Queensland University of Technology (QUT) has a research project called CAUSEE where they have tracked thousands of Australian startups to find out how many got funded (a very tiny percentage of all startups), what they were planning to build, whether they succeeded or failed, how quickly they got going and so on.

Their justification was that too often we look at successful startups and then do whatever they did. Which isn't always bad advice, but there's the danger that it might be like looking at lottery winners to see what they "did" to make their win. How do we know that the stuff that those startups did actually made a difference?

For example, the CAUSEE study showed that (in Australia) retaining a lawyer is a good predictor that the startup will fail. Accessing any government service other than the R&D tax concession is a good predictor of failure. Writing a step-by-step business plan is a good predictor of failure.

But that's a bit Australian-specific. Is it the same in the USA? Is it the same in Europe? Is it the same in developing nations?

Should YC Research study scientifically what makes a startup successful in different economies? It would tie in nicely to YC's other work!

It would also help guide investment (and government policy) if we had hard data on what works and what doesn't.