sum is what a plan actually costs in time, and it is a smaller number than most owners fear and a larger one than most calendars have room for by accident.

Table 11.1 — The ten writing sessions: deliverable, hours, verification criterion

DO THE MATH Ten sessions × the hours each: what a plan really costs The total of the hours column is 3.0 + 3.0 + 3.0 + 2.5 + 3.0 + 3.0 + 4.0 + 4.0 + 3.0 + 3.0 = 31.5 hours. At an owner's loaded cost of $58 an hour, that is 31.5 × 58 = $1,827 of the most expensive time in the business. Add a professional review of the model — a bounded engagement rather than an open one — and the direct cost rises by whatever that engagement is quoted at; chapter 12 explains how to scope and obtain that quote, and this book prints no fee ranges because they vary by market and by year. Now the comparison that matters. In the running case the plan's second version reduced the financing request from $180,000 to $132,000. On a five-year term at the assumed rate, the interest avoided on the $48,000 not borrowed is roughly $8,100 over the life of the loan — which is more than four times the cost of writing the plan properly, from one change alone. And the comparison that matters more. The same plan found that the business was at 99.96% of its capacity ceiling. Without that finding, the year's sales effort would have been spent on demand the business could not serve. Thirty-one and a half hours is what it cost to find out.

11.1.2 Spacing the sessions

Two or three sessions a week, over four weeks, is the working rhythm. Faster than that and the evidence-gathering between sessions does not happen; slower and each session begins with twenty minutes of remembering where you were.

Sessions 7 and 8 — the model — should sit close together, ideally two days apart, because the model is the one artifact that needs to be held whole in the mind.

11.2 The Task Table, in La Machine Format

11.2.1 What the format is for

A task table states, for each recurring or one-off task, what triggers it, how often it runs, how much of it is automated, and what proves it is done. It is the operational form of everything chapter 10 described, and it is what a plan hands to the business on the day it is delivered.

The paragraph below introduces Table 11.2, the task table for the running case. Its five columns are the task, its trigger, its frequency, its automation level and its verification criterion — the definition of done. Five columns rather than ten is deliberate: a table wider than about six columns is unreadable on a printed page, so the role and the tools for each task follow in prose beneath the table rather than being squeezed into it. The product's own list of plan tasks carries more than five — an identifier, the name, the business, the project, the owner, the tools, the automation, the due date, the frequency, the priority, the wave and the status — which is readable on a screen that scrolls sideways and not on paper.

Table 11.2 — The task table: trigger, frequency, automation level and definition of done

Who does each, and with what. The four one-off tasks belong to the owner, except the evidence folder, which an administrator can assemble from a list. Reading the day's queue and the five numbers belongs to whoever holds the daily pass — the owner in a business of four, an operations lead above that. Logging conversations belongs to whoever has them. The monthly client report belongs to the administrator, which is what makes it survive a busy month. Reconciliation and the lender report belong to whoever keeps the books, working from the accounting system. Checking risk triggers belongs to the risk owner named in each register row. The quarterly review and the annual refresh belong to the owner and cannot be delegated, because they are the two moments where judgment is the entire content. As for tools: the model and the funnel live in a spreadsheet, the document in a word processor with version history, the daily and weekly reads on the operating screens described in In La Machine, and the reports wherever the accounting data already is. Chapter 13 names products and their pricing models, dated.

11.2.2 One-off, recurring, and the difference that matters

A one-off task closes. A recurring task returns to a healthy repeating state and never becomes done — which is correct and which surprises people the first time they see a task list where several items are permanently in a state that is not "done."

Getting this distinction right in the task list is what prevents the most common failure of a task system in a small business: a list that looks like a backlog of failures, is therefore demoralizing, and is therefore not opened.

11.3 The Daily and Weekly Cadence While Writing

11.3.1 During the sprint

Daily, ten minutes: gather whatever evidence arrived — a quote, a statistic, a competitor's price — into the evidence folder with its date. Evidence gathered daily and filed immediately is evidence you can cite; evidence gathered at the end is evidence you have to find again.

Weekly, thirty minutes: review what the sessions produced against the ten-session plan, and reschedule rather than absorb any session that did not produce its deliverable.

11.3.2 After delivery

Daily, ten minutes: the operating pass — what is awaiting a decision, what is overdue, what is due today, what failed. This is not plan work; it is the routine the plan's tasks join.

Weekly, twenty minutes: read the five numbers with last week's values beside them, and protect the outreach block, which is the activity that is displaced first whenever a week gets busy.

11.3.3 Protecting the block

Every cadence in this chapter is displaced by the same thing: work that is urgent and arrives with somebody attached to it. Four habits protect a block, and they are worth more than any amount of resolve.

Attach it to something that happens anyway. The daily pass attaches to the coffee, the shop opening, the first email — an event that occurs whether or not you feel like it. A block attached to a time of day competes with the day; a block attached to an event inherits the event's reliability.

Put it in the calendar as an appointment, not a reminder. A reminder is dismissed; an appointment is rescheduled, and rescheduling leaves a trace that tells you at the end of the month how many times it happened.

Name the deliverable in the calendar entry. "Session 7 — revenue, costs, income statement" is a block you can tell whether you completed. "Business plan" is not.

Make the first action small enough to start. The reason a three-hour session gets postponed is rarely the three hours; it is that the first action is undefined. Write the first action in the entry: "open last year's invoices and count the contract renewals."

KEY CONCEPT A cadence is a period, a duration and one named output Any of the three missing and it is not a cadence. A period without a duration expands until it is abandoned. A duration without a named output produces a meeting rather than a result. And an output without a period happens once. The five cadences of this chapter each carry all three: ten minutes daily producing an empty approval queue or a written reason, twenty minutes weekly producing five numbers read against last week's, one hour monthly producing three written lines, ninety minutes quarterly producing a one-page record with decisions, and six to ten hours annually producing a new version with a change note. Write yours the same way, in that shape, and the question of whether they are working becomes answerable.

11.4 The Monthly Cadence After Delivery

One hour a month, in three parts.

Reconcile forecast against actuals, ten days after month end when the books are close enough. State a variance for revenue and for cash. Two numbers, not twenty.

Check every risk trigger in the register, and date the check. Most months this takes six minutes and finds nothing, which is the point: the value is entirely in having looked.

Read the plan's own assumptions that are marked low confidence. Three rows, checked against a month of reality.

The output is three lines in a running file: what the variance was, whether any trigger fired, and whether any assumption has visibly moved. Those three lines are what makes the quarterly review possible in ninety minutes rather than three hours.

11.5 The Quarterly Review Cycle

The review itself is described in chapter 10 — six moves, ninety minutes, ending in decisions with owners and dates. What belongs here is its scheduling.

Hold it two weeks after quarter end, which is late enough that the numbers exist and early enough that the quarter it will change has not run away. Put all four in the calendar at once, at the start of the year, on dates.

The review before the fiscal year end is different from the other three: it is the one that decides whether the annual refresh will be an amendment or a rewrite, and it should include an explicit twenty minutes on that question.

11.6 The Annual Refresh

11.6.1 Refresh, not rewrite

An annual refresh takes six to ten hours, against thirty-one and a half for a new plan. It keeps the structure, the sources and the history, and it updates what actually changed. The exercise is not peculiar to business plans: quality-management standards require the people who run an organization to review its system at planned intervals, against the objectives set and the evidence of how it performed, which is the same discipline applied to a different object (International Organization for Standardization, 2015).

Six steps. Update the actuals, replacing forecast columns with what happened. Revise the assumption log, marking each row as confirmed, revised or removed, with the evidence. Rebuild the forecast from the revised assumptions rather than from last year's forecast. Re-derive anything that depends on the forecast: the capacity check, the break-even, the request if there is a new one. Update the market section only where evidence changed, which is usually two or three figures rather than the whole section. Issue a new version number with a date and a one-page note of what changed and why.

That last step is what makes the refresh cumulative. A business with four annual versions of its plan, each with a change note, has something no single plan can provide: a record of how its own understanding of its market evolved.

11.6.3 What a refresh actually finds

Fictional figures from the running case. Nordec's first refresh, six weeks after the fiscal year end, took seven hours and produced a change note of eleven lines. Four of them are worth reproducing, because they show what the exercise is for.

Confirmed: the average contract price. Assumed at $685; the register's twelve-month average was $691. Left unchanged, with a note that the drift is upward and should be watched rather than acted on.

Revised: billable utilization. Assumed at 68%; measured at 65.4%. The two new technicians spent more of their first quarter traveling than the model allowed, because the second territory had not yet been routed. The forecast for year two was rebuilt at 66%, and the routing became a project.

Revised, in the other direction: days sales outstanding. Assumed to fall from 47 to 40 by month nine; it reached 38 by month eight. The milestone had an owner and a criterion, and it worked better than assumed — which is worth recording explicitly, because a business that only records the assumptions it got wrong learns a distorted lesson.

Removed: the price-refusal risk. The staged increase by renewal date produced one refusal out of nineteen renewals. The register row was closed with its outcome, rather than deleted, so that next year's version can see that the risk was named, addressed and did not materialize.

The refresh also produced one structural finding that no single year could have shown: the two years' assumption logs, read side by side, revealed that every estimate about the business's own operations had been optimistic and every estimate about the market had been conservative. That pattern is more useful than any individual correction, and it is only visible because the logs were kept.

11.6.2 When to rewrite

Rewrite when the structure changed, as chapter 10 said: a new line of business, a different customer, an acquisition, the loss of a capability the plan depended on. A rewrite also makes sense when the plan is being written for a different reader than the original — a plan written for a lender being adapted for a partner buying in is a different document.

CASE IN POINT Nordec Entretien: the Friday session and the one-deliverable rule This case is fictional; the company, the people and the figures are invented. Sophie Nadeau wrote the first version of her plan in eleven days, in the evenings. The second version took five weeks and she wrote it on Friday afternoons and Tuesday mornings — ten sessions, two a week, each three hours, each with one deliverable written at the top of the page before she started. Two sessions did not produce their deliverable. Session 5, the go-to-market, ran out of time because she could not reconstruct her own conversion rates from memory and had to go back through a year of calendar entries and invoices. Session 8, the second half of the model, produced a cash flow that did not tie to the balance sheet and she stopped rather than continuing. Both were rescheduled rather than absorbed. The reconstruction of the conversion rates became a separate two-hour task the following week, and it produced the funnel of chapter 5 — with real numbers, which is the reason the sales section survived a lender's reading. The cash flow was fixed in a session with her bookkeeper, who found the error in eleven minutes: the loan principal had been treated as an expense. Her observation afterwards was about the rule rather than the schedule. Writing the deliverable at the top of the page before starting meant that when a session was going badly, she could tell — and stopping a session that is going badly is cheap, while discovering at the end of a plan that a section is thin is not.

