Here's a question worth sitting with: if 90% of startups fail, why does almost all startup advice come from the 10% that succeeded?
The answer is obvious once you say it out loud. But the implications are stranger than they first appear — and they go well beyond "watch out for survivorship bias."
The Story Crowds Out the Statistics
When a founder tells you what made their company work, they're not lying. They're doing something more insidious: they're reconstructing a coherent narrative from a chaotic sequence of events, and then presenting that narrative as strategy.
A recent Forbes piece draws on Morgan Housel's The Psychology of Money to make this point sharply. The Bill Gates example is instructive: Gates is routinely described as brilliant and driven — which he was — but he also attended one of the very few high schools in the late 1960s that had access to a computer terminal. That circumstance gave him thousands of hours of practice when almost no one else had any. Most retellings skip that part. The story becomes about decisions and traits, not about the conditions that made those decisions possible.
This is the mechanism behind base rate neglect: vivid, specific stories override statistical frequencies. Kahneman and Tversky identified this formally in 1973 — people anchor on narrative descriptions and systematically ignore how often outcomes actually occur across the full population. The description of a diligent, analytical trader feels like signal. The base rate — that 93% of individual derivatives traders lost money over three years in one large-scale study — is the actual signal. The description is noise.
Startup advice has the same structure. The story of how Airbnb survived near-death or how Stripe found product-market fit is vivid and specific. The base rate — that most companies following similar paths failed — is invisible, because failed companies don't write retrospectives that get turned into frameworks.
What the Research Actually Shows
Here's where it gets interesting. MIT Sloan research on founder choices found that founders who incorporated in Delaware and acquired patent and trademark protection were dramatically more likely to achieve an equity growth event. That's a real finding with real predictive power.
But notice what it is: a base rate finding. It describes what's statistically associated with success across a large population of companies. It doesn't tell you why any individual company succeeded. And it certainly doesn't tell you that following those steps causes success — it might be that both behaviors are proxies for a third thing (founder sophistication, access to good legal counsel, serious intent) that's doing the actual work.
The advice that spreads virally is almost never this kind. It's the specific tactical story: "We did X and it worked." That advice is drawn from a sample of one, filtered through hindsight, and stripped of the base rate context that would tell you whether X actually matters.
The Trap Is the Specificity
I'd argue the more specific and actionable the advice sounds, the more suspicious you should be. "Incorporate in Delaware" is a base rate finding — boring, structural, probably worth doing. "We sent handwritten notes to our first 100 customers and that's what created our early community" is a story from a survivor. Maybe it worked. Maybe it worked for them, in their market, at that moment, with that particular customer base. Maybe a hundred other companies did the same thing and you've never heard of them.
A Medium piece reviewing 30 startup origin stories makes the point that the real histories — with pivots, near-deaths, and dumb decisions — get edited out of the versions that circulate. What remains is the clean arc. And the clean arc is precisely what makes the advice feel transferable when it isn't.
Try This
Next time you encounter startup advice that feels actionable, ask one question before you apply it: what's the base rate for this approach across all the companies that tried it, not just the ones I've heard of?
If you can't answer that — and usually you can't — treat the advice as a hypothesis worth testing in your specific context, not a principle worth following. The story is evidence. It's just much weaker evidence than it feels like.
The frameworks that actually generalize are usually the boring ones. The advice that feels like insight is usually just a survivor talking.