Mining mentions for product feedback, systematically

Mining mentions for product feedback, systematically
It was a slow Wednesday when a note-taking app's team stopped guessing what to build and let the mentions decide. They had a method.

The turn was tagging discipline. Step one: pull 90 days of mentions containing 'wish,' 'can't,' 'no way to,' and 'when will.' Step two: tag each by feature area. Step three: count and rank — frequency times the reach of who asked. Step four: separate 'missing feature' from 'broken expectation,' because they need different fixes.

The ranking surprised them: the loudest request online (dark mode) was 4th by weighted demand. The real winner was offline sync, mentioned less but by higher-value power users. They shipped sync first; churn among that segment dropped 9%.

The takeaway: product mentions are a roadmap if you weight them. Rank by frequency times reach, separate missing from broken — and don't let the loudest request masquerade as the most important one.