पाठशाला Pathshala · विचार Vichār, The idea · Lesson 29 · Scale

When the market moves under you: rereading the thesis

Run an annual thesis review that scores last year’s assumptions against this year’s numbers, names the ones that broke, and turns each into something the company stops doing.

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

An aerial black and white view of a river winding in tight meanders through grassland. — Photograph: Yulia Opanasyuk · Pexels

Markets rarely announce that they have moved. Customers buy a little less often, a deal takes a month longer to close, a rule that a business depended on is amended in a bill most people did not read. By the time the change is obvious the company has spent a year executing a strategy written for a market that no longer exists.

This lesson is the annual review that catches it. It treats the thesis as a set of assumptions with numbers attached, shows two ways markets move and one Indian case where the move took two days, gives the question that unlocks the hardest decision, and ends with the part most reviews skip: a list of what the company will stop doing.

A thesis is a list of assumptions

Rita McGrath and Ian MacMillan’s discovery-driven planning, first set out in Harvard Business Review in 1995, starts from the observation that a new venture’s plan rests on assumptions rather than knowledge. Their method, as McGrath’s firm describes it, builds an assumptions checklist, the list of things that need to prove true for the project to succeed, ranked so that the deal killers and the assumptions that can be tested cheaply come first. They pair it with a reverse income statement: start from the profit required and work back to what would have to be true to earn it.

A company at scale needs the same discipline turned backwards. Last year’s [strategy memo](/library/strategy-memo-thesis-your-company-runs-on) contains a diagnosis, and the diagnosis contains numbers: how many buyers need the product, what they will pay, what it costs to win one, how long a sale takes, which rule or platform the business rests on. Those five are the load-bearing assumptions in most companies. The review checks each one against this year’s actual, and nothing else, before anyone is allowed to talk about plans.

How markets move: slowly, then by statute

Most moves are slow. A customer segment matures and buys less often; a channel saturates and the cost of acquiring a customer climbs a little each quarter; a competitor drops its price and the company’s own price holds while its win rate falls. None of these produces a single bad month. Each shows up only when this year’s number is set beside the one the thesis assumed.

The silhouette of a rooster weather vane against a cloudy sky. — A vane turns with every change of wind. A thesis does not turn unless someone reads it against this year’s numbers. Photograph: Mat Brown · Pexels

An invented example shows the shape. A Hyderabad company selling attendance software to schools assumed last year that it would win a school for ₹40,000 of sales and marketing, at ₹1.2 lakh a year, in a sales cycle of three months. This year the actual cost of winning a school is ₹64,000, the price has slipped to ₹84,000 because two free products arrived, and the sales cycle is unchanged. Month by month nothing looked alarming. Side by side, two of the company’s five load-bearing numbers have moved by 30 and 60 per cent, and its payback has gone from four months of revenue to more than nine.

Some moves are sudden, and India supplies a recent and complete example. The Promotion and Regulation of Online Gaming Bill, 2025 was introduced and passed in the Lok Sabha on 20 August 2025 and passed in the Rajya Sabha the next day. It prohibits offering online money games, whether they depend on skill, chance or both, along with advertising them and facilitating payments for them; offering such games carries up to three years in prison, a fine of up to ₹1 crore, or both. A company whose thesis rested on real-money gaming had its central assumption removed in two sitting days of Parliament. The lesson on [regulatory arbitrage](/library/regulatory-arbitrage-ideas-expiry-date) covers how to see such dates coming; this one covers what to do on the morning after.

The question Intel asked in 1985

The hardest part of a thesis review is not finding the broken assumption. It is admitting what it means for the business the leadership built. Andy Grove and Gordon Moore faced it at Intel in 1985, at the start of what NPR later called the company’s radical turnaround. As NPR reported, Grove asked Moore what would happen if somebody took them over and got rid of them: what would the new guy do? Moore answered: get out of the memory business. Grove agreed, and suggested that they be the ones to do it. Under the new plan Intel laid off more than 7,000 people, almost a third of its workforce. The question works because it lets the people who built a business answer as outsiders would.

Peter Drucker had a version of the same question for every organisation. In The Theory of the Business he wrote that every three years an organisation should challenge every product, every service, every policy and every distribution channel with the question: if we were not in it already, would we be going into it now? He was blunt about the alternative, as quoted by the American Management Association: without systematic and purposeful abandonment, an organisation will be overtaken by events.

Running the review

Give it one day, once a year, a month before the strategy memo is rewritten. The pre-read is two pages: last year’s memo, and a table of the five load-bearing assumptions with the number assumed and this year’s actual beside it. The finance lead prepares the table, not the founder, so that the numbers arrive before the interpretation does.

Keep the room small: the founders, the leaders of each business line and finance, and one person from outside the leadership who will ask the obvious question, a board member or an adviser who knows the market. The outsider’s job is Grove’s question in human form. Agree at the start that the day ends with two documents and only two: the scored table and the stop-doing list. Plans come later.

Score each assumption as actual over assumed. Within 25 per cent either way it holds. Between 25 and 50 per cent it is strained and the section of the memo that rests on it is rewritten. Beyond 50 per cent it is broken, and that includes the lucky direction: a company that sold twice the volume it planned at a price it did not expect has a thesis it does not understand, and will not know why when the luck stops. The figure is set to a typical year: demand slightly soft, price 30 per cent below plan, the cost of winning a customer 60 per cent above it.

If we were not in this already, would we go into it now? Ask it every year, and act on the answer.

With one assumption broken and one strained, the verdict is to rewrite the sections that rest on them, not the whole thesis. Push the rule or platform slider to zero, as real-money gaming companies saw in August 2025, and a second broken assumption makes the verdict a full rewrite. That is the point at which Grove’s question stops being a thought experiment.

The stop-doing list

A review that ends with a new plan and no deletions has failed. Every broken assumption should produce at least one item on a stop-doing list: a product line wound down, a segment no longer sold to, a channel switched off, a hiring plan cancelled. Each item carries a date, the owner who will carry it out, and the people and rupees it frees, written as a number. The freed resources are then allocated in the new memo, in the open, so that the team sees the stop and the start as one decision.

Stopping is harder than starting because every item on the list has a constituency inside the company. That is why the list is written in the review, by the people in the room, against the numbers in the table, rather than left to the planning cycle where each item will be defended one at a time.

In the school software example the stop-doing list writes itself once the table is on the wall. The free products have taken the small private schools, so the company stops selling to schools under 500 students, ends the field sales team in two cities where those schools were its only market, and moves the six people it frees to the larger school groups where price has held. The freed cost is about ₹70 lakh a year, written as a number, and the new memo says where it goes.

An annual calendar for the thesis

Hold the review in the eleventh month of the financial year. In the week before, the finance lead fills in the five assumptions. On the day, score them, ask Grove’s question of each business line, and write the stop-doing list. In the following fortnight rewrite the strategy memo, carrying forward what held, rewriting what was strained, and replacing what broke. Between reviews, watch the rule and platform assumption monthly, because that is the one that can break in two days. Keep each year’s scored table; three years of them show whether the company’s judgement about its own market is getting better.

Law and figures were checked in October 2026 against the sources below. The scoring bands are this library’s judgement. Nothing here is legal or investment advice.