I haven't found a method with solid evidence that it predicts what software to build.

I haven't found a method with solid evidence that it predicts what software to build. So I'm testing mine against history.

Building is cheap now. Deciding what to build is the hard part, so I read what evidence exists:

• Outcome-Driven Innovation's often-cited 86% success rate is based on 21 client projects of the firm behind the method. • A re-examination of the 77 cases cited for disruptive innovation found only 9% fit all four elements of the theory. • In a meta-analysis of 233 studies, the strongest correlate of new-product success was product advantage. The samples are mostly manufacturing and B2B. • I couldn't find one verified case of "picked an idea from store data, then made money."

My own first test was humbling: of the 100 highest-scoring candidates from the Abandoned Users signal, an AI agent researching each one on the web judged 0 worth building.

The rule I now apply to anyone's hit rate, mine included: assume random picks succeed 5–20% of the time. A hit rate that isn't compared to that baseline tells you nothing.

So I'm backtesting. We have daily Chrome Web Store snapshots back to September 2023. Pretend it's a past date, run today's rules, check whether a newcomer reached 10,000+ users in those needs within 12 months, and compare against randomly picked needs. Signals that don't beat chance get published as such.

How do you decide what to build next?

Data: Chrome-Stats, processed by us