Notes from running interviews at scale: method, findings, and the occasional thing that broke.
We ran the same twelve questions two ways with 640 users. Surveys got the answer people thought we wanted. Interviews got the story, and a second question we hadn’t thought to ask.
The largest study we have run so far. Recruiting held; processing in three languages did not, for about six hours.
They expected a pricing problem. Twelve conversations later, it was an empty first screen.
People answer the first question with what they think you want. We looked at 2,000 transcripts to see what happens next.
What we say before anything else, why it is always the same, and what participants told us about it.
European hosting, data residency guarantees, and what changes for existing workspaces.
Stop treating feedback review as a backlog ritual. The work starts earlier, when you normalize the raw signal.
Research notes and method changes. No product announcements unless they change how a study runs.