11.7 Board and Lender Reporting

11.7.1 The obligation is a date

A credit agreement typically requires financial statements on a schedule — annually, and often quarterly or monthly for an operating line. That schedule is a covenant, and delivering late is a breach even when the numbers are good.

Put every reporting date in the calendar the day the agreement is signed, with a reminder ten days before, and treat the delivery as a task with a verification criterion: delivered on or before the date in the agreement.

11.7.2 What a reporting pack contains

Four things, and consistency matters more than elaboration. The statements in the form the agreement requires. The covenant calculation, showing each ratio and its threshold — computing it yourself rather than leaving it to the lender is both faster and better received. A brief commentary, three or four sentences, on any material variance. And the updated forecast if anything has changed materially.

A pack that arrives in the same format every period, on the date, with the covenant arithmetic already done, changes the relationship with a lender more than any single meeting.

11.7.3 Board and advisory reporting

Where a board or advisory board exists, its pack is the quarterly review's one-page record plus the five numbers plus whatever decision requires its input. Sent five working days before the meeting, so that the meeting can discuss rather than absorb. → see Book H6 (Corporate Governance, Shareholder Agreements and Minute Books).

11.8 What Can Be Automated, and What Must Not Be

11.8.1 The test

One question separates the two: does this task require a judgment that would change with the circumstances? If yes, it must not be automated, though its inputs can be. If no, it is a candidate.

By that test, gathering, assembling, formatting, reminding, checking that something exists, and sending a recurring report on a schedule are all candidates. Deciding whether an assumption is still true, whether to reduce a target, whether to accept a covenant, or what to say to a client whose contract is at risk are all judgments and remain human.

11.8.2 The gate that must exist

Where automation touches anything that leaves the business or spends money, a human gate belongs in front of it. This is not a general principle about caution; it is a specific and defensible design. Five kinds of action warrant a stop: publishing something visible outside, sending in volume, spending money, sending a proposal that commits the business to terms, and deleting data. Four kinds do not: collecting, analyzing, qualifying and drafting — because each of those is private and reversible, and nothing has left.

That distinction is worth adopting whatever tools you use. The product described in In La Machine applies the same logic one automation at a time rather than as a single rule: the worst outcome an automation would produce if it misfired is written in the registry before it is built, and that sentence is what decides whether the automation carries the approval flag. → see Book J7 (Business Process Automation) for the discipline and Book N3 (Workflows, Automation and Human Approval) for the product's own treatment.

THE RULE A lender report is due on the date in the covenant, not when convenient Reporting obligations are covenants, and a covenant breach is a default even where every payment has been made and every ratio is comfortable. Three habits prevent the entire category of problem. Extract every date from the agreement the day it is signed and put it in the calendar with a ten-day reminder — do not rely on the lender's reminder, which arrives when you are already late. Deliver the same pack every period, so that assembling it is a routine rather than a project. And compute the covenant ratios yourself and include them, which converts an inspection into a report and, incidentally, means you learn about a breach before your lender does.

11.9 The Workflows Worth Building

Four workflows repay the effort of building them in a small business, and they are worth building in that order.

The monthly report to clients. Data already exists in the operating record; the assembly is mechanical; the sending is on a fixed date. In the running case it is the mitigation for the largest risk in the register, which means automating it protects $9,336 a year of expected loss for a few hours of setup.

The reporting reminder chain. Every reporting date, ten days ahead, with the pack's checklist attached. Purely mechanical and it prevents a covenant breach.

The weekly numbers digest. The five numbers, assembled and delivered on Monday morning, with last week's values beside them. The assembly is the automatable part; the reading is not.

The evidence collector. A single place where a statistic, a quote or a competitor's price goes with its date, reachable in fifteen seconds from wherever you are. Low technology and high return, because the reason plans contain unsourced figures is almost always that filing the source was inconvenient at the moment it arrived.

What is not worth automating in a business of this size: anything that runs less than monthly, anything whose inputs are not already in a system, and anything whose output requires a judgment before it is useful. → see Book J7 (Business Process Automation).

11.9.1 The four workflows, specified

A workflow is worth specifying before it is built, in four lines: the trigger, the steps, the gate, and what proves it ran. The four below are written that way so they can be handed to whoever builds them.

The monthly client report. Trigger: the first working day of the month. Steps: pull the previous month's interventions per building from the operating record; compute the count, the median response time and the outstanding items; assemble one page per client from a fixed template; queue for review. Gate: a person reads the batch before it sends, because this is a send in volume and a wrong figure reaches every client at once. Proof: the send log shows one report per portfolio client, dated on or before the fifth working day.

The reporting reminder chain. Trigger: ten days before each reporting date extracted from the credit agreement. Steps: raise a task with the pack's checklist attached, addressed to whoever keeps the books; escalate to the owner at five days if the task is not in progress. Gate: none needed — the workflow raises a task and never sends anything outside. Proof: every reporting date in the year has a task raised at least ten days ahead.

The weekly numbers digest. Trigger: Monday, early. Steps: read the five sources; assemble the five values with the previous week's beside them and the threshold from Table 10.2; deliver to whoever holds the weekly read. Gate: none — nothing leaves the business and nothing is decided. Proof: a digest exists for every Monday of the quarter, and each carries five values rather than four.

The evidence collector. Trigger: a person sends a link, a figure or a quote to one address. Steps: store it with the date received, the source and a one-line note; file it under the plan section it belongs to. Gate: none. Proof: every citation in the finished plan can be traced to an item in the collector with a date.

Three of the four have no gate, and that is the pattern: the automatable parts of planning work are the ones that gather and assemble rather than the ones that decide or send. The one that does send is the one that needed the gate, which is the test of §11.8.1 applied rather than recited.

WATCH OUT The workflow that decays between uses A routine that runs monthly stays healthy because somebody notices immediately when it breaks. A routine that runs twice a year breaks silently — a changed field name, a moved file, an expired credential — and is discovered at the moment it was needed, which is by definition the worst moment. Two rules follow. Do not automate anything that runs less often than monthly; do it by hand from a checklist, which never breaks. And for anything that does run monthly, check the output rather than the completion: a workflow that reports success while producing an empty report is the failure mode that costs the most, because the success message stops anybody from looking.

11.10 Cadence in Three Situations

11.10.1 Solo

Ten sessions is still ten sessions, and the realistic spacing is two a week over five weeks. The cadence after delivery compresses: daily ten minutes, weekly twenty, monthly forty, quarterly ninety. That is about seven hours a quarter, which is the entire ongoing cost.

The specific risk for a solo operator is that the review has no other attendee and therefore no accountability. The countermeasure is to send the one-page record to somebody — an accountant, a peer, a mentor — which converts it from a note to yourself into a small commitment.

11.10.2 Small team

The writing sessions are led by one person with contributors, and two sessions should include the whole team: session 6, where capacity meets the forecast, and session 10, the reconciliation pass.

The daily and weekly cadences are held by whoever runs operations. The quarterly review includes everybody who owns an objective, a project or one of the five numbers.

11.10.3 Established business

The sessions are distributed across functions and the risk is integration rather than production: four well-written sections that assume different things. Add an explicit integration session between sessions 6 and 7, before the model is built, where the assumptions from each function are reconciled into a single log.

At this size the reporting cadence is usually externally imposed, and the plan's own cadence should be aligned to it rather than run in parallel.

WHO DOES THIS Who holds each cadence, in three situations The daily pass belongs to whoever opens the business: the owner when solo, the operations lead in a team, a named person at any size. It cannot rotate, because its value is entirely in the accumulated sense of what normal looks like. The weekly numbers belong to the same person, with one adaptation in a team: each of the five numbers has an individual owner who is asked about theirs, so that the reading is a short round rather than a monologue. The monthly reconciliation belongs to whoever keeps the books, and it is the one cadence worth buying from outside if nobody internal can hold it reliably. The quarterly review and the annual refresh belong to the owner, always, and cannot be delegated. They are the two moments in the year where the content is judgment rather than execution, and delegating them produces a plan that describes somebody else's business.

QUÉBEC & CANADA NOTE The dates that already structure a Québec business's year A cadence works better aligned to obligations that exist anyway. Four families of dates matter. Corporate income tax filing and installment dates follow the fiscal year end and are set federally and provincially; the federal filing and remittance dates for corporations, for the GST/HST and for payroll deductions are published by the Canada Revenue Agency (Canada Revenue Agency, 2026a), and the corresponding Québec frequencies and deadlines by Revenu Québec (Revenu Québec, 2026). Sales tax remittance — GST and QST — runs monthly, quarterly or annually depending on the registration, and each remittance is a cash event that belongs in the monthly pass. Source deductions and employer contributions are remitted on a frequency set by the size of the payroll. And employment dates shape available capacity: the reference year for annual leave runs from 1 May to 30 April unless an agreement fixes another start, under sections 66 and following of the Act respecting labour standards, CQLR c N-1.1 (consolidation updated to April 7, 2026, read September 9, 2026), and the eight statutory holidays of chapter 6 fall on dates that are known a year ahead. Setting the four quarterly reviews beside these, once, means each review happens when the information exists rather than when the calendar happened to be free. → see Book G5 (Business Taxes in Québec and Canada).

TOOLS OF THE TRADE What a cadence needs from your tools Five capabilities, whatever the products. You need recurring tasks that return to a healthy repeating state rather than appearing overdue, or the list becomes unreadable within a quarter. You need a place that shows what is due and overdue today, without being sought. You need a calendar that holds the dates — reviews, reporting obligations, renewals — with reminders far enough ahead to act. You need a running file for the monthly three lines, which is the cheapest artifact in this chapter and the one that makes the quarterly review short. And you need the five numbers readable in one place, ideally assembled rather than collected by hand each week, because a weekly read that takes twenty minutes to assemble becomes a monthly read within two months. Chapter 13 names products; the operating screens described below cover three of the five.

