Product discovery and user research

Use this when the user names interviews, user research, a survey, a beta, or a pile of feedback to make sense of. The output is either the plan for the research or the ranked themes that come out of it — plus what changes on the roadmap.

What you need first

The question the research must answer, the users to talk to, and any notes or transcripts on hand. No users to reach means you draft the recruiting ask and the screener first. No transcripts means you plan the interviews before you synthesize anything.

Plan it, run it, mine it

1. Name the one product outcome this research serves, then the one question it must answer. A session that tries to answer three learns nothing. Write what a yes and a no would each change.
2. Plan the interviews: screened participants, one story opener — "tell me about the last time you..." — plus follow-up probes, never a long question list. No leading questions, and no pitching the solution mid-call. Pilot the guide on one or two sessions, then run thirty to sixty minutes with a neutral interviewer. Keep recruiting self-scheduling through in-product intercepts or a rotating panel.
3. Work transcripts one at a time: verbatim quotes with speaker, role, and date; numbers kept separate with their time period.
4. Mine for themes. Rank by count, lead each theme with the two or three quotes that carry it, and call out language that repeats across users — the same phrase from four different people is worth more than any single quote. Three accounts saying it is a pattern; one is an anecdote.
5. Map themes onto the outcome tree: outcome, then opportunity, then solution, then the assumption