── ── Startup data
Why Startups Actually Die: Numbers from 26,724 Company Histories
July 11, 2026 · 7 min read
Across the 26,724 startup histories deciqAI has profiled, 'ran out of money' is the named cause of death in only ~8% of shutdowns. The dominant patterns: never finding traction (45%), competition (32%), and going in without a defensible moat (43% of failures vs 14% of winners).
Every founder has read a 'top reasons startups fail' listicle. Most are built on a few hundred self-reported post-mortems — written by the people with the least distance from the failure. We took a different route: deciqAI has profiled 26,724 company histories as part of a 1.3M+ entry knowledge base, annotating how each company started, what it built, and — where the record is clear — how it ended. This article is what the aggregate shows, with the dataset's limits stated up front.
The dataset — and its honest limits
Three caveats before any number. First, the sample skews toward companies visible to the venture ecosystem (YC, VC portfolios), not a census of all businesses. Second, annotation is LLM-assisted; every statistic quoted here was individually re-verified by a human against the underlying records. Third, these are descriptive correlations in our dataset — not causal claims. And one framing number: of the 26,724 startups we've profiled, only about 19% have a clear win and ~6% a confirmed shutdown. Most stories are still unresolved. Full methodology at deciqai.com/methodology.
'Ran out of money' is a symptom, not a disease
Of the profiled startups that shut down, 'ran out of money' was the named cause for under 1 in 10 (~8%). Far more simply never got traction (45%) or got crushed by competition (32%).
| Named cause of shutdown (our dataset) | Share |
|---|---|
| Never got traction | 45% |
| Crushed by competition | 32% |
| Ran out of money | ~8% |
Read that as a hierarchy of diseases and symptoms. Running out of money is almost always the terminal event, rarely the root condition. Fundraising buys time to find demand; it doesn't find demand for you. If the runway conversation is dominating your week while the traction conversation isn't happening, the data suggests you're managing the symptom.
Competition kills more than founders plan for
After 'simply went quiet,' the single most-named reason the startups we profiled died was competition — 32% — ahead of team, product-market fit, or money. 'Build it and they will come' ignores that they already have ten other options. Being better isn't enough if nobody notices you're different; the failure records are full of products that were marginally better and entirely unnoticed.
Moats show up on the winning side
The sharpest winner-versus-failure gap in our dataset is the moat question. Among the startups we've profiled, 43% of the ones that failed had no defensible moat — versus 14% of the ones that won. Break moats into types and the pattern holds: a proprietary-data moat showed up in 17% of the winners but only 6% of the failures (roughly 3x), and 25% of winners had a network-effect moat versus 15% of failures.
| Moat pattern (our dataset) | Winners | Failures |
|---|---|---|
| No defensible moat | 14% | 43% |
| Proprietary-data moat | 17% | 6% |
| Network-effect moat | 25% | 15% |
Features get copied; data compounds
A feature advantage survives until a competitor's next sprint. A data or network advantage gets stronger the longer you run. When you choose what to build next, weight the options by whether they compound — the winners in our dataset disproportionately picked the compounding thing.
Commercial DNA on the founding team
Founders with a business/commercial profile show up about 2x more often among the winners we profiled than the failures (20% vs 9%). Again: descriptive, not destiny. But it rhymes with the traction numbers above — technical brilliance with nobody owning go-to-market is the configuration the failure records keep describing. Someone on the founding team has to be obsessed with getting customers, not just shipping.
B2C is the harder road in this data
Among the failures we profiled, 41% were consumer (B2C) versus 28% of the winners; winners skewed B2B (49% vs 38%). Consumer means fighting for attention against everything else on the phone, and solo founders routinely underprice that distribution cost. If you're choosing between a B2B and a B2C version of the same idea, this dataset says the B2B road is more traveled by the companies that made it.
What to do with this on a Monday morning
Translate the failure data into decision bandwidth. The named killers — no traction, competition, no moat — are all questions of demand, differentiation, and compounding advantage. Those are the decisions that deserve the founder's personal hours. The pricing exceptions, support escalations, and sprint approvals eating your calendar are not on this list; we wrote a separate piece on the seven decision categories to systematize or delegate (see 'The Founder Bottleneck' below). And keep the survivorship caveat in view: with only ~19% clear wins in the dataset, the loud winners you're copying are the exception. The quiet middle — still unresolved, still deciding — is where most companies actually live.
FAQ
Why do most startups fail?
In the 26,724 company histories deciqAI has profiled, the most-named causes of shutdown are never finding traction (45%) and competition (32%). 'Ran out of money' is named in only ~8% of cases. These are descriptive patterns in our dataset, not causal claims.
Do startups really fail because they run out of money?
Rarely as the root cause. Among the profiled startups that shut down, running out of money was the named cause for under 1 in 10 (~8%). It's usually the terminal symptom of an upstream problem — most often no demand or losing to competition.
What moats correlate with startup success?
In our dataset, 43% of failures had no defensible moat versus 14% of winners. A proprietary-data moat appeared in 17% of winners vs 6% of failures (~3x), and network-effect moats in 25% of winners vs 15% of failures.
Is B2B or B2C more likely to succeed?
Among the startups we've profiled, failures skewed consumer (41% B2C vs 28% of winners) and winners skewed B2B (49% vs 38%). That's a distributional observation in our dataset — consumer distribution costs are the usual suspect, not an inherent law.
Where does this data come from?
From 26,724 company histories profiled in deciqAI's knowledge base (1.3M+ entries total). Annotation is LLM-assisted with every published statistic re-verified by a human; the sample skews toward venture-visible companies. Full methodology at deciqai.com/methodology.