पाठशाला Pathshala · विचार Vichār, The idea · Lesson 18 · Build
Secrets: what you know that the market does not
Every company worth building rests on a belief most informed people think is wrong. Write yours as one sentence, then gather the evidence that it is true and the evidence that it is unpopular. You need both.
Pathshala, The Founder Library · 11 October 2026 · 8 min read
Peter Thiel’s favourite interview question asks what important truth very few people agree with you on. Most founders answer it with a trend that everyone already believes: AI will change everything, Bharat is the next billion. A trend everyone believes is not a secret. It is the reason the market is crowded.
This lesson gives the idea of a secret a working form. It shows how to state the belief your company rests on in one sentence, how to gather evidence that it is true, how to gather evidence that it is unpopular and how to keep checking both as the company grows.
What a secret is, and what it is not
Thiel’s Zero to One, written with Blake Masters and published in September 2014, grew out of a course he taught at Stanford. Its argument is that great companies are built on secrets: things that are true, important and not yet widely known. He divides them into secrets about nature, which science uncovers, and secrets about people, which concern what people do not know about themselves or will not say. Most business secrets are of the second kind. They are truths about how customers actually behave, what an industry quietly accepts, or what an incumbent cannot do without hurting itself.
A secret is not a fact no one has access to. It is usually hiding in plain sight, available to anyone who looks closely at an unglamorous place. It is not an opinion either. An opinion that cannot be checked is a guess. A secret is a claim that can be tested, which most informed people would bet against today.
Nor is a secret the same as an idea. The idea is what you will build. The secret is the belief that makes the idea work when others think it will not. Two founders can have the same idea and different secrets, and the one with the truer secret usually wins.
Right and lonely: the two-by-two that explains why
The investor Howard Marks drew the reason in his memo Dare to Be Great II of April 2014. Put conventional and unconventional behaviour on one axis and favourable and unfavourable outcomes on the other. Conventional behaviour with good outcomes produces average good results. Conventional behaviour with bad outcomes produces average bad results. Only unconventional behaviour can produce above-average results, and it can also produce below-average ones. Non-consensus ideas, Marks wrote, have to be lonely, and the investor who holds them needs the ability to look wrong for a while.
Translated to a startup: a consensus belief that is right attracts every competitor and every investor, and the returns are competed away. A non-consensus belief that is wrong is how most contrarian companies die. The only box that produces a company worth a decade of your life is the non-consensus belief that turns out to be right. That is why both halves of the evidence matter. A true belief everyone shares is a crowded market. An unpopular belief that is false is a mistake.
Where founders in India find them
Paul Graham’s advice in How to Get Startup Ideas of November 2012 is to live in the future, then build what is missing. His companion essay, Schlep Blindness of January 2012, explains why the missing things stay missing. For over a decade, he wrote, every programmer who had processed payments online knew how painful it was, and nobody fixed it, because people’s dislike of tedious work is largely unconscious. Your unconscious will not even let you see ideas that involve painful schleps. Stripe’s founders saw it.
In India the schlep is often where the secret lives. Four places repay attention.
Inside an unglamorous trade. Founders who have worked in a textile market, a transport yard, a hospital billing desk or a district court know how the work is really done, which is rarely how the industry report describes it. The [founder-market fit lesson](/library/founder-market-fit-unfair-advantage-test) covers how to tell whether that knowledge is an advantage or a nostalgia.
In the rulebook. A new RBI direction, a GST notification or a change to a scheme’s eligibility can make something possible that the market still believes is impossible. Few people read the circulars. Fewer read them and ask what business they create.
In operations everyone avoids. Cash collection, vernacular customer support, field verification, last-mile returns. Investors often treat these as reasons not to back a company. A founder who has found a way to do one of them cheaply has a secret that looks, from the outside, like a cost.
With customers the incumbent has priced out. If the leading company in a category serves the top of the market well and the rest badly, the belief that the rest will pay something for a different product is usually unpopular and sometimes true.
Writing the secret as one sentence
Write it in three clauses. Most people who know this market believe A. We believe B. We know because of C. The first clause names the consensus, and it should be a view a well-informed person in the industry would actually hold, not a straw man. The second clause is the secret. The third is the evidence, and it should name something you saw or measured rather than something you feel.
