Interview findings have a well-known weakness: they’re convincing but hard to measure. Product teams, rightly, ask how much something is worth before changing the roadmap. My answer is to turn the finding into a testable hypothesis.
From finding to hypothesis
At Gameram, I turned interview findings into A/B tests. What people told us became a hypothesis: if the interview shows what gets in the way, changing that point should change the measured behavior.
One of those tests increased conversion by 30%.
The test doesn’t replace the interview. The interview says what to test and why. The test says how much.
When the numbers come first
It works the other way too. After dark mode launched, the data showed many people were still on the light theme. In interviews with active and new users, the design worked well. The problem was something else: many people didn’t know they could change the theme.
A survey of about 650 users confirmed it: 74% of those who knew about the option used dark mode, versus 54% of those who didn’t. The data showed the outcome. The research explained the reason. Read the full case.
What I take from this
- Write the finding as a hypothesis. “If we change this, we expect to see that” makes it clear what to measure.
- Let the interview pick the test. Testing what people said gets in the way costs less than testing everything.
- Go back to people when the number surprises. An unexpected result is a new research question.
A good insight becomes an experiment. And a good experiment goes back to research.
Some details have been omitted for confidentiality.