Most startups that fail did not have to.

Some failure belongs to innovation. A company tests what matters, finds out honestly that the market is not there, and stops. That kind should never be removed, because it is the price of trying something new.

The other kind comes from committing people, time and money to beliefs nobody tested while testing was still cheap. From the inside the two feel identical. This page, and the book, are about the second kind.

The companies I looked at after Tørn were not failing for many different reasons. They were failing for one reason underneath very different stories.

The reason is that a startup moves past a stage of its development before it has secured the proof that stage required. It hires, builds, spends, and raises, as if the proof were already in hand. Months or years later the missing proof surfaces, and by then the hires, the building, the spending, and the story told to investors have hardened into something too expensive to unwind. The death certificate says poor product-market fit, or premature scaling, or broken economics, or the wrong team. These are not separate reasons, but the same mistake, caught at different stages and given different names.

That is what happened at Tørn. We built supply before we understood demand. We committed to economics that worked on the first customers and broke when we tried to acquire more of them. We expanded into a second country before the first one worked the way we believed it did.

The same pattern runs through new ventures inside established companies. There the budget comes from the parent company, and the next stage-gate plays the part of the next funding round.

Around every startup sits a layer of people meant to help it navigate uncertainty: investors, boards, accelerators, advisors, state funders. You would expect some of them to catch what the founder cannot see from inside. Usually they do not, and it is not carelessness.

Almost everyone around the founder has borrowed the venture capital model, even when their own economics look nothing like it. That model works by backing many companies, expecting most to fail, and letting a few large outcomes pay for the rest. Tolerating spend on untested assumptions is not a flaw in it; it is how it finds the rare outlier. But almost no one else in the picture has a portfolio like a venture capital fund does. The founder has one company. The angel wrote a few cheques. The employee took equity instead of salary. The state fund and the family office use the venture filters without the portfolio diversification that makes those filters safe. The logic that is sound for the fund was never sound for the people standing inside any one of its companies.

And even when someone does see the weakness, the moment capital is committed the role shifts from scepticism to support. The company becomes part of the investor's own story, and questioning it openly gets complicated. So the founder cannot see it from inside, the people outside are not built to look, and the few who glimpse it have neither the licence nor the tools to act.

For two decades, cheap capital absorbed the cost of skipping this foundational work. A company that scaled on untested assumptions could keep raising while the assumptions either came good or the money kept flowing. That has ended. Follow-on rounds are harder, exits are smaller and rarer, and companies that used to survive being wrong for years now run out of money first. The cost of a missing proof, which the system used to absorb, now lands directly on the company.

The principle is old: test your assumptions before you commit money to them. What has been missing is the order. Which assumptions matter at which stage, and what counts as having tested one well enough to commit to the next.

A startup moves through stages, and each stage asks for one kind of proof before the company has the right to move on.

Proof of pull. Before building in earnest: that a specific kind of person wants the value enough to act on it, repeatedly, not just say they like it.

Proof of value deliverability. Before scaling delivery: that the value still arrives when the founder is not personally in the room holding it together.

Proof of business. Before pouring in capital: that the economics work on a small cohort, that a customer is worth more than it costs to win and serve them.

Proof of scale readiness. Before expanding into a new market or geography: that what worked here works there, tested on a small version first rather than assumed.

The proofs build on each other, and they are gathered while finding out is still cheap. Skipping one does not save time. It makes the next one impossible to gather honestly, and pushes the cost of being wrong into the period when it is most expensive to fix.

This is not slower. The founder who has done the work spends the next year building, not discovering at great cost what a few weeks could have told her.

The full method, how to gather each proof and what it looks like when it fails, is the book.