$10M → $100M

This document has two halves.

The free half is Chapters 0 and 1. It scores whether your company is ready for any of this, on five gates, with numbers instead of opinions. Then it has you draw a map of where your revenue is dangerously concentrated. Then it shows you the one conversion this whole book is about: turning what one person knows into what the company knows.

Read only that far and you still walk away with a score, a map, and one area named as the place to start. Most companies at $30M have none of the three.

The paid half is Chapters 2 through 12, plus the template pack. That is the machine. How to hand over judgment instead of tasks. How to hear the truth once you have forty polite people between you and the customer. How to become the only real option in a game you defined. How to turn your reputation from something you personally carry into a job with owners and a budget. How to catch demand you currently find by luck. How to pick a way of funding growth on purpose instead of drifting into one. How to redesign the company when routine work costs almost nothing. And how to tell, from evidence rather than revenue, that you actually got there.

The decision in this chapter: is your company ready to turn what it knows into a system, or would this book just make it more fragile?

Every case in this book is a composite. The numbers are rounded, the industries sometimes moved, the details changed. What is not changed is the shape of the mistake. I have either made it, watched a client make it, or paid for it.

The idea, in one page

Playbook I was about earning belief. Playbook II was about building confidence — yours, your team's, your market's — by repeating the promise-and-deliver loop until it got boring.

This book is the step after that, and it is a different kind of work.

Confidence is personal. It lives in you and in your best three or four people. It is the thing you notice in a meeting and act on without being able to explain how you noticed it. Confidence grows slowly at best. It goes on holiday. It gets sick. It resigns on a Tuesday and carries eleven years of judgment out of the building in a cardboard box.

Knowledge is institutional. It lives in the product, in the standards, in the brand. It lives in the new hire who, in month four, makes the call you would have made — not because she asked you, but because the company taught her how you think. Knowledge outlives people. Including you.

All of Playbook III is one conversion: turn what you know into what the company knows.

Most founders refuse to pay for that conversion, and the reason is almost never money. It is identity. At $10M you are still the smartest person in the room about your own business, and that feels good. At $100M, if you are still the smartest person in the room about your own business, you have built yourself a very expensive job.

What changes at this size

Three things, and they change together.

• Your mistakes multiply

  At $1M a bad hire costs you a quarter. At $30M a bad standard spreads through 140 people, gets baked into onboarding, and costs you two years. You will not notice for about eighteen months. The feedback gets slower right when the stakes get higher. That is why this book moves slower than the first two. Fewer attempts, heavier decisions, and no way to iterate your way out of a structural mistake.

• How you fund the company becomes a strategy

  Below $10M, money is mostly a survival question. Above it, the choice between staying self-funded, raising, investing, or running a fund is the biggest single lever on where you end up. It is also the choice most often made by drifting.

• Nobody tells you the truth anymore

  Not because your people are dishonest. Because politeness stacks. Forty people sit between you and the customer, each one softens the message a little, and the fire reaches you as a warm patch. Truth has to be engineered now, the way you engineered your finance reporting. On purpose, with owners, on a schedule.

How the chapters work

One decision, one tool, one 90-day move per chapter. No theory without a tool attached.

Each chapter ends with the same four sections.

• Diagnostic

  Where you actually are, counted or scored, never guessed.

• The 90-day move

  What to run, in what order, with a test that tells you when you are done.

• Bias watchlist

  The specific mental trap that ruins this chapter's work, and how to block it. These are not filler. Believing you control more than you do, judging decisions by how they turned out, defending a system because you built it, throwing more money at a losing bet — these are the failure modes of people who have already succeeded. Nobody warns you about them until they get expensive.

• Annotated example

  One situation shown three ways: unacceptable, adequate, excellent, with the reasoning that separates them. This is PEM Layer-2 format, and the book does what it asks of you.

At this size, ninety days is not a sprint. It is a bet. Twelve chapters, six parts, and the honest path to $100M is about eight of those bets. Two years if things go well, three or four if they go normally. Anyone selling you a faster number is selling you something.

Read this before the diagnostic

Do not start this book because you crossed $10M. Start it because you pass the gate below.

Everything in Part IV and Part V is leverage: capital, category, big contracts, AI agents, cash reserve. Leverage multiplies whatever sits underneath it. If what sits underneath is your personal attention, you are about to build something that needs more of you at exactly the moment it can hold less of you. That failure is not slow and it is not cheap. It has eaten companies I know that looked fine the quarter before.

The independence diagnostic

Five gates, each scored 0 to 4. Twenty points maximum. Score from real records: your standards library, cohort reports, where leads came from, version history. Anything you score from memory, score one point lower.

First, the 30-day disappearance test. You are unreachable for thirty days. No phone, no exceptions, no marathon briefing session the week before. What breaks? Write the list down before you read Chapter 1. That list is the real table of contents for your next year.

Two of these need a note before you score them.

