Problem 1 of 4
Value
More features. The same customer problem.
Shipping more features does not guarantee a better customer experience. When building gets cheaper, choosing the right problem matters more.
01 / The effectiveness gap
More output. The same unanswered question.
AI can lower the cost of building features. Customer value still depends on choosing and solving a worthwhile problem. Fewer than a third of ideas tested at Microsoft improved their target metric, and roughly a third made the product worse.
The organization that shipped the wrong things slowly can now ship them at scale. As more ideas become feasible, deciding which ones deserve investment matters more.
Reported feature effectiveness across three companies
02 / How the wrong solution gets locked in
The business case locks in the answer.
Most features start life in a business case. There is usually research behind it, but with software nobody can be sure the solution works until customers use it.
Say research shows customers find setup too hard. Checklists are an established pattern for this, so the business case funds one. It ships on time and setup completion barely moves. Customers were stuck at account verification, and the checklist leaves that step just as hard.
The change to make
Let the evidence change the solution.
Let the team investigate the problem. Watching customers struggle with verification gives it a better question to test: would a simpler verification step help more people complete setup?
The team can test that assumption before committing to a full build. It follows the result through to customer use and keeps improving. Research, design, and development stay connected, with room to change the solution as evidence arrives.