WHAT CHANGED Covenant reporting has become more frequent for small facilities Where a small commercial facility once carried an annual reporting obligation, many now carry quarterly or even monthly requirements, particularly on operating lines secured by receivables. The practical consequence for a plan is that the reporting pack must be cheap to produce, because it will be produced twelve times a year rather than once: same format, same source data, covenant arithmetic already built. A pack that takes four hours to assemble is a pack that will be late, and lateness is a breach regardless of the numbers. Confirm your own obligations from the agreement rather than from what was usual. As of September 2026.

Costly Mistakes

1. Writing the plan against a deadline instead of in sessions. It gets written in the last week, and the financial section — which comes last — gets whatever hours remain.
2. Absorbing a failed session into the next one. Ten sessions become four, with six sessions of debt, and the debt is always paid by the model.
3. Treating recurring tasks as overdue. The list looks like a backlog of failures, becomes demoralizing, and stops being opened within a quarter.
4. Rewriting the plan annually instead of refreshing it. It costs thirty-one hours instead of eight and discards the record of what was assumed, which is what would have improved the new assumptions.
5. Missing a reporting date. It is a covenant breach and therefore a default, regardless of how good the numbers are and how comfortable the ratios.
6. Automating a judgment. Deciding whether an assumption still holds, whether to reduce a target, or what to say to a client at risk are judgments; automating their inputs is useful and automating the decision is not.
7. Building a workflow for something that runs less than monthly. The setup cost exceeds the saving, and the workflow decays between uses until it fails at the moment it is needed.

Key Takeaways

• Ten sessions with one deliverable each produce a plan in about thirty-one and a half hours, spread over four to six weeks.
• A session that does not produce its deliverable is rescheduled rather than absorbed, because absorbing creates debt that the financial section always pays.
• The task table has five readable columns — task, trigger, frequency, automation level, definition of done — with role and tools in prose beneath it.
• Five cadences keep a plan alive: ten minutes daily, twenty weekly, one hour monthly, ninety minutes quarterly, and six to ten hours annually.
• The annual refresh keeps the structure and updates what changed, and issuing a version with a change note is what makes the plan cumulative across years.
• Reporting dates in a credit agreement are covenants; late delivery is a default even when every ratio is comfortable.
• The automation test is whether the task requires a judgment that would change with circumstances; five kinds of action warrant a human gate and four do not.
• Four workflows repay their setup in a small business: the client report, the reporting reminder chain, the weekly numbers digest and the evidence collector.

Your Action Plan

1. Book the ten sessions in the calendar, on dates, with the deliverable in the title. Deliverable: ten calendar entries over four to six weeks. Timing: fifteen minutes. Verification criterion: each entry names what will exist when it ends.
2. Write the task table for your own business. Deliverable: a five-column table with your one-off and recurring tasks, plus a paragraph naming who does each and with what. Timing: ninety minutes. Verification criterion: every row's definition of done is checkable by somebody else.
3. Set the four cadences with their durations. Deliverable: recurring calendar entries for the daily, weekly, monthly and quarterly passes, each naming what it opens. Timing: twenty minutes. Verification criterion: the quarterly entries fall two weeks after each quarter end.
4. Extract every reporting date from your agreements today. Deliverable: a list of dates with what is owed to whom, each with a ten-day reminder. Timing: forty minutes. Verification criterion: each date is in the calendar and names the pack it requires.
5. Start the monthly three-line file. Deliverable: a running file with the variance, the trigger check and the assumption movement for the current month. Timing: forty minutes for the first entry. Verification criterion: the entry is dated and the trigger check names each trigger.
6. Choose one workflow to build, and build only that one. Deliverable: one automated or semi-automated routine, from the four in §11.9, with the human gate where it belongs. Timing: two to four hours. Verification criterion: it has run twice without intervention, and you can say what it saves.

In La Machine

Every route named in this section was verified in the product's route inventory on 2026-09-09 (La Machine, 2026a), and what each page holds is described in the product's user guide (La Machine, 2026b).

• The writing sprint as a project. /projets holds it as a one-off; each of the ten sessions is a task at /taches whose verification criterion is the session's deliverable. No page creates such a task, so the ten are loaded through the assistant, through the programming interface or by a script, and each of those paths refuses a task that carries no criterion. The recurring cadences are recurring tasks, which return to the status Active (recurring) rather than appearing overdue.
• The daily pass. /aujourdhui gathers what is awaiting approval, what is overdue, what is due today, what was completed today and this week, what has failed recently, and what is running or queued. Three limits belong with that list: the page has no filter of its own, it gives the count of the tasks due today without listing them, and it exports nothing. One caution belongs with it as well, because the tile that counts what was completed today compares days in universal time, so a reader in Québec who opens the page in the evening can see zero after a full day of work.
• What ran, and what it produced. /executions holds runs with their states; a task carries the definition of the work and each occurrence produces a distinct run. /registre holds the workflow registry, where an entry is To build, Under construction, In production, Paused or Removed — and being listed is not being in production, which is why the page's subtitle gives the total and, separately, the number in production.
• The gate. Five actions stop for a person — publishing, sending in volume, spending, sending a proposal, deleting data — and four never do: collecting, analyzing, qualifying and drafting. There is no page of approval requests as such: a waiting run carries Approve and Reject beside it on /executions and on /aujourdhui, and the same request also reaches the approver on a messaging thread whose every message is labeled with the business it belongs to. Approving writes the approver, the time and a note on the run itself; the buttons on the daily page act at once and send no note.
• The trace and the decisions. /journal records every creation, change and deletion with what, who and when, including a launch that was refused, and it names the actor of each event — a person, the assistant, the system or the engine — with filters on the actor, the entity, the action and the period; arbitrages holds decisions that could not be settled in the moment.
• Small supports for the sprint. /liens holds the addresses reopened daily — during a sprint, the statistical agency, the register and the lender's document list; /presse-papier holds notes that can be attached to a project or a task; /outils is the inventory of what the business runs on, where each tool carries one of six statuses, from To evaluate to Active, and where the version and the folder of a locally installed tool are measured on the machine rather than declared.
• What it does not do. La Machine does not assemble a reporting pack, does not compute covenant ratios, and does not hold your credit agreement's dates as obligations. It has no plan-refresh function and no version history for a manuscript, and no page of its tools inventory holds a cost or a renewal date. Across the productivity module the assistant only lists, counts and reads: none of its capabilities on that side writes anything.

Self-Check

1. Name the deliverable of session 1 and of session 8, and say what happens when a session ends without producing its deliverable.
2. Give the five columns of the task table and explain why the role and the tools are in prose rather than in two more columns.
3. State the five cadences with their durations, and total the ongoing quarterly cost in hours.
4. Give the test that separates what can be automated from what must not be. Apply it to "decide whether to reduce a target."
5. Your credit agreement requires quarterly statements. Describe what you would do the day it is signed, in three actions.

Chapter References

Act respecting labour standards, CQLR c N-1.1, ss. 60, 66–69 (Québec). LégisQuébec, consolidated version updated to April 7, 2026. Read September 9, 2026.

Canada Revenue Agency. (2026a). Filing and remittance due dates for corporations, GST/HST and payroll deductions. Retrieved September 9, 2026.

International Organization for Standardization. (2015). ISO 9001:2015 — Quality management systems — Requirements, clause 9.3 (management review). ISO.

La Machine. (2026a). Documentation — route reference inventory (211 routes: 191 Core, 20 Pilotage paths). Internal product documentation, read September 9, 2026.

La Machine. (2026b). Guide utilisateur [User guide] (FAQ: reading the Today page; pilotage, tasks and their statuses; executions and recurrences; approval by message; the workflow registry; links and notes; the tools inventory). Internal product documentation, read September 9, 2026.

Revenu Québec. (2026). Remittance frequencies and filing deadlines for businesses. Retrieved September 9, 2026.

Chapter 12 — Expertise, Roles and When to Get Help

A business plan requires six skills and almost nobody has all six at a professional level. That is not a problem; it is a purchasing decision, and it is one that small businesses routinely make badly in two opposite directions — buying a whole plan from a consultant when what was needed was four hours on the model, or buying nothing at all and discovering in a lender's office that the cash flow does not tie to the balance sheet. This chapter is about making that decision precisely: what each professional actually delivers, what to ask for, and how to tell a good engagement from an expensive one.

What You'll Be Able to Do

• Name the six skills a plan requires, and rate your own honestly against each.
• Distinguish what you can learn in a week from what you should buy.
• Choose the right accounting engagement — compilation, review or audit — and know what each one does and does not assert.
• Brief an advisor in one page, so that what comes back is usable rather than generic.
• Recognize the warning signs of an engagement that will disappoint, before it starts.
• Use the public advisory resources that already exist, and know what each is for.

Why It Matters

The financial section is where outside help pays for itself most reliably, and it is also where owners most often decline it. The reason is understandable — the model looks like arithmetic, and arithmetic feels like something one ought to be able to do — but the value a competent reviewer adds is not arithmetic. It is knowing what a lender checks, which classifications are conventional, where the statements must tie, and which assumption in your sector will be questioned first. Two hours of that is worth more than twenty hours of your own careful checking.

There is an opposite failure that costs more. A plan written entirely by a consultant is a plan you cannot defend, and the meeting where that becomes apparent is the meeting that mattered. A lender asks why the utilization rate is 68% and the answer must not be "that's what our consultant used."

Finally, the professionals you name in the plan are themselves information. A reader seeing an accountant, a lawyer or notary, an insurance broker and a banker named with their firms concludes that the business has the relationships a business of that size should have. It takes four lines.

In This Chapter

• 12.1 The Skills a Plan Actually Requires
• 12.2 What You Can Learn in a Week, and What You Cannot
• 12.3 The Accountant and the CPA: What They Add and When
• 12.4 The Business Consultant: Deliverables, Fees and Warning Signs
• 12.5 Lawyers, Notaries and What They Catch
• 12.6 Public Advisors and Development Bodies
• 12.7 Incubators, Accelerators and Mentors
• 12.8 Working With a Writer or an Editor
• 12.9 What the Software Does and Does Not Replace
• 12.10 Choosing, Briefing and Paying an Advisor
• 12.11 Expertise in Three Situations

12.1 The Skills a Plan Actually Requires

Six, and they are genuinely different from one another.

Writing clearly. Producing prose that a stranger understands on one reading, with a specific claim per paragraph. Learnable, and improved fastest by reading aloud.

Reading a financial statement. Knowing what each line means, what ties to what, and what a ratio tells you. Learnable in a week of deliberate effort and worth more than any other single skill in this list.