Gate 2 is a floor, not a scale. If net revenue retention is under 100%, score it 0, and no chapter in this book will fix it. Building a knowledge system inside a company that loses clients faster than it grows them just speeds up the loss. Go back to Playbook II, Chapter 5, and rebuild the review pipeline first.

Gate 4 is the one people grade too kindly. Level 3 means they work inside your written standards and decide without asking you. Level 4 means they handle situations nobody wrote down, and you hear about it afterwards without being alarmed. The test is not whether they could. It is whether they did, last quarter, and whether you can name the case.

Scoring

• 17–20 — Ready. Your problem is sequence, not readiness

  Your company already learns. Read straight through, pick your funding path with your eyes open, and treat Chapters 7 and 9 as this year's real work.

• 12–16 — The intended reader

  Some of what you know is now the company's, and some of it is still yours alone. Run Parts I and II in order. Do not touch Part IV until Chapters 1 and 2 have produced something you can hold. Leverage on a half-converted company is the exact mistake this band exists to prevent.

• Under 12 — Not yet, and I am not going to soften that

  Go back and finish Playbook II. Adding leverage to a founder-dependent machine makes it more fragile, not bigger. Fund money, enterprise deals and category plays all amplify whatever is underneath them. A year spent consolidating costs you time and nothing else. The alternative costs you the company.

Write your score and today's date somewhere you will find them again. Chapter 12 runs the same five gates against a harder target. The gap between the two readings is the only proof that any of this worked.

The fragility heat map

Run this next to the diagnostic. The score tells you what you can build. The map tells you what could take it away.

Draw your revenue as blocks: by client, by channel, by country, by currency, by key person. Then draw your delivery capacity the same way. Shade anything over 25% of a block red.

I learned this in 2022, badly. We had a company where 60% of delivery capacity sat inside one 40km radius. When the power grid went down, our backup plan turned out to be a shared belief that we had one. Two weeks of improvising, three lost contracts, one permanent lesson. A dependency you have not drawn is a dependency you have not seen.

The map takes forty minutes and it is the cheapest tool in this book. Chapter 11 tells you what to do about the red blocks. For now, just draw them.

Tool 0 — the independence diagnostic and fragility heat map: five gates scored 0 to 4, plus your revenue and capacity drawn as blocks with anything over 25% shaded. Blank version in the template pack.

The decision in this chapter: does your company's judgment live in people, or in systems that build people?

Where most $10M companies actually are

You have documentation. You are proud of it. It runs to 340 pages in Notion, it was written eighteen months ago by whoever had the calmest week, and about 30% of it describes a process that no longer exists.

That is not a learning organization. It is a snapshot with a search box.

The test is not whether standards exist. The test is whether they update themselves through use. In a company that learns, a project that goes wrong produces two things: a fixed client and a revised standard that makes the same failure harder next quarter. If your post-mortems produce Slack threads and feelings, your knowledge is evaporating about as fast as you create it.

That evaporation shows up on no report you read, which is how it runs for years. Nobody files a variance for judgment that walked out the door.

The four-step conversion

The cycle you need is old and well understood, and almost nobody runs it on purpose.

• From instinct to writing

  Someone's instinct becomes a written standard. This is the hardest step and the one people skip, because being good at something feels like nothing at all from the inside. Your best account director cannot tell you why she reprices a deal on the second call. She can only do it. Getting it out of her head means sitting beside her through four calls and asking "what did you just notice?" until the answer stops being "I don't know."

• From writing to combined

  Her standard meets the delivery lead's standard meets the finance model. This is where contradictions show up, and the contradictions are the point. If your sales standard promises a fourteen-day turnaround and your delivery standard budgets nineteen days, you have just found the source of a churn number you have been blaming on client expectations.

• From combined to instinct again

  People work through the standards until the standards become their own judgment. Reading does not do this. People remember what they produce and forget what they read. A standard handed over as a manual gets skimmed. The same standard handed over as ten worked examples plus "now do the eleventh and we will compare" gets learned in a fortnight.

• From their instinct to new knowledge

  Their judgment, working on cases you never described, produces instincts you do not have. That is where the loop closes and starts paying you back.

Most companies do the first two steps badly and then stop. You end up with a pile of documents and no judgment.

Level 4 is the target

Playbook II got you to Level 3. You set the standards, other people work inside them, you check the results. Good. That is where leverage starts.

Level 4 is different in kind. At Level 4, people write their own standards and handle cases nobody anticipated. You find out afterwards. You are sometimes surprised, and the surprise is usually a good one.

The gap between 3 and 4 is not skill. It is permission plus the why. A Level 3 person knows what good looks like. A Level 4 person knows why the standard exists, which means they can spot when the standard is wrong for the case in front of them and adjust without asking. Teaching the why is slower, more annoying, and the entire job.

In practice: for each of your top eight leaders, you should be able to name the area where they now write the standard and you review it, rather than the other way round. If you cannot name eight, name the three you have and put the other five on a development plan with dates on it. A plan without a date is a compliment.

