“I’m not ready to predict our success rate will stay as high as 50%. That first batch could have been an anomaly. But we should be able to do better than the oft-quoted (and probably made up) standard figure of 10%. I’d feel safe aiming at 25%.” – Paul Graham, March 2007
Seven years later, how have Y Combinator companies done?
YCList names 92 YC companies in the three years between winter 2006 and summer 2008. We can divide them into five categories:
Successful: The company has a large and growing userbase, big revenues, late-stage VC rounds, or other clear evidence of winning.
Product acquisition: The company was bought, and the acquirer keeps selling the product.
Stagnant: The website is still up, but there’s no evidence the company is profitable or has lots of users.
HR acquisition: The company was bought, but the product was shut down shortly afterwards.
Dead: The website is down; the service is no longer available.
Looking at the numbers, we find:
Successful: 12 (13%)
Product acquisition: 6 (6%)
Stagnant: 13 (14%)
HR acquisition: 14 (15%)
Dead: 47 (51%)
This gives us a success rate of about 20%. One notes that the outcome of Paul Graham’s original company Viaweb – a product acquisition – is actually by far the least likely scenario.
The Rationalist Conspiracy 
Where does the 20% number come from?
I see Successful (13%) + Product acquisition (6%) + HR acquisition (15%) = 34%
HR acquisitions are failures; they’re what companies do when the founders have no other choice, and investors typically lose money on them. The engineers usually get cushy big-corp programming jobs, but they can have those whenever they want anyway.