I cataloged 127 failed startups. The #1 failure pattern surprised me — it wasn't PMF.
After 20+ years in engineering and watching startups rise and fall (including working through a few failures myself), I spent the last few weeks doing something I wish existed when I started: cataloging 127 failed startups and organizing them by WHY they actually died.
Everyone talks about product-market fit as the startup killer. But after going through $23.4B in burned capital, the most expensive failure pattern wasn't PMF.
It was Big Tech competition.
"The Crushed" category — startups killed when Google, Apple, or Amazon shipped their product for free — accounts for $11.7B across just 10 companies. Google+, Vine, BlackBerry, and others all had real products. Real customers. Real traction. Then a feature update made them irrelevant overnight.
Here are the 7 archetypes I found:
- The Hallucination — No real market need (Quibi, Juicero, Theranos) — $3.1B burned
- The Bonfire — Ran out of cash, usually by scaling pre-PMF (WeWork, Jawbone, Fab.com) — $6.2B
- The Crushed — Out-competed by Big Tech (Google+, Vine, BlackBerry) — $11.7B (largest)
- The Math Problem — Unit economics that never worked (MoviePass, Webvan, Pets.com) — $1.8B
- The Lemon — Product or tech that didn't work (Segway, Magic Leap, Nikola) — varied
- The Civil War — Founder/team conflict (Friendster, Clinkle) — not about the product at all
- The Outlaw — Legal or regulatory issues (Napster, Zenefits, Backpage)
A few things that stood out:
-
WeWork and MoviePass fail for completely different reasons. Grouping them as "ran out of money" misses the lesson. One is a real estate company with bad accounting; the other is a math problem that could never be fixed.
-
Quibi raised $1.75B for something nobody needed. The product worked. The marketing worked. The premise didn't. Classic Hallucination.
-
Theranos is fascinating — it crosses Hallucination AND Outlaw. The technology never worked, but the fraud puts it in a different category.
I organized all 127 into a structured database — each entry has the funding raised, what they did, why they failed, and the key lesson for founders. You can use it as a pre-mortem tool: run your startup idea against 127 historical failures before you build.
I put it on Gumroad if anyone wants the full dataset: bishtpradeep.gumroad.com/l/dead-startup-database
But honestly, even just naming the 7 archetypes has been useful for me. When I look at new startup ideas now, I ask: "Which of these 7 ways is this most likely to fail?" It changes how I evaluate things.
Curious what IH folks think: which failure pattern do you see most in early-stage startups you've been part of or watched?
the Crushed category being #1 is wild but makes total sense when u think about it. like Vine had EVRYTHING - users, culture, creators - and then instagram just copied stories and it was over
this is honestly why i think building in weird niches is the move for indie hackers. big tech doesnt care about your daily cryptic crossword puzzle app or your artcle-to-audio converter. theyre too busy fighting each other over AI chatbots and social media. the stuff thats too small or too weird for google to copy is exacty where indie builders thrive
the Hallucination one is scary tho. quibi is the perfect exmaple - the product was fine, the marketing was fine, the PREMISE was wrong. no amount of executon can fix a bad premise. thats the one that keeps me up at night lol
id add an 8th archetype tbh - "The Ghost" - startups that were actually good but just never figured out distrbution. nobody knew they existed. died in silence. thats probably the most common indie hacker failure and its not in the VC dataset because they never raised
great post, bookmarking the framework
"The Ghost" is a great callout — and you're right that it's underrepresented in my dataset precisely because the VC-backed failures dominate the historical record. There's no TechCrunch obituary for a bootstrapped product that died in silence.
Here's what I keep thinking about though: The Ghost often is a Hallucination in disguise. Most founders who say "nobody knew about us" built something for a market that wasn't actively searching. Real demand pulls products out of obscurity — people Google for solutions. When distribution completely fails, sometimes the signal is "people weren't looking for this" not just "people couldn't find us."
But for indie/bootstrapped? You're 100% right. Distribution failure as a standalone killer is probably the most common IH failure mode, and it won't show up in a VC dataset because there's no fundraising announcement to catalog.
Actually now I'm thinking about a v2 of the database that includes bootstrapped failures. The challenge is sourcing them — they leave almost no trace. If anyone here has experienced or knows of a clear Ghost failure, I'd be curious to hear about it.