What 50 dead startups taught me about advice

By Juan · September 10, 2025 #survivorship-bias#patterns#data

I had a free weekend in March and I read 50 startup postmortems in a row.

Not the famous ones. Just whatever I could find — CB Insights’ lists, Failory, the Indie Hackers “shutdown” tag, blog posts founders wrote on their way out the door. Companies that had raised between $0 and $30M, mostly the lower end. Companies you’ve never heard of, by definition.

I went in expecting to read a lot of variations on “we ran out of runway” and “we couldn’t find product-market fit.” I got something quite different.

What actually killed them

Out of 50 postmortems, here’s what the founders themselves named as the primary cause:

15 — co-founder conflict, divorce-style breakdowns. By far the largest single category.
9 — burnout. The founder ran out of personal energy before the company ran out of money.
7 — built a product the founders wanted but no users did. Classic.
6 — capital structure / cap table problems. Bad early deals that strangled later ones.
4 — competition (a bigger player launched the same thing).
3 — tried to scale too early; the unit economics never worked.
3 — the market itself disappeared (regulation, Apple deprecation, etc.).
2 — fraud or trust collapse with customers.
1 — pure bad luck (one company’s only customer went bankrupt).

The honest postmortems almost always named more than one cause, but I tried to pick the one the founder said was decisive.

What’s missing from this list

Notice what’s not on it:

“We didn’t move fast enough.”
“We didn’t iterate hard enough.”
“We didn’t validate demand.”
“We didn’t have the right metrics.”
“We didn’t build a moat.”

These are the failure modes that successful founders write essays about. They’re the failure modes that get talked about at conferences. They are mostly not the ones actual companies died from.

The actual top cause was two humans who couldn’t keep working together. The second was one human who couldn’t keep going. The third was the classic build-the-wrong-thing.

The advice I’d been absorbing for a decade was almost entirely about the third bucket. It was about as useful for the first two as a dietitian’s advice is for a marriage.

Why successful founders write the advice they write

When a founder succeeds and writes about the lessons, they’re writing from the survivor side of the survivorship-bias filter (more here). They survived, so by definition they didn’t die from any of the causes above. They might have come close to product-market fit problems, so they have content there. They might have had near-burnout episodes they overcame, so they have content there.

But the founders who lost their co-founder mid-Series A don’t have a startup to write about anymore. The ones who burned out are mostly in therapy, not on stage. The ones with cap table catastrophes are usually under NDAs. The ones whose market disappeared are working on different things and don’t want to revisit.

So the advice you read is selected for the topics survivors can talk about. The topics that actually killed most companies are mostly missing.

What this changes

If you’re starting a company, the rational thing isn’t to optimize for “ship faster” or “have a better moat” — those are necessary but they’re not what kills you. The rational thing is to spend disproportionate time on the things actually killing companies that aren’t in the official advice corpus.

That looks like:

Co-founder agreements that include divorce clauses you discuss before you need them. Talk about money, control, equity vesting on bad scenarios, what happens if one of you wants to quit.
A real plan for personal sustainability. Not “we’ll grind for two years” — actually figure out what your week looks like at month nine when the novelty is gone.
A cap table you understand by heart, with someone independent who’s not your investor checking it.
A clear-eyed read on whether the platform/regulator your company depends on could deprecate you.

This is much less fun than “ship faster.” It’s also much more closely correlated with not dying.

What this changes about reading

I read fewer founder bios now. I don’t think they’re useless — there’s craft in there, and some of the technical stuff transfers. But I treat them the way I treat lottery winner interviews: interesting, not instructive.

Instead, I read more postmortems. Not for the lessons — most postmortem lessons are wrong or obvious — but to calibrate my model of what actually goes wrong. The shape of failure is the data. The “what I learned” sentences are commentary you can mostly skip.

If you’ve been reading mostly success stories, try a weekend of reading mostly failure ones. The map of the territory will look very different.

— Juan