Most people say that 90% of start-ups fail, but they don’t mention how many start-ups entrepreneurs attempt on average. If:
0.most founders only attempt one startup (and first-time startups have a 90% chance of failing),
But, founders who found multiple startups have a better chance of success.
Then, the inside view that (you should be able be able to succeed at a startup if you do a lot of them in your 20s) and the outside view of (90% of startups fail) can actually be compatible.
You could make your model a bit more precise by noting:
Chance of all your startups failing (during life-time) = P(1st startup is a failure) P(2nd-startup is a failure) ETC.
If:
each start-up is independent of each other. (A pretty big assumption. I would expect people to get better over time. You can gauge how much.)
Each startup takes 2 years. (You can obviously change this number around).
Then, if somebody just did 6 startups consecutively, their probability of success would be 1-.9^6 = .47. There’s certainly room for a healthy amount of optimism in this model. If you model it more like 1-(.9.7.6*.6) = .73, then those seem like pretty good odds to do lots of startups.
I’d like to see people play around with numbers like these. It’s an outside view model. You can use your inside view to predict more specific things (how long a start-up takes, how much you learn from your failures, etc).
This is better than having a model:
P(Devise an idea for a product that creates demand.) = .90
P(Build it) = .90
p(Market and sell it) = .90
P(Things run smoothly (some might call this luck) ) = ?
P(success) = .9.9.9*? = higher than what’s actually true. (Though, you could contend that plenty of people can’t devise an idea for products that creates demand, and that if you can do this, then you have a much better chance than the average start-up).
Paul Graham said something like “Startup founders are / (have to be) optimists”. I’m wondering how accurate people think P(2nd-failure | 1st failure) = .7 is. It seems like it’s still pretty skeptical (closer to .9 chance of failure rather than .1 chance of failure of the purely inside view model), but is still probably optimistic compared to YC’s ~80% failure rate.
Your post is exactly why “how many startups can I conceivably do” is an important question. If failed startups take on average 5 years to fail, which is a reasonable assumption for a semi-successful but ultimately failed startup, then doing 4 startups takes 20 years of your life. For most people, working 20 years at a startup and making relatively low wages is not feasible or desirable.
Most people say that 90% of start-ups fail, but they don’t mention how many start-ups entrepreneurs attempt on average. If: 0.most founders only attempt one startup (and first-time startups have a 90% chance of failing),
But, founders who found multiple startups have a better chance of success. Then, the inside view that (you should be able be able to succeed at a startup if you do a lot of them in your 20s) and the outside view of (90% of startups fail) can actually be compatible.
You could make your model a bit more precise by noting:
Chance of all your startups failing (during life-time) = P(1st startup is a failure) P(2nd-startup is a failure) ETC.
If:
each start-up is independent of each other. (A pretty big assumption. I would expect people to get better over time. You can gauge how much.)
Each startup takes 2 years. (You can obviously change this number around).
Then, if somebody just did 6 startups consecutively, their probability of success would be 1-.9^6 = .47. There’s certainly room for a healthy amount of optimism in this model. If you model it more like 1-(.9.7.6*.6) = .73, then those seem like pretty good odds to do lots of startups.
I’d like to see people play around with numbers like these. It’s an outside view model. You can use your inside view to predict more specific things (how long a start-up takes, how much you learn from your failures, etc).
This is better than having a model: P(Devise an idea for a product that creates demand.) = .90 P(Build it) = .90 p(Market and sell it) = .90 P(Things run smoothly (some might call this luck) ) = ?
P(success) = .9.9.9*? = higher than what’s actually true. (Though, you could contend that plenty of people can’t devise an idea for products that creates demand, and that if you can do this, then you have a much better chance than the average start-up).
Paul Graham said something like “Startup founders are / (have to be) optimists”. I’m wondering how accurate people think P(2nd-failure | 1st failure) = .7 is. It seems like it’s still pretty skeptical (closer to .9 chance of failure rather than .1 chance of failure of the purely inside view model), but is still probably optimistic compared to YC’s ~80% failure rate.
Serial entrepreneurs are no more likely to succeed than first time entrepreneurs.
Any thoughts about what to conclude from this? It might imply that success in business is mostly luck.
Your post is exactly why “how many startups can I conceivably do” is an important question. If failed startups take on average 5 years to fail, which is a reasonable assumption for a semi-successful but ultimately failed startup, then doing 4 startups takes 20 years of your life. For most people, working 20 years at a startup and making relatively low wages is not feasible or desirable.