Then add a fourth line: if we are right, D follows. D is the business. If the secret is true, what product does it make possible, for whom and why can the consensus not build it quickly? A secret with no business attached is an interesting observation. A business with no secret underneath is a bet on execution against people who share every one of your beliefs.
A worked example: the traders in Surat
A team spends a year in Surat’s textile markets before writing any code. Their sentence: most investors believe small textile traders will not pay for software. We believe they will pay a monthly fee for software that chases the money buyers owe them, because unpaid credit is the worry that keeps them up at night and bookkeeping is not. We know because of what we saw in thirty shops.
The evidence is specific. Of thirty traders interviewed, nineteen kept a separate notebook of dues apart from their accounts. Twelve employed someone, a relative or a part-time clerk, whose main job was to call buyers who had not paid. Six paid a refundable ₹1,000 deposit to join a pilot. The consensus belief is about software in general, which traders associate with accounting and the tax consultant. The secret is about one job, collection, that traders already pay a person to do.
If they are right, a collection product sold at less than a clerk’s monthly salary, with reminders in Gujarati and Hindi and a UPI link in every message, can win shops that every accounting product has failed to convert. The figures are illustrative, but the shape is the one to copy: a consensus stated fairly, a narrower belief against it, and evidence that is behaviour rather than opinion.
A true belief everyone shares is a crowded market. An unpopular belief that is false is a mistake. You need a belief that is both true and lonely.
Proving it twice: true, and unpopular
First, that it is true. Evidence comes in rungs, and the higher rungs count for more. A story is the lowest: one customer said something striking. A pattern across twenty conversations is better, especially if you heard it without asking a leading question. Behaviour is better still: people paid, switched, kept a notebook, employed a person, or changed how they work to get around the problem. A number from a public source that the consensus has not read closely is strong. A test you ran, with a price and a result, is strongest. A secret should reach the behaviour rung before you build a company on it, and the test rung before you raise money on it.
Be suspicious of evidence that only you could have gathered and only you have interpreted. Ask someone outside the team to read the interview notes cold and say what they think the customers meant.
Then, that it is unpopular. This half is skipped more often, and it matters as much. Four tests.
The five-person test. Tell the sentence to five informed people: an investor in the sector, an operator from an incumbent, an industry consultant, two founders in adjacent markets. If four of them agree at once, it is consensus. If most disagree and cannot quite say why your evidence is wrong, you may have a secret.
The funding test. Search funding announcements and the startup databases for companies built on the same belief. If several are funded, the belief is consensus among investors at least, and the question becomes why you will beat them.
The incumbent test. Ask why the leading company in the category has not acted on the belief. A good answer is structural: acting on it would hurt their margins, cannibalise a product or offend their best customers. A weak answer is that they have not thought of it.
The timing test. If the belief is true and unpopular, something probably changed recently that the consensus has not absorbed. Name it, with a date and a number, as the [why now lesson](/library/why-now-timing-argument) describes. A secret with no why-now is often an old idea that failed for reasons you have not yet found.
Spending a secret, and a quarterly review
Secrets decay. Every customer you win is evidence the market can see, and competitors will eventually read it. The head start a secret gives you is worth only what you build during it: the customer relationships, the data, the distribution or the brand that remain when the belief has become consensus. The [moats lesson](/library/moats-that-matter-in-year-one) covers which of those you can begin in year one.
Tell investors the secret. They cannot back a company whose central belief they do not know, and the good ones have heard thousands and will not steal it. Tell the team, because a team that knows the secret makes better small decisions. Be more careful with partners who could act on it faster than you can.
The secret file. Keep one page. At the top, the sentence in its three clauses and the line on what follows. Beneath it, two columns: evidence it is true and evidence it is unpopular, each item dated and placed on its rung. Every quarter, add what you learnt, strike out evidence that no longer holds, and answer two questions in writing. Is the belief still true? Is it still unpopular? When the second answer becomes no, the secret has become the consensus, and the company needs either the moat it built during the head start or a new secret.
The worked example is illustrative. Sources were checked in October 2026. Nothing here is investment advice.