Building a model. Constructing a spreadsheet whose assumptions are separate, whose formulas are visible, and whose statements tie automatically. Partly learnable; the structural discipline is the hard part.

Gathering evidence. Finding a countable population, obtaining an official figure with its reference year, conducting twenty customer conversations and counting the answers. Learnable, and mostly a matter of persistence.

Estimating operations. Knowing how long work takes, what a utilization rate really is, and where capacity is actually lost. Not learnable from a book — it comes from having done the work, which is why an owner who came up through the trade has an advantage a business graduate does not.

Negotiating. Reading terms, understanding what is negotiable, and knowing what a guarantee commits. Partly learnable; the reading of legal documents is where outside help is cheapest relative to what it prevents.

12.1.1 Rating yourself honestly

Score each of the six from one to five for your own situation, in writing, and take the lowest two seriously. The point is not accuracy; it is that the exercise forces a decision about where help is worth buying, which otherwise gets made by whichever section feels most uncomfortable at the time.

The two that most often score lowest in a small business are building a model and negotiating, and they are the two with the cheapest, most bounded external solutions.

12.2 What You Can Learn in a Week, and What You Cannot

12.2.1 Worth learning

Reading a statement. A week of deliberate study — your own statements, line by line, with your bookkeeper answering questions — produces a permanent capability that improves every future decision. This is the highest-return week available to a small business owner. → see Book G1 (Accounting Fundamentals for Business Owners).

Basic modeling structure. The discipline of separating assumptions from calculations is a concept rather than a technique, and it can be learned in an afternoon and applied forever.

Finding official data. Where the statistical portals are, how classification codes work, what a reference year means. An afternoon.

12.2.2 Worth buying

The correct treatment of anything unusual. A capital lease, a related-party transaction, an asset purchase, a grant with conditions — each has a treatment that is conventional and non-obvious, and getting it wrong is visible to any professional reader.

Knowing what a lender checks. This is knowledge held by people who see files from the other side, and it is not documented anywhere you can read it.

The reading of a security document or a guarantee. Bounded, fast, and it prevents the category of surprise that lasts years. → see Book A5 (Financing a Business), whose chapter 3 sets out the instruments themselves and the security taken against them, the personal guarantee included.

12.2.3 The middle case

Between the two sits the most common and least examined option: buying an hour rather than an engagement. Many professionals will review a model or a section for a small, defined fee, and the results are frequently better than a larger engagement because the question is precise. The obstacle is usually that nobody asks, on the assumption that professionals only sell packages.

12.2.4 The week that teaches you to read a statement

The highest-return learning in this chapter is specific enough to schedule, and it fits in five sessions of about ninety minutes each. Use your own statements throughout; a textbook's example teaches the vocabulary and your own statements teach the business.

Session one — the income statement, line by line. Take your most recent annual statement and write, beside each line, what it actually contains. Where you cannot say, mark it and ask your bookkeeper. Most owners discover two or three lines they had assumed contained something else.

Session two — the balance sheet, and the one identity. Assets equal liabilities plus equity, always. Identify each item, say whether it is current, and compute two ratios: current assets over current liabilities, and total liabilities over equity. You now have the two numbers a lender computes first.

Session three — the connection between them. Find net income on the income statement and find where it went on the balance sheet. Find amortization and find the accumulated amortization it feeds. These are two of the three ties from chapter 7, and seeing them once in your own statements is worth more than reading about them five times.

Session four — the cash flow. If your accountant produces one, read it against the income statement and list every difference. If they do not, build one from your bank statements for three months, and compare it with the profit for those months. The gap is the education.

Session five — the questions. Take the marked items from all four sessions to your bookkeeper or accountant in one meeting. This is the session that converts the exercise from reading into understanding, and it is the reason the marks matter.

DO THE MATH What the week of learning is worth Five sessions of ninety minutes is 7.5 hours. At a loaded $58 an hour, the time costs $435, plus one hour of your accountant's time for session five. Against what? Three things, and each is measurable. The classification error. In the running case, the loan principal treated as an expense made the entire first cash flow meaningless and contributed to a refusal. An owner who had done session four would have found it, because the cash flow would not have matched the bank. The ratio conversation. Knowing your own current ratio and debt-to-equity before the meeting changes the first ten minutes of it from an examination into a discussion. There is no invoice for that and every owner who has experienced both knows the difference. The permanent capability. Every subsequent decision — a lease against a purchase, a hire against a subcontractor, a price change — is made with a clearer sense of what it does to the statements. The week is paid once and the capability lasts. The comparison worth making is not against zero. It is against the alternative, which is buying the same understanding one question at a time, at professional rates, for years — which is what most small businesses do, and which costs several times $435 in the first year alone.

12.3 The Accountant and the CPA: What They Add and When

12.3.1 Bookkeeper, accountant, CPA

The three terms are used loosely and they mean different things. A bookkeeper records transactions and produces the internal records that everything else is built on. An accountant prepares statements and returns and advises on structure. A Chartered Professional Accountant holds a professional designation, is subject to a regulatory body, and can perform engagements that carry a defined level of assurance — which is the part that matters for a plan.

12.3.2 The three levels of assurance

This is the distinction most business owners have never had explained, and it determines what a lender is actually getting.

DEFINITION Compilation, review and audit: three levels of assurance In a compilation engagement, governed in Canada by CSRS 4200, the practitioner assists management in preparing financial statements from information the client provides. The practitioner does not verify or test the data and expresses no assurance on the statements. In a review engagement, governed by CSRE 2400, the practitioner performs inquiry and analytical procedures and expresses limited assurance — a negative form of conclusion, stating that nothing has come to their attention causing them to believe the statements are materially misstated. In an audit, governed by the Canadian Auditing Standards, the practitioner performs substantially more work and expresses reasonable assurance — a positive opinion that the statements present fairly, in all material respects. The three standards are the ones published for Canadian practitioners (CPA Canada, 2026a), read September 9, 2026. The three cost very different amounts and assert very different things; a lender asking for "statements" may accept any of them, and asking which one they require before commissioning it is a two-minute question that occasionally saves a great deal.

The paragraph below introduces Table 12.1, which sets the three levels side by side. Its columns give what the practitioner does, what is asserted, who typically requires it, and what it means for a plan. Read the third column: it is where the practical decision is made, because the level required is determined by the reader rather than by the business.

Table 12.1 — Three levels of assurance: what each does, asserts and is required for

12.3.3 What to actually ask for

For a business plan, three bounded requests cover almost everything an accountant should be doing.

"Review the structure of my model and tell me whether the statements tie." A small, defined piece of work. Ask for it in writing and specify that you want the ties checked at each year end.

"Tell me whether my assumptions are within a defensible range for this sector." They see other files; you see one.

"Confirm the treatment of the capital purchase and the loan." The two items most often modeled incorrectly, and both are conventional matters with a right answer.

What not to ask for: that they write the financial section. It is your section and you will defend it.

12.4 The Business Consultant: Deliverables, Fees and Warning Signs

12.4.1 What a good consultant delivers

A consultant is worth engaging for a bounded deliverable that requires expertise you do not have and will not need again. Market research in an unfamiliar sector, a process design, a specific analysis, a facilitated planning session.

The engagement should name the deliverable, the date, the fee and its basis, what the consultant needs from you and by when, and what happens if the scope changes. Anything vaguer produces a disappointment that neither party can point at.

KEY CONCEPT You buy a deliverable, never hours An engagement defined in hours has no natural end and no test of completion. An engagement defined by a deliverable has both. The distinction survives every fee model — a consultant can be paid hourly and still be engaged for a deliverable, and the two are frequently confused. State what will exist when the work is done, in a sentence you could show a third party, and the whole relationship becomes checkable. The corollary matters for the plan: the consultant's deliverable is an input to your plan, not the plan. A market study, a process map, a facilitated session — these are materials you then use. A finished plan delivered by somebody else is a plan you cannot defend in the meeting where it counts.

12.4.2 Fee models

Four models exist and this book prints no amounts, for the reason it has given throughout: fee levels vary by market and by year, and a figure without a dated source is exactly what the rest of this book refuses. What is stable is the structure, and knowing it lets you read a proposal.

A fixed fee for a defined deliverable transfers the estimating risk to the consultant and is usually the right model for planning work. An hourly rate with an estimate and a cap is appropriate where the scope genuinely cannot be known in advance; insist on the cap. A retainer buys availability over a period and suits an ongoing advisory relationship rather than a one-time deliverable. A contingent or success fee ties payment to an outcome such as obtaining financing, and it carries a conflict worth naming: the consultant is paid for the approval rather than for the accuracy, which is not the same objective as yours.

Ask for the fee model in writing before the work begins, with what is included and what would be extra. → see Book I8 (Freelancers, Agencies and Outsourcing).

12.4.3 Warning signs

Five, and each is visible before the engagement starts.

A proposal with no deliverable named, only an approach and a duration.

A template answer to a scoping question. If the proposal could have been written before speaking to you, the deliverable will read the same way.

Reluctance to name the person doing the work. In small consulting engagements, the individual matters more than the firm.

A promise about the outcome — that financing will be obtained, that a grant will be approved. Nobody controls that.

Payment entirely in advance for a deliverable that does not yet exist. Staged payment against milestones is normal and protects both parties.

CASE IN POINT Nordec Entretien: the mandate that produced a refusal This case is fictional; the company, the people and the figures are invented. Before her first plan, Sophie Nadeau engaged a consultant to write it. The proposal described an approach and a four-week duration, named no deliverable beyond "a bankable business plan," and did not say who would do the work. She paid half in advance. What came back was twenty-eight pages, well formatted, containing a market section built top-down from a national figure, a revenue forecast that grew 30% a year with no driver behind it, and a financing request of $180,000 that appeared on page four without derivation. The consultant had asked her nine questions in total, by email. The plan was declined, and the reason the account manager eventually gave was not that it was a poor business — it was that the ask did not follow from anything in the document and that the forecast implied roughly twice the billable hours the operations section described. Two things are worth extracting. The engagement was structurally wrong before it began: no named deliverable, no named person, payment in advance, and — decisively — the consultant was buying the outcome of "a plan" rather than an input Sophie would use. And the second version cost less in money and more in her own hours: she wrote it herself in ten sessions and bought four hours of her accountant's time to check the model's structure and the treatment of the vehicle purchase. Her accountant found the loan principal treated as an expense in eleven minutes, which was the error that had made the first cash flow meaningless.

