1,310 Ways Startups Fail (and How to Avoid Them)
We analyzed over a thousand startup post-mortems to build the largest failure pattern database. Here's what we found.
We spent months building the largest startup failure pattern database we know of. Here's what 1,310+ post-mortems taught us.
The Data
We analyzed startup post-mortems from CB Insights, Failory, IndieHackers, Hacker News, and dozens of other sources. Each failure was coded into specific, actionable patterns — not vague categories like "ran out of money" but precise failure modes like "entered consolidating market with undifferentiated product."
The Top 10 Failure Patterns
1. No market need (23%) — Built something nobody wanted. The classic. Still the #1 killer. 2. Ran out of cash before PMF (18%) — Had a real problem, couldn't find the solution fast enough. 3. Wrong team composition (14%) — Technical founders with no distribution. Sales founders with no product. 4. Got outcompeted (11%) — Incumbent added the feature. Funded competitor moved faster. 5. Pricing/cost model broken (9%) — Unit economics never worked. Couldn't charge enough. 6. Timing was off (7%) — Right idea, wrong year. Market wasn't ready. 7. Regulatory/legal (5%) — Didn't check until it was too late. 8. Couldn't scale (4%) — Product worked for 10 users, broke at 1,000. 9. Founder conflict (3%) — Co-founder disagreements killed more startups than competition. 10. Pivoted too late (2%) — Knew it wasn't working, kept going anyway.
What Surprised Us
Pattern combinations are more lethal than individual patterns. A startup with "no clear differentiation" might survive. Add "entering consolidating market" and the kill rate jumps to 94%.
The same 30 patterns account for 80% of failures. Out of 1,310+ patterns, the top 30 are responsible for the vast majority of deaths. This is why our free tier — which checks against the most common patterns — catches most red flags.
B2B startups fail differently than B2C. B2B dies from long sales cycles and wrong buyer targeting. B2C dies from CAC and retention. Our Guardian knows the difference.
How We Use This
Every idea submitted to Greenbelt is checked against the full pattern database. The Guardian agent doesn't just know these patterns exist — it understands the conditions that trigger them and can identify when your specific idea matches.
This isn't a checklist. It's adversarial intelligence.