पाठशाला Pathshala · विचार Vichār, The idea · Lesson 28 · Scale

The strategy memo: writing the thesis your company runs on

Write a two-page strategy memo a new VP could execute from: the diagnosis, the thesis, three bets, the non-bets, and kill criteria set before anything starts.

Pathshala, The Founder Library · 11 October 2026 · 7 min read

A compass and a set square resting on a sketchpad. — Photograph: Liezl Wilken · Pexels

Somewhere around fifty people the founder stops being able to carry the strategy in conversation. Decisions start to be made by people who were not in the room when the reasons were given. From that point the company runs on whatever is written down, and in most companies nothing is.

This lesson is about the document that fills the gap: two pages, written by the founder, that a new VP could execute from in their first month without asking what the founder meant. It sets out what the memo is for, its six parts, the two parts most memos leave out, and how to write and test it. The figure is a calculator for the part that decides whether the memo has teeth.

What the memo is for

Richard Rumelt gave the clearest account of what a strategy contains in McKinsey Quarterly in 2011. Its kernel has three parts. A diagnosis: an explanation of the nature of the challenge. A guiding policy: the overall approach chosen to overcome the obstacles the diagnosis found. Coherent actions: steps coordinated with one another to carry out the policy. He was as precise about bad strategy: fluff that hides the absence of analysis, failure to face the challenge, mistaking goals for strategy, and objectives that are a sprawling list of tasks.

Roger Martin made the companion point in Harvard Business Review in 2014: planning makes for more thorough budgets but must not be confused with strategy, which is about making bets. A budget with growth targets is a plan. It becomes a strategy only when it says what the company is betting on, what it is not, and why. The memo exists to say those three things in writing, so that they survive the founder’s absence from a meeting.

It helps to say what the memo is not. It is not the vision, which describes the world the company wants and changes rarely. It is not the operating plan, which turns the memo into quarterly targets, hiring and budgets. It is not the board deck, which reports progress. The memo sits between the vision and the plan and answers one question for a year: given what is true now, where will this company put its scarce people and money, and where will it not? A reader who finishes it should be able to predict how the founder would decide a question the memo never mentions.

The two pages, part by part

Diagnosis. A paragraph on what is true about the market and the company, with numbers. Not ambitions: facts the reader can check. Thesis. One sentence saying why the company wins, of the form: because of X that we have and others lack, customer Y will choose us for job Z over the next three years. Bets. Three at most, each with an owner, the people and rupees it gets, the metric it will move and the date it will be judged. Non-bets. What the company will not do this year, with the reason. Kill criteria. For each bet, the number below which it stops on its review date. Resource split. What share of people and money goes to the core and to each bet.

A surveying total station on a tripod at a construction site. — A survey fixes the lines before the first trench is dug. The memo fixes the bets before the first sprint is planned. Photograph: Khoiru Ummah Kendari 24 · Pexels

A worked example, invented for the purpose. A Bengaluru company sells inventory software to mid-sized distributors and has ₹18 crore of annual recurring revenue, growing at 45 per cent. Its diagnosis: three-quarters of new revenue comes from distributors in four states; net revenue retention is 112 per cent; deals with the largest distributors take seven months and win a third of the time. Its thesis: because it holds the daily stock data of 900 distributors, it can sell working-capital credit to them through lending partners better than any lender can underwrite alone. Its bets: a credit product with two lending partners; a self-serve tier for small distributors; a second-language interface for two more states. Its non-bets follow.

The resource split closes the memo and keeps it honest. In the example, 70 per cent of engineering and sales stays on the core inventory product, which still produces nearly all the revenue; 15 per cent goes to the credit product; 10 per cent to the self-serve tier; 5 per cent to the language work. In rupees, the credit bet gets ₹2.4 crore of the year’s ₹16 crore operating budget. A memo that lists three bets and gives each of them nothing in particular has made no bets at all, because every bet will be starved by the core the first time a large customer escalates.

Non-bets: the section that does the work

A memo without non-bets is a wish list, because every plausible idea fits somewhere under a thesis. Non-bets are what the company will decline this year even when a customer, an investor or a board member asks: in the example, no enterprise tier for the largest distributors, no expansion outside India, no lending from the company’s own balance sheet. Each gets a sentence of reason. The reason matters more than the item, because it lets a manager decide the cases the memo did not foresee.

Non-bets are also the cheapest test of whether leadership agrees. Ask each senior leader to write down, alone, the three things the company is not doing this year. If the lists do not match, the memo has not been written yet, whatever exists on the drive.

The most common failure is the thesis that is really a goal. Become the leading inventory platform for Indian distributors is a goal; it says where the company wants to be and nothing about why it will get there or what it will give up. Rumelt’s test applies: if the sentence does not name the obstacle and the approach to it, it is not yet a thesis. Rewrite it until a competitor reading it would learn something they did not want you to know.

Kill criteria: a number and a date, written before the start

Every bet will look promising at its review, because the people running it want it to and because early numbers are noisy. The defence is to decide in advance what failure looks like. The takeaways from Lenny Rachitsky’s conversation with Annie Duke, the former professional poker player who wrote Quit, put it in two lines: use pre-mortems to set kill criteria, and commit in advance to reassess when the red flags appear, because people wait for incontrovertible proof before they quit. A kill criterion has three parts: the metric, the date, and the number below which the bet stops and its people return to the core.

In the example, the credit product is disbursing ₹4 lakh a month through its lending partners at the start, with a target of ₹25 lakh a month in nine months and a kill line at 60 per cent of target. The figure is set to those numbers. At 15 per cent monthly growth the projection lands near ₹14 lakh, below the kill line of ₹15 lakh: the memo says the bet stops in month nine unless growth rises. The target needs about 23 per cent a month. That one number is what the bet’s owner should be held to every month.

A strategy memo without non-bets and kill criteria is a list of hopes. The two sections nobody wants to write are the ones that make it a strategy.

Writing it: prose, rewritten

Write the memo as prose. Slides let a founder place bullet points next to one another without saying how they connect, which is exactly where weak strategy hides. Jeff Bezos explained in his 2017 letter to shareholders that Amazon writes narratively structured six-page memos instead of presentations, that great memos are written and rewritten and shared with colleagues who are asked to improve them, and that a great memo probably should take a week or more. A founder’s first strategy memo will take longer. Two pages is the right length for a company under a few hundred people; anything longer is usually hiding a decision not yet made.

Hands typing on a vintage typewriter under the glow of a lamp. — Prose forces the connections that bullet points let a writer skip. The first draft of a strategy memo is never the one that ships. Photograph: Min An · Pexels

Send the draft to three people with one question: what would you do differently on Monday if this were true? If the answer is nothing, the memo describes the company rather than directing it. Rewrite until each reader names something they would start and something they would stop.

The new-VP test, and the memo’s calendar

The test of the finished memo is the newest senior hire. Give it to them in their first week and ask, a day later, for five things from memory: the thesis, the three bets with their owners, the non-bets, the kill line of one bet, and one thing they would now decline. If they get four, the memo works. If they cannot name the non-bets, rewrite that section first.

Then put the memo on a calendar. Write it in the month before the financial year, so the budget follows the strategy rather than the reverse. Read it aloud at the start of every quarterly review and mark each bet against its kill line. Rewrite it once a year with the [thesis review](/library/when-market-moves-under-you), which asks which assumptions in the diagnosis broke and what the company should stop doing. Keep every old version; the archive is the company’s record of what it believed and when.

The example company is invented for illustration. The figure’s rule for kill lines is this library’s suggestion, not a standard. Nothing here is investment advice.