12.5 Lawyers, Notaries and What They Catch

12.5.1 What they add to a plan

Three things, each bounded.

Reading the security and the guarantee before signing, which is the single highest-value legal hour in this entire process. The four questions of chapter 8 are the brief.

Confirming the corporate facts — that the register agrees with reality, that the shareholder arrangements are what you believe, that any lease the plan depends on says what you think it says.

Flagging what the plan assumes and the law does not permit. A plan that assumes an employee can be released without notice, or that a restrictive covenant will be enforced as written, is assuming something a lawyer would correct in a sentence.

12.5.2 The Québec distinction

In Québec, notaries and lawyers are two distinct professions with two distinct regulators. Notaries are public officers with exclusive jurisdiction over certain instruments — real property transactions and mortgages, among others — and also practise in corporate and commercial matters. Lawyers represent parties and appear before courts. For a plan involving premises or a mortgage, a notary is the relevant professional; for a dispute or a contract negotiation, a lawyer. Both are verifiable through their respective orders' public registers. → see Book H1 (Contracts and Commercial Agreements) and Book A4 (Legal Structures, Registration, Permits and Licences).

THE RULE A service contract has obligations on both sides, set by law An engagement with any professional is a contract of service, and in Québec the Civil Code of Québec sets out its framework at articles 2098 and following (consolidation updated to April 7, 2026, read September 9, 2026). Three consequences are worth knowing before you sign anything. The provider has an obligation to act in the best interests of the client and to inform them of the progress of the work, which means "I did not know it was going badly" is a defect on their side rather than yours. The provider generally has the free choice of the means of performance — you buy the result, not the method, which is another reason to specify the deliverable rather than the hours. And the client may unilaterally resiliate the contract, subject to paying for what has been done and, in some cases, for injury caused — which means a bad engagement can be ended, at a cost you should ask about at the outset rather than discover. Get the engagement in writing, with the deliverable, the fee basis and the ownership of the work product stated. → see Book H1 (Contracts and Commercial Agreements).

12.6 Public Advisors and Development Bodies

A significant amount of competent advice is available at low or no cost, and it is systematically underused because it is not marketed.

The Business Development Bank of Canada publishes free templates and guides, including the business plan template and the financial plan template cited throughout this book, and it lends and advises (Business Development Bank of Canada, 2026a, 2026b, 2026c). Its published materials are a reasonable starting structure for anyone who wants a form to work against, and they were read on September 9, 2026.

Investissement Québec delivers loans, guarantees, equity and government programs, and its advisors can tell you which programs a business is eligible for — which is information that is genuinely hard to obtain from published pages alone (Investissement Québec, 2026).

Regional and local development bodies — community futures organizations, local investment funds, chambers of commerce, sector associations — provide advisory services, and their value is concentrated in local knowledge: who lends to businesses like yours in this region, which programs are actually being used, which advisor is worth the call.

Futurpreneur Canada pairs its lending with mentoring for up to two years, matched to each borrower (Futurpreneur Canada, 2026). For an entrepreneur aged 18 to 39, that mentoring is arguably worth more than the loan.

All four are named here as sources of advice; the money itself, and the choice between one source and another, belong to the financing book. → see Book A5 (Financing a Business), whose chapter 3 covers the lenders — banks, the Business Development Bank of Canada, credit unions and the federal guarantee program — together with the instruments, the security and the covenants that come with them, and whose chapter 4 covers the government programs and grants.

Two habits make these resources useful. Go with a specific question, not with "I need help with my business plan" — an advisor with a precise question gives a precise answer. And verify anything they tell you about a program's parameters at the program's own page, because advisors are as susceptible as anyone to a figure that has changed.

12.6.1 The five questions worth asking a public advisor

A public advisor's value is concentrated in knowledge that is not published, and five questions extract most of it. Ask them in this order, in a single call of about thirty minutes.

"Which programs is a business like mine actually eligible for?" Not which exist — which apply, given the sector, the size, the region and the stage. This is the question their published pages cannot answer.

"Which of those are actually being used by businesses my size this year?" A program that exists and is rarely approved is a poor use of three weeks of application effort, and advisors know which is which.

"What is the realistic timeline from application to funds?" This changes your cash flow, and it is routinely underestimated by everyone except the people who process the files.

"What causes applications like mine to be refused?" Better than asking what makes them succeed, because the failure modes are fewer and more specific.

"Who else should I be talking to?" The single most productive question in the set, because the network of regional bodies, sector associations and lenders is largely invisible from outside and entirely visible from inside.

Two things to bring: a one-page summary of the business — which you already have from chapter 3 — and a written note of what you have already tried. Two things not to bring: the full plan, which nobody will read in the meeting, and an expectation that the advisor will make an introduction on the first call.

WATCH OUT The grant that costs more than it pays Public funding is genuinely valuable and it has a cost that is invisible until you are in it: the application effort, the reporting obligations that follow, the constraints on how the money may be spent, and the timeline, which is frequently months. Before committing to an application, do the arithmetic honestly — the hours to apply, at your loaded rate, plus the hours of reporting over the life of the agreement, against the amount net of anything you would have spent anyway. For a small grant with heavy reporting, that arithmetic can come out negative, and it is far better to discover it in advance than after two months of work. The question to ask the advisor is blunt and entirely acceptable: how many hours does a typical application and its reporting take? → see Book G7 (Grants, Subsidies and Public Programs).

12.7 Incubators, Accelerators and Mentors

Incubators provide space, shared services and a cohort, usually for early-stage businesses, sometimes in exchange for equity and sometimes not. Accelerators provide a time-bounded program with intensive support, generally for businesses with growth ambitions and generally in exchange for equity. Neither is designed for an established local service business, and applying to one because it is prestigious is a poor use of a month.

Mentors are different and available to everybody. A mentor is somebody who has run a business of your kind and will talk to you regularly. The value is specific and hard to obtain elsewhere: they have already made the decision you are facing, and they will tell you what it actually cost.

Two practical points. Ask directly, because most people say yes and almost nobody asks. And come with a question and leave with a decision, because a mentoring relationship with no agenda decays into pleasant conversation within three meetings.

12.8 Working With a Writer or an Editor

A writer or an editor can genuinely improve a plan, and the distinction between the two matters.

An editor works on text you wrote: structure, clarity, consistency, and the removal of the fog. This is a good purchase, because the argument stays yours and the document gets better. Ask for a copy edit against a brief that names the audience and the length.

A writer produces text from your material. This is a poorer purchase for a business plan specifically, for the reason §12.4.1 gave: you cannot defend a document you did not write. Where a writer is used, the sections to hand over are the descriptive ones — the company section, the offer description — never the analysis or the financials.

The one place a writer earns their fee reliably is the executive summary, and only after you have written a draft. Compressing 25 pages into 560 words is a specialist skill, and an editor working from your draft will do it better than you will.

12.9 What the Software Does and Does Not Replace

12.9.1 What it does well

Software genuinely replaces four things. Arithmetic, reliably and without complaint. Format, producing a document that looks professional without design skill. Structure, in the sense that a template ensures no section is forgotten. And assembly, gathering data that already exists into a report.

Generative tools have added a fifth: drafting, meaning producing readable prose from notes. Used as a first draft that you then correct, edit and take responsibility for, this saves real time on the descriptive sections.

12.9.2 What it does not replace

Judgment about what is true. A tool cannot tell you whether 68% is the right utilization rate for your business; only your records can.

Evidence. No tool produces the twenty customer conversations, the countable population, or the dated quote from a named supplier.

Accountability. Somebody signs the plan and answers the questions. That is a person.

Sector knowledge held by people who see other files. What a lender checks first, which assumption is questioned in your sector, what a covenant in your industry usually requires — this is tacit knowledge held by practitioners.

12.9.3 The specific caution about generated text

Two failure modes matter for a plan specifically. Generated prose can produce confident, plausible figures that have no source, which is precisely the defect chapter 4 spends six pages preventing — every figure must trace to an assumption with a source, and a figure that arrived in a draft without one must be removed or sourced before delivery. And generated prose tends toward a generic register: sentences that could belong to any business, which is the exact quality that the substitution test of chapter 3 exists to detect.

The practical instruction is narrow. Use whatever tools help you produce a first draft faster, then apply the same two tests you would apply to a consultant's draft: can every figure be traced? and would this sentence become false with a competitor's name in it? The tools are named, with their pricing models and dates, in chapter 13.

12.10 Choosing, Briefing and Paying an Advisor

12.10.1 Choosing

Three questions, asked of anyone you are considering. Who exactly will do the work? Can you show me something comparable you have done? What do you need from me, and by when? The third is the most revealing: a professional who has thought about your engagement knows what they need; one who has not will say "just send whatever you have."

Verify professional registration where one exists — an order's public register takes two minutes — and ask for two references from businesses of your size rather than from their largest client.

METHOD Briefing an advisor in one page A one-page brief produces a better result than an hour of conversation, and it takes twenty minutes to write. Seven blocks. One: the deliverable. What will exist when this is done, in one sentence. Two: the audience. Who reads it, and what decision they must make. Three: the context. What the business does, in three sentences. Four: what you have already. The draft, the model, the statements, the evidence folder — listed, so nothing is redone. Five: the specific questions. Three to five things you want answered, phrased as questions rather than as topics. Six: the constraints. The date, the budget or the fee basis, the format. Seven: what you will do with it. Which section it feeds, and what happens next. The deliverable is one page, sent before the first meeting. The verification criterion is that the advisor's first response engages with your specific questions rather than restating their offering.

12.10.2 Paying

Stage payments against milestones rather than paying entirely in advance or entirely on completion. A common structure is a portion on engagement, a portion on delivery of a draft, and the balance on acceptance — with acceptance defined by the deliverable rather than by satisfaction.

Agree in writing who owns the work product and whether you may reuse it, because a market study you cannot include in a plan is not much use. Under a contract of service, the question is settled by what the contract says, so it should say something.

12.10.3 Ending an engagement that is not working

Say so early, in writing, naming the specific gap between the deliverable and what has arrived. Most engagements that go wrong go wrong slowly and could have been corrected at the halfway point, and the reason they are not is that nobody wants the conversation.

Where it cannot be corrected, a client may end a contract of service, subject to paying for the work performed and, in some circumstances, for injury caused. Ask about the terms of termination at the outset rather than at the moment you need them.

