Most founders learn from the startups that made it.
A Second World War lesson about bomber armour shows why the startups that failed often hold the more useful lessons, and how to study them in practice.
Every founder has a shelf of success stories. The Airbnb founders selling cereal boxes, the Monzo waiting list, the university dropout who built a unicorn from a bedroom.
We read them, we copy them, and we quietly assume that following the same steps will bring the same result.
There is a problem with that. For every startup that made it onto a podcast, many more tried similar things and shut down without anyone noticing.
Their stories rarely get told, so their lessons never reach you. You end up learning from a sample already filtered for luck, timing and survival.
This trap has a name: survivorship bias. One of the clearest examples comes from a group of statisticians trying to keep bomber crews alive during the Second World War.
The planes that came back
By 1943, Allied bombers were being shot down over Europe in heavy numbers. Commanders wanted more armour on the planes, but armour is heavy.
Add too much and a bomber flies slower, burns more fuel and carries less. So the question was simple: where should the armour go?
The obvious answer came from the planes that made it home. Their bullet holes were mapped, and the damage clustered on the wings, the tail and the main body.
The natural conclusion was to reinforce those areas.
Abraham Wald, a Hungarian-born mathematician at the Statistical Research Group at Columbia University, spotted the flaw. Every plane in the data had survived.
If returning bombers rarely had holes around the engines, it was not because engines were rarely hit. It was because planes hit in the engines mostly did not come back.
The holes on the survivors showed where a plane could take damage and still fly home. The clean areas showed where it could not.
Wald's advice was to protect the places where the returning planes had no damage at all. His wartime memorandum, A Method of Estimating Plane Vulnerability Based on Damage of Survivors, turned that idea into hard maths.
The lesson is simple. The most important data is often the data that never made it back.
How survivorship bias shows up in startups
Startup culture runs on survivor stories. Podcasts interview founders who exited, LinkedIn rewards posts about funding rounds, and business books study companies that are still standing.
So founders absorb a set of rules that look proven: drop out, move fast, raise big, ignore the doubters.
The catch is that thousands of founders did exactly those things and failed. You never hear from them, so every rule looks safer than it really is.
This shows up in three common ways:
Confusing traits with causes. Many successful founders are stubborn. So are many who ran out of money because they refused to change course. A trait found on both sides is not the reason one side won.
Copying the tactic and ignoring the context. A move that worked for one company, in one market, at one moment, gets repeated as a law. The many companies that tried it and got nothing stay invisible.
Misjudging the odds. If you only ever see winners, the game looks easier than it is. You take bets you cannot afford to lose, because the people who lost them are not around to warn you.
The better question for founders is simple: which startups did not make it, and what actually went wrong?
How many startups really fail
The startups that fail are not a small group. They are the majority.
According to the Office for National Statistics, only 38.4% of UK businesses born in 2019 were still active five years later. Most made it through year one, then the losses built steadily.
These figures cover all VAT or PAYE-registered businesses, not just venture-backed startups. The shape is the same though: the survivors are outnumbered.
So what brings them down? CB Insights analysed 111 founder post-mortems. The two most-cited causes were running out of cash or failing to raise (38%) and no market need (35%).
Its founder, Anand Sanwal, made a sharp point about the first one. Running out of cash is usually a symptom, not the cause.
Something else went wrong first, with the product, the market or the team. That is the real cause of failure.
It rarely looks dramatic from the outside. It is a slow problem nobody prepared for, because the success stories never mentioned it.
No market need is also the cheapest failure to test for, by validating your startup idea before you build.
How to learn from startups that failed
Studying failure does not mean being negative. It means looking at the whole sample before you decide where to focus.
Keep a failure file for your industry. For every success story you save, save a failure from the same space. Administrators' reports filed at Companies House often explain in plain language what went wrong.
Talk to founders who shut down. Ask what they believed at the start that proved wrong, when they first saw the problem, and what they would protect first if they started again.
Run a pre-mortem. Before a launch, hire or raise, imagine the startup has failed two years from now. Write down the three most likely reasons, including team risks like the wrong co-founder fit.
Study the failures in your own data. Churned customers, lost deals, inactive trial users and investors who said no can teach you what happy customers never will. Call five of them this month.
Compare, do not just copy. Ask what the winners did that the losers did not. A habit shared by both groups tells you nothing.
Treat every success rule as a hypothesis. When you hear "charge more" or "raise early", ask how many people followed it and failed. If nobody knows, test it small before it shapes your startup pitch deck.
The mindful side: you are probably doing better than you think
Survivorship bias does not only distort strategy. It also distorts how founders feel about themselves.
When your feed is full of overnight successes, your own progress looks painfully slow. You end up comparing your messy middle with someone else's highlight reel.
The founders who struggled through the same eighteen months and then quietly closed are not posting. The comparison is rigged from the start.
A few things worth holding on to:
Slow is normal. Many companies that look like overnight successes had years of unseen work behind them. You are only seeing the part that made it into the story.
Success and failure are events, not identities. Cazoo founder Alex Chesterman had already built LoveFilm and Zoopla. The same founder can have big wins and a painful collapse.
Your nagging worries are data. If one risk keeps coming back to you at 2am, it might be your real weak spot. Do not dismiss it just because someone said "ignore the doubters".
Write down why you are doing this. On a hard week, your own reasons will steady you far better than someone else's success story.
The bottom line
Success stories are still worth reading. They just need to be read alongside the failures, not instead of them.
Keep learning from the winners, but give the failures equal time, because they show you where the real damage happens.
This week, pick one company in your space that shut down and find out what went wrong. Then check whether your own startup is exposed to the same problem.