Many projects at once: the advantage nobody can copy

Here is the part that becomes a real moat. It takes years, which is exactly why it holds.

When forty projects run the experience → understanding → impact loop at the same time, and what they learn flows into one standards system, you start seeing patterns nobody else in your market can see. Not because you are cleverer. Because you have forty observations of the same thing happening at once and your competitor has four, one after another.

A logistics software company I worked with, about $22M revenue, started tagging every implementation with the client's internal blocker at week six. Eleven months later they had 190 tagged implementations. They found that one specific way of buying predicted a fourfold rise in arguments about scope. They rewrote their qualification rules around it. Churn in that segment fell from 19% to 6% inside three quarters.

No competitor could copy that. The asset was not the insight. The asset was the 190 tagged observations and the habit that produced them. The insight itself was two sentences long and would have been worthless without four years of tagging underneath it.

What it costs

Be honest with yourself about the bill.

Doing this properly eats about 10–15% of your senior people's time for the first year, and shows almost nothing for the first two quarters. Boards hate that. Founders under revenue pressure quit at month five, right before it starts to pay.

If you cannot protect that time, do not start. A half-built knowledge system is worse than none, because people learn that the standards are theatre. That lesson is much harder to unteach than it was to teach.

Diagnostic

Count these from records, not impressions.

• Standards created or seriously revised in the last twelve months: ____
• How many of those were written by someone other than you: ____
• Areas where the judgment sits in exactly one person's head: ____
• Leaders who own an area where they write and you review: ____ of your top eight
• Date of the last standard revision caused by a project going wrong: ____

Then find yourself on this ladder. Standards exist and nobody has updated one in six months: you have documents, not a system. Standards get updated but only you write them: the ceiling is your own calendar. Others write and you review: Level 3.5, a good place to be. Others write, handle cases nobody anticipated, and tell you afterwards: Level 4, and Chapter 2 is your next move.

The 90-day move

Pick one area. Not five. Pick the one where you get pulled in most often for the least strategic reason, which is usually pricing exceptions or escalated delivery calls.

1. Days 1–20

   shadow the two people who do it best. Record the sessions. Ask "what did you notice?" after every decision, until the answer stops being "I don't know."

2. Days 21–45

   they draft the standard using the four layers from Playbook II. You do not write it. You mark it up, which is a different job and produces a different result.

3. Days 46–75

   build eight worked examples with the reasoning attached — three excellent, three adequate, two bad — taken from real files, not from imagination.

4. Days 76–90

   two people who were not involved run live cases against the standard. Every place they hesitate is a hole in the document. Fix the hole, not the person.

Completion test: one area has a standard written by someone else, tested by two people who did not write it, and fewer things in that area now get escalated to you.

Bias watchlist

• The curse of knowledge

  You cannot imagine not knowing what you know, so your standards skip the steps that matter most. Test every standard on somebody hired in the last six months. Their confusion tells you about the standard, never about them.

• Experts underestimating difficulty

  Your best people assume that what is easy for them is easy. That produces documents that are accurate and useless.

• Loving what you built

  You will overvalue the standards you wrote yourself and resist replacing them with better ones from your team. Watch for it in review meetings, where it disguises itself as having high standards.

• Doing too much at once

  Five areas feels serious and produces five half-finished standards, which teaches everyone that standards never get finished.

Annotated example: writing down a pricing instinct

Composite: a B2B services company, $24M revenue, where pricing exceptions reached the founder nine times a quarter.

• Unacceptable

  One line in the sales handbook: "Price against value, not hours." Everyone agrees with it. Nobody can act on it. Six months later discounting sits at 22% and nobody can say why. Why it fails: a principle got mistaken for a standard. It names the destination and gives no route, so every rep invents their own route, and the average of those routes is the discount rate.

• Adequate

  A four-page pricing standard with a decision tree: client size, cost of the problem, urgency, three price bands. Reps follow it and discounting drops to 14%. Then a client turns up with an odd structure — a public tender with a fixed ceiling and a big hidden implementation cost. The tree gives no answer, and the deal lands on the founder's desk. Why it is only adequate: the standard covers the ordinary case and leaves the profitable edge undefined, and the edge is where both the margin and the escalations live.

• Excellent

  The same decision tree, plus nine real deals written up, three of them losses. Each one is written by the account director who ran it: what she saw, what she assumed, what the client actually did. Plus one written rule for the edge: when the case does not fit the tree, price against what the buyer's alternative would cost them, and write up your reasoning within 24 hours. Those write-ups go into the library. Eleven months later the library holds forty cases, the founder sees two escalations a quarter instead of nine, and two account directors are rewriting the tree themselves. Why it works: the standard now produces more standard. Odd cases stopped being interruptions and became the raw material for the next version.

Tool 1 — the knowledge audit and Level 3→4 development plan: four columns per area showing where the judgment lives, plus one page per leader naming the area they own, the standard they will write, and the review date. Blank version in the template pack.