12.11 Expertise in Three Situations

12.11.1 Solo

Buy two things and nothing else: four hours of an accountant's time on the model's structure and the treatment of anything unusual, and one legal hour on any security or guarantee document. Everything else is learnable or postponable.

The free resources matter more at this size than at any other, and they are used less. A local development body, a mentor, and a bank advisor cost nothing and will each answer a question you would otherwise get wrong.

12.11.2 Small team

The same two purchases, plus one worth considering: a facilitated session for the moment where the capacity ceiling meets the sales forecast, if that conversation has stalled internally. A neutral third party is often what unblocks it, and it is a two-hour engagement rather than a project.

Assign the relationship with each professional to one person, so that questions are asked once and answers are recorded.

12.11.3 Established business

You have the relationships already, and the question is whether they are being used at the right moment. A common pattern at this size is an accountant engaged only for year-end work who would have caught a modeling error in twenty minutes if asked in March.

The purchase worth adding is an outside reader for the finished plan: an advisor, a board member or a peer who reads it cold and lists every inconsistency. It is the cheapest quality step available and it substitutes for the reconciliation pass being done by somebody who wrote the document.

The paragraph that follows introduces Table 12.2, which sets out the five professionals a plan might involve. Its columns give what each one delivers, the typical engagement shape, how they are usually paid, and the trigger that says it is time to call them. Read the trigger column: it is the practical content of this chapter, because knowing when is harder than knowing who.

Table 12.2 — Five professionals: deliverable, engagement, payment basis and the trigger to call

Figure 12.1 puts the same information on a timeline instead of in a table. The horizontal axis is the ten writing sessions of chapter 11, and each professional appears at the session where calling them is most useful, with the question to ask written beneath. The point of drawing it this way is that the when is what most owners get wrong: the accountant called at session 10 finds errors that were free to fix at session 7, and the lawyer called after signing finds nothing at all.

Figure 12.1 — Original figure. Who to call and when: the five professionals placed on the ten writing sessions, each with the question to ask.

WHO DOES THIS Who does what in three situations In a solo business, you hold every role and the purchasing decision is about which two hours to buy. The answer is nearly always the model review and the security document. In a small team, one person owns each professional relationship and the questions are collected before the call rather than asked as they arise. This alone typically halves what the relationships cost. In an established business, the professionals are already engaged and the failure is timing rather than access. Set the model review as a milestone in the writing sprint, not as a step after it. In every case, the person who signs the plan must be able to answer for every number in it. That is the line no engagement crosses, whatever it costs.

TOOLS OF THE TRADE What to ask of any professional you engage Six questions, before the engagement rather than during it. What exactly will exist when this is done? — the deliverable, in a sentence. Who will do the work? — a person, not a firm. What do you need from me, and by when? — the answer reveals whether they have thought about your file. What is the fee basis, and what would make it change? — in writing. Who owns the work product and may I reuse it? — settled at the start, not at the end. What happens if I need to stop? — asked calmly now rather than urgently later. Six questions, ten minutes, and they prevent the majority of engagements that disappoint.

QUÉBEC & CANADA NOTE Verifying a professional before you engage them Each regulated profession maintains a public register, and checking one takes about two minutes. The Ordre des comptables professionnels agréés du Québec lists CPAs and the firms permitted to perform public accounting engagements — which matters, because the assurance levels of §12.3.2 may only be performed by practitioners authorized to do so (Ordre des comptables professionnels agréés du Québec, 2026). The Barreau du Québec lists lawyers (Barreau du Québec, 2026) and the Chambre des notaires du Québec lists notaries (Chambre des notaires du Québec, 2026); both indicate whether the member is entitled to practise. The Autorité des marchés financiers lists insurance and financial-product representatives and firms (Autorité des marchés financiers, 2026). Management consultants are not a regulated profession in Québec, which is precisely why the engagement letter and the deliverable matter more there than anywhere else; CMC-Canada administers a voluntary certification, which is a signal rather than a license, and its holders are bound by a published code of conduct (CMC-Canada, 2026). Verify the registration, ask for two references from businesses your size, and get the engagement in writing. → see Book H8 (Regulated Activities, Professional Orders and Industry Rules).

WHAT CHANGED Advisory services now include more than they did Two shifts are worth knowing when scoping an engagement. Accounting practices increasingly offer bounded advisory work — model reviews, forecasting support, covenant modeling — alongside the compliance work that was historically their whole offering, which means the four-hour engagement recommended in §12.11.1 is easier to buy than it was. And remote engagements have become ordinary, which widens the pool beyond your own region and makes it practical to find someone with experience in your specific sector rather than simply someone nearby. Neither changes what to ask for; both change what is available when you ask. As of September 2026.

Costly Mistakes

1. Buying a whole plan from a consultant. You cannot defend a document you did not write, and the meeting where that becomes apparent is the meeting that mattered.
2. Declining a model review to save money. Two hours from somebody who reads statements for a living catches the classification and tie errors that make a cash flow meaningless.
3. Engaging without a named deliverable. The work has no test of completion, and the disappointment that follows cannot be pointed at by either party.
4. Commissioning the wrong assurance level. An audit where a compilation would have been accepted costs several times as much and asserts nothing the lender asked for; ask which one they require first.
5. Calling the lawyer after signing. The four questions of chapter 8 are worth an hour before, and nothing at all after.
6. Paying entirely in advance. Stage against milestones, with acceptance defined by the deliverable rather than by satisfaction.
7. Accepting a figure from a generated draft without sourcing it. Confident, plausible and unsourced is exactly the defect this book has spent nine chapters preventing.

Key Takeaways

• Six skills are required and almost nobody holds all six; rating yourself honestly turns a discomfort into a purchasing decision.
• Reading a financial statement is the highest-return week of learning available to a small business owner.
• Compilation, review and audit assert no assurance, limited assurance and reasonable assurance respectively, and the level required is determined by the reader rather than by the business.
• The three bounded requests worth making of an accountant are: check the ties, judge the assumptions against the sector, and confirm the treatment of the capital purchase and the loan.
• You buy a deliverable, never hours — and the deliverable is an input to your plan rather than the plan itself.
• Five warning signs are visible before a consulting engagement starts, and the most reliable is a proposal that could have been written before speaking to you.
• Public advisory resources are competent, low-cost and underused; go with a specific question and verify program parameters at the source.
• Software replaces arithmetic, format, structure, assembly and first drafts; it does not replace judgment, evidence, accountability, or knowledge held by people who see other files.

Your Action Plan

1. Rate yourself on the six skills, in writing. Deliverable: six scores with one line each. Timing: twenty minutes. Verification criterion: the two lowest are named, and each has a decision beside it — learn, buy, or accept.
2. Ask your lender which assurance level they require. Deliverable: a written answer naming compilation, review or audit. Timing: one email. Verification criterion: you have it before commissioning anything.
3. Book four hours of an accountant's time at session 7, not session 10. Deliverable: a scheduled engagement with the three bounded requests written out. Timing: fifteen minutes to arrange. Verification criterion: the engagement names the ties, the assumptions and the treatment of the capital purchase.
4. Write the one-page brief before any advisor conversation. Deliverable: seven blocks on one page. Timing: twenty minutes. Verification criterion: the advisor's first reply engages with your specific questions.
5. Verify the registration of every regulated professional you engage. Deliverable: a note of the register checked and the date. Timing: two minutes each. Verification criterion: each name appears in its order's public register as entitled to practise.
6. Make one call to a public advisor with a specific question. Deliverable: a question written down beforehand, and the answer recorded. Timing: thirty minutes. Verification criterion: the question was answerable in one call, and you know what you will do with the answer.

In La Machine

Every route named in this section was verified in the product's route inventory on 2026-09-09 (La Machine, 2026a), and what each page holds is described in the product's user guide (La Machine, 2026b).

• The people the business works with. crm/agents, crm/agent-profile and crm/agent-import hold agents, and crm/clients, crm/client-details and crm/client-profile hold clients; none of these pages has a menu entry, and they are reached from a contact, from the list or from the tree view. An advisor is recorded as a client, with Add Client, which asks for a first name, a last name and an email address, and offers a source, tags and an internal note; the agents page cannot add anybody, because its Add Agent button is disabled on the stated ground that creating an agent is a design decision for the organization rather than an action on a screen. No field holds an engagement date, so the dates and the questions go in the note and in the tasks below. crm/organizations and crm/organization-profile are not a directory of outside firms: they read the workspaces you belong to and the people invited into them.
• The engagement as work. An advisory engagement is a one-off project at /projets, with the brief, the questions and the deliverable as tasks at /taches, each carrying its own verification criterion, which the product refuses to omit when a task is created. No page creates such a task: the entry is made through the assistant, through the programming interface or by a script. /aujourdhui surfaces what is overdue, which in an engagement is usually the material the advisor is waiting on from you; the page has no filter of its own, so what it shows is the whole of the plan's overdue work rather than the engagement's alone.
• Where the questions are collected. /presse-papier holds notes that can be attached to a project or to a specific task — which is exactly what a running list of questions for the next accountant call is, and attaching it to the task is what makes it findable by whoever opens that task.
• Getting help with the product itself. support holds support, tutorials holds tutorials, test-guide holds the guide, and wish-list and feature-wish hold requests for features, each item carrying a status that runs from idea through to delivered. → see Book N1 (Getting Started with La Machine) for how help is obtained inside the product.
• What it does not do. La Machine does not hold engagement letters, does not track professional registrations, and has no directory of advisors. It will not tell you which assurance level a lender requires. What it holds is the record of who you work with and the work each engagement generates.

Self-Check

1. Name the six skills a plan requires, and say which one cannot be learned from a book and why.
2. A lender asks for "financial statements." Name the three levels of assurance and say what you would do before commissioning anything.
3. Give three bounded requests worth making of an accountant during a writing sprint, and say at which session each belongs.
4. List four of the five consulting warning signs. Which one is visible from the proposal alone?
5. A generated first draft contains the sentence "the regional market has grown 11% annually since 2022." State the two tests you would apply before that sentence stays in the plan.

Chapter References

Autorité des marchés financiers. (2026). Register of firms and representatives. Retrieved September 9, 2026.

Barreau du Québec. (2026). Tableau de l'Ordre — public register of lawyers. Retrieved September 9, 2026.

Business Development Bank of Canada. (2026a). Business plan template for entrepreneurs (downloadable template with guide, instructions, glossary and worked example). Retrieved September 9, 2026.

Business Development Bank of Canada. (2026b). Financing solutions for small and medium-sized businesses; advisory services. Retrieved September 9, 2026.

Business Development Bank of Canada. (2026c). Free financial plan template for entrepreneurs. Retrieved September 9, 2026.

Chambre des notaires du Québec. (2026). Public register of notaries. Retrieved September 9, 2026.

Civil Code of Québec, CQLR c CCQ-1991, arts. 2098–2129 (contract of enterprise or for services). LégisQuébec, consolidated version updated to April 7, 2026. Read September 9, 2026.

CMC-Canada. (2026). Certified Management Consultant designation and code of conduct. Retrieved September 9, 2026.

CPA Canada. (2026a). CSRS 4200 — Compilation engagements; CSRE 2400 — Engagements to review historical financial statements; Canadian Auditing Standards. Retrieved September 9, 2026.

Futurpreneur Canada. (2026). Frequently asked questions (mentoring for up to two years, matched to each borrower). Retrieved September 9, 2026.

Investissement Québec. (2026). Financial products and advisory services. Retrieved September 9, 2026.

La Machine. (2026a). Documentation — route reference inventory (211 routes: 191 Core, 20 Pilotage paths). Internal product documentation, read September 9, 2026.

La Machine. (2026b). Guide utilisateur [User guide] (FAQ: CRM and contacts; projects and task chains; getting help; the wish list). Internal product documentation, read September 9, 2026.

Ordre des comptables professionnels agréés du Québec. (2026). Public register of members and firms. Retrieved September 9, 2026.

Chapter 13 — Tools: Functions and Popular Tools

The tool question is the one business owners ask first and should ask last. Asked first, it produces a subscription to something that does part of a job nobody has defined; asked last, it produces a short list of things you already own. This chapter names the six functions a plan-writing toolkit must cover, in a form that will still be true in five years, and then names products — with their pricing models rather than their prices, because a price printed without a dated vendor page is precisely the kind of number the rest of this book refuses.

What You'll Be Able to Do

• Name the six functions a plan-writing toolkit covers, independent of any product.
• Choose on the function rather than on the brand, and say what each tool must do before naming one.
• Apply the free-and-already-paid rule, which for most small businesses settles four of the six functions immediately.
• Say what a spreadsheet does better than anything else, and what it does badly.
• Situate La Machine accurately in the chain, including the four functions it does not perform.
• Assemble a toolkit in four decisions rather than in eleven subscriptions.

Why It Matters

Two costs follow from getting this wrong, and neither is the subscription fee.

The first is fragmentation. A plan whose evidence is in one place, whose model is in another, whose tasks are in a third and whose reports come from a fourth is a plan that decays, because keeping four things current is four times the work of keeping one. The fee for the fourth tool is trivial; the maintenance is not.

The second is the model you cannot send. A financial model built inside a proprietary planning application may be unexportable in a form a lender's analyst can open and check. Since chapter 8 established that the model will be requested, that is a constraint worth knowing before you build in it rather than after.

There is also a quieter reason. Most small businesses already own tools covering four of the six functions in this chapter, and the money spent on new ones is frequently spent to avoid learning the ones they have.

In This Chapter

• 13.1 The Six Functions a Plan-Writing Toolkit Must Cover
• 13.2 Function 1 — Writing and Version Control
• 13.3 Function 2 — Financial Modelling
• 13.4 Function 3 — Market and Competitive Research
• 13.5 Function 4 — Presentation and Pitch
• 13.6 Function 5 — Execution: Objectives, Projects and Tasks
• 13.7 Function 6 — Reporting and Review
• 13.8 Popular Tools, by Function
• 13.9 The Free and Already-Paid Rule
• 13.10 La Machine in This Toolkit
• 13.11 Choosing a Toolkit in Three Situations

13.1 The Six Functions a Plan-Writing Toolkit Must Cover

Functions are stable; products are not. A toolkit assembled around functions survives a product being discontinued, repriced or acquired, which happens often enough to matter over the life of a business.

Function one: writing and version control. Produce long-form text, revise it, and know what changed between versions.

Function two: financial modeling. Build a model whose assumptions are separate, whose formulas are visible, and whose statements tie.

Function three: market and competitive research. Find countable populations, collect evidence with its citation, and keep the two together.

Function four: presentation and pitch. Produce a short version that reads without a presenter.

Function five: execution. Turn objectives into projects and tasks with owners, dates and verification criteria, and show what is due.

Function six: reporting and review. Read the five numbers, produce the reporting pack, compare forecast with actuals.

Figure 13.1 shows the six functions as a chain, with what each one produces and where it hands off to the next. The handoffs are the point: the research function produces the evidence the modeling function consumes, the modeling function produces the figures the writing function quotes, and the execution function consumes the milestones the writing function produced. A toolkit whose handoffs require retyping is a toolkit that will not survive the first amendment.

Figure 13.1 — Original figure. The six functions as a chain: what each produces, where it hands off, and the return path that makes this year's actuals into next year's evidence.

13.2 Function 1 — Writing and Version Control

13.2.1 What it must do

Five things. Produce long-form structured text with headings, tables and images. Keep a history of versions you can return to, so that version 1.0 can be opened a year later. Allow more than one person to work without overwriting each other. Export to PDF with tables intact, since a re-flowed financial table is worse than no table. And work offline, at least partly, because a plan is frequently written in the places where connectivity is worst.

13.2.2 What matters less than people think

Templates matter less than they appear to. A template guarantees that no section is forgotten, which chapter 2's table of contents already does at no cost. Design features matter less still: a plan is judged on its arithmetic, and a beautifully formatted document with an untied cash flow fails in exactly the same way an ugly one does.

The paragraph below introduces Table 13.1, the writing and version-control tools most commonly used by small businesses. Its columns give the tool, what it is best for, its pricing model — a model, not an amount — and a note on its limitation. No price is printed anywhere in this chapter, for the reason given at the outset; verify current pricing on the vendor's own page and record the date you read it.

Table 13.1 — Writing and version control: what each is best for, and its pricing model

13.3 Function 2 — Financial Modelling

13.3.1 What it must do

Six things, and they are the six from chapter 7. Keep assumptions in one place, separate from the statements. Show its formulas, since a model whose logic is hidden cannot be reviewed. Produce monthly output for at least the first year. Tie the three statements automatically. Support scenarios without duplicating the model. And be sendable, because a lender will ask.

13.3.2 Why the spreadsheet still wins

A spreadsheet does all six, it is understood by every accountant and every analyst who will read your model, and it exports to a format anybody can open. Those three properties are why it remains the default for small-business planning despite forty years of alternatives.

Its weaknesses are real and worth naming. Errors are easy and invisible: a formula that points at the wrong cell produces a plausible number. Structure is not enforced: nothing stops you from typing a value into a formula. And collaboration is awkward beyond two people.

13.3.3 Six habits that make a spreadsheet model reviewable

The weaknesses above are addressed by discipline rather than by software, and six habits cover nearly all of it. They cost nothing and they are what an accountant is looking for when they open your file.

An assumptions tab, first. Every input lives there, once, with its label, its unit, its source and its date. Nothing downstream contains a typed number.

One color convention, applied. Inputs in one color, formulas in another, links to other tabs in a third. This is the single fastest way for a reviewer to spot a hard-coded value where a formula belongs, and it takes no time once adopted.

One row per concept, one column per period. Resist the temptation to compress two ideas into one row; a row that holds both a rate and an amount cannot be checked.

A check row on every statement. A row that computes something that must equal zero — assets minus liabilities minus equity, or closing cash minus balance-sheet cash. If it is ever not zero, it is visible immediately rather than three months later.

No hidden rows or columns. A hidden row is a place where an error lives undisturbed, and a reviewer who finds one wonders what else is hidden.

A version and a date in the file name. The same convention as the document, for the same reason: two versions circulate, and somebody comments on the wrong one.

Applied together, these six turn a spreadsheet from a private calculation into an auditable model — which is exactly the difference between a file your accountant can review in an hour and one they decline to review at all.

KEY CONCEPT The spreadsheet is not the weak link; the undocumented assumption is Blaming the spreadsheet for modeling errors is a category error. The errors that matter in a small-business plan are almost never arithmetic — they are an assumption typed directly into a statement, a classification changed between years, a growth rate applied where a driver belonged, a loan principal treated as an expense. None of those is prevented by a different application, and all of them are prevented by the assumption log of chapter 7 and the rule that no number is hard-coded inside a formula. Buy structure from your own discipline first; if you then want to buy it from a product as well, that is a reasonable second purchase and a poor first one.

The paragraph that follows introduces Table 13.2, the modeling tools. Its columns are the same four: the tool, what it is best for, the pricing model, and a note. Read the note column with §13.3.1's sixth requirement in mind — being sendable — because that is where the dedicated planning applications differ most from the general ones. The row worth opening first costs nothing: the Business Development Bank of Canada publishes a free financial plan template, alongside the business plan template this book has cited from chapter 2 onward (Business Development Bank of Canada, 2026a, 2026c).

Table 13.2 — Financial modeling: what each is best for, and its pricing model

13.4 Function 3 — Market and Competitive Research

13.4.1 What it must do

Four things. Reach official statistical sources with classification codes and geography. Collect interview and survey responses in a countable form. Store evidence with its citation and the date read, so the two never separate. And keep a competitor record that is updated rather than rewritten, because the useful information is the change over time.

The first of these is free and public, and it is the one most often replaced by a paid subscription that answers a different question.

The paragraph below introduces Table 13.3, which lists research sources and tools. Note that the first four rows are public and free, and that they are where the bottom-up funnel of chapter 4 actually comes from; the paid rows serve narrower purposes.

Table 13.3 — Market and competitive research: what each is best for, and its pricing model

13.5 Function 4 — Presentation and Pitch

13.5.1 What it must do

Three things, and one of them is unusual. Produce ten to fifteen slides that read without a presenter, since chapter 3 established that the deck is now often a screening document. Keep the figures identical to the document, which is a discipline rather than a feature. And export to PDF, because that is how it will be sent.

The paragraph that follows introduces Table 13.4, the presentation tools. The differences between them matter less here than in any other function, which is itself worth knowing: this is the function on which to spend the least time deciding.

Table 13.4 — Presentation and pitch: what each is best for, and its pricing model

13.6 Function 5 — Execution: Objectives, Projects and Tasks

13.6.1 What it must do

Six things, and they are the six from chapter 10. Hold a task with an owner, a date and a verification criterion, ideally refusing to complete one whose criterion is not met. Distinguish recurring work from overdue work. Support dependencies. Surface what is due and overdue today. Keep a trace of what changed, by whom and when. And hold the numbers you chose to watch.

Most general task applications do the first four partially and the last two not at all. The verification criterion in particular is rare: most tools treat "done" as a checkbox rather than as a claim that must be justified.

The paragraph below introduces Table 13.5, the execution and reporting tools, which covers functions five and six together because in practice the same product often serves both. Read the fourth column against the six requirements above: the gap between what a tool does and what the function needs is usually filled by a discipline rather than by a second product.

Table 13.5 — Execution and reporting: what each is best for, and its pricing model

13.7 Function 6 — Reporting and Review

13.7.1 What it must do

Four things. Produce the reporting pack in the same format every period, so that assembling it is a routine. Compare forecast with actuals without retyping either. Compute the covenant ratios, or at least hold the inputs in one place. And hold the monthly three lines — the variance, the trigger check, the assumption movement — somewhere they accumulate.

The fourth is the cheapest and the most neglected. It can be a text file.

13.7.3 Assembling the pack without rebuilding it

The reporting pack of chapter 11 is produced twelve times a year in many small facilities, which makes its assembly cost the thing to optimize rather than its appearance.

Three structural choices make it cheap. Build it once as a template with linked figures rather than as a document retyped each period; the figures come from the accounting export and the model's actuals tab, and nothing is keyed twice. Keep the covenant arithmetic in the model, in its own small block, so that each period's ratios are computed rather than recalculated. And fix the commentary at three or four sentences, on the same three topics every period — the variance, anything that changed materially, and what you are doing about it — because a commentary whose scope is decided fresh each month takes an hour and one whose scope is fixed takes ten minutes.

A pack assembled this way takes about thirty minutes a period after the first one. A pack rebuilt each time takes two to four hours, and is therefore late, which chapter 11 established is a covenant breach regardless of the numbers inside it.

13.7.2 Where the actuals come from

From the accounting system, which is the one tool in this chapter that is not optional. A business whose books are not current cannot compare forecast with actuals, cannot compute a covenant ratio, and cannot answer a lender's question without a delay that is itself information.

Getting the books current is therefore a prerequisite for the entire cadence of chapter 11, and it is the first thing to fix if it is not true. → see Book G1 (Accounting Fundamentals for Business Owners).

13.8 Popular Tools, by Function

The five tables above are the list. Three observations about reading them.

The tables give pricing models, not prices. Every vendor's prices change, frequently more than once a year, and a printed amount is stale before the book is opened. Verify on the vendor's own page, record the date you read it, and put the figure in your own assumption log rather than trusting any third-party summary — including this one.

The lists are not exhaustive and do not need to be. Each function has more products than are listed; what matters is that you can state what the function must do before you compare products, which is what §13.2 through §13.7 provide.

Being listed is not an endorsement. These are tools commonly used by small businesses for these functions, named so that the functions are concrete. A tool that is not listed and covers the function is a perfectly good choice.

THE RULE No price is printed without a dated vendor page This rule governs the whole book and it applies most visibly here. A price stated without its source and its date cannot be verified, cannot be updated, and will be wrong within a year — which makes it worse than no price at all, because a reader who checks and finds it wrong discounts everything around it. Where a vendor's current price is needed for your own plan, obtain it from the vendor's own pricing page, record the date you read it, and enter it in the assumption log with that date. Where a price cannot be obtained — increasingly common, as more vendors move pricing behind a contact form — state the model instead: per user per month, tiered by feature, one-time license, included with a suite you already hold. The model is what determines how the cost behaves as you grow, and it is the part that actually affects the forecast.

DEFINITION Template, model, scenario, version A template is a pre-structured document or spreadsheet supplying the shape of a deliverable without its content. A model is a set of linked calculations that turns assumptions into statements; its defining property is that changing an input changes every dependent figure. A scenario is one coherent set of assumption values, so that a model with three scenarios has one structure and three input sets rather than three copies. A version is a dated snapshot of a document or model as it stood at a moment, kept rather than overwritten, which is what makes the annual refresh of chapter 11 cumulative.

13.9 The Free and Already-Paid Rule

13.9.1 The rule

Before buying anything, use what is free and what you already pay for. Applied to the six functions, this settles most of them for most small businesses.

Writing: you almost certainly already have a word processor. Modeling: you almost certainly already have a spreadsheet. Research: the official portals are free and they are the primary sources. Presentation: included in the same suite as the word processor. That is four of six, at no additional cost.

The two that may justify a purchase are execution and reporting, and even there the free-and-already-paid rule applies first: a task application included in a business suite you already hold covers most of what chapter 10 requires.

13.9.2 When buying is justified

Three conditions, and it is worth requiring all three rather than any one.

The function is genuinely not covered by anything you have, rather than covered awkwardly.

The gap costs measurable time, which you can state — hours a month, not "it would be nicer."

The tool will be used by whoever actually does the work, which is the condition most often violated: a tool chosen by the owner and used by nobody else costs its subscription and produces a second place where things are recorded.

DO THE MATH What a subscription actually costs over three years The published monthly figure is not the cost, and the arithmetic below is worth doing before any purchase. The multipliers are the point; substitute the current price from the vendor's own dated page for the letter P. The license. A per-user subscription at P per user per month, across 4 users, over 36 months, is 4 × 36 = 144 user-months, or 144 × P. A tool at a modest monthly price is a four-figure commitment over a plan's horizon. The setup. Configuring it, importing what exists, and agreeing conventions: realistically 8 to 20 hours at a loaded $58, or $464 to $1,160, paid once. The learning. Four people at 3 hours each is 12 hours, or $696, plus a period of reduced output nobody measures. The maintenance. Someone keeps it current — roughly 1 hour a month, which over 36 months is 36 hours, or $2,088. The exit, if it comes. Getting your data out and into something else, once, at 10 to 30 hours. The total, excluding the license itself: about $3,250 to $3,950 of time over three years, against a license of 144 × P. For most small-business tools, the time exceeds the license — which reverses the usual comparison, in which a business agonizes over a monthly fee and ignores the hours. The practical consequence is the third condition of §13.9.2. A tool nobody else opens costs the license and the setup and the maintenance, and returns nothing, which is why "who will use it" decides the purchase more than any feature comparison.

WATCH OUT Buying planning software before knowing what the plan must contain The sequence that wastes the most money in this chapter is buying a tool and letting its structure define the plan. Planning applications embed assumptions about what a plan contains, and those assumptions are frequently written for a different kind of business — a venture-backed startup rather than an operating service company, or an American lending context rather than a Canadian one. The result is a plan with sections nobody asked for and without the section your lender opens first. Write the table of contents of chapter 2 before evaluating any tool, and then ask each candidate one question: can it produce this? The order matters; reversed, the tool answers a question you had not yet asked.

13.10 La Machine in This Toolkit

La Machine sits in functions five and six, and it is worth being precise about what that means for a business plan — including the four functions it does not perform at all.

What it does. It holds objectives as projects, each either a chain or a one-off, and the work as tasks, each requiring a verification criterion at creation: a task submitted without one is refused, with a message saying that such a task cannot be completed. No page creates a plan task at all, so the entry is made through the assistant, through the programming interface or by a script, and each of those paths demands the criterion. It is worth knowing where the enforcement stops, because chapter 10 leans on it: the assistant refuses to mark done a task whose criterion is empty and offers to drop it instead, while the drawer lets a criterion be emptied afterwards and the Mark as done button on screen does not check that one is there. It distinguishes recurring work from overdue work, so a task list of healthy repeating items does not read as a backlog of failures. It supports dependencies, and it makes them behave correctly: a dependent task waits for a done task, and a pivotal task that is dropped leaves its dependants blocked. It surfaces the day's demands on a single screen. It keeps a trace of every creation, change and deletion with what, who and when — including refusals. And it holds a registry of automations with each one's real state, plus a record of open decisions.

What it does not do, said plainly. It does not write the plan and holds no business-plan template. It does not model financial projections: no income statement, no cash flow forecast, no balance sheet, no break-even, no scenarios. It does not size a market, connect to statistical agencies, or hold industry data. It does not assemble a reporting pack or compute a covenant ratio. It has no version history for a manuscript and no objectives object as such — an objective becomes a project, which is the closest equivalent. Across the productivity module the assistant is read-only: its capabilities there list, count and read tasks, projects, alerts and alarms, and none of them writes.

What that means practically. Functions one through four are covered by the tools of §13.2 through §13.5, and function five is where La Machine does work a general task application does not, because of the verification criterion and the distinction between recurring and overdue. Function six is shared, and shared unevenly: the operating numbers are counted on its screens rather than assembled by it, since its dashboard measures the plan's own progress and offers no period, no export and no field for a figure you chose yourself, while the financial numbers come from the accounting system.

The honest summary is that a business using La Machine still writes its plan in a word processor and builds its model in a spreadsheet. What changes is what happens after the plan is delivered, which chapter 10 argued is the part most plans lose. → see Book N2 (Piloting Tasks, Executions and Projects) and Book N3 (Workflows, Automation and Human Approval).

CASE IN POINT Nordec Entretien: the spreadsheet that lasted three years This case is fictional; the company, the people and the figures are invented. Sophie Nadeau's model was a spreadsheet with six tabs: assumptions, revenue, costs, income statement, cash flow, and a tab of the actuals pasted in monthly. It was built in the spreadsheet application she already owned as part of the suite that ran her invoicing, and it cost nothing additional. In its second year she looked at three dedicated planning applications after a peer recommended one. All three produced a more polished document than hers. Two of the three could not export a model her accountant could open and check, which ruled them out on the sendability requirement of §13.3.1. The third could, and she did not switch, for a reason that had nothing to do with features: her bookkeeper knew the spreadsheet, updated the actuals tab monthly without being asked, and would not have updated a new application. What did change was function five. The plan's milestones had lived in the document, which meant nobody looked at them after March. Moving them into projects and tasks — with the criteria written as chapter 10 requires — was the change that made the second year's review possible in ninety minutes rather than in an afternoon of reconstruction. Her summary, at the second annual refresh: "The spreadsheet was never the problem. The problem was that the plan and the work lived in different places, and only one of them had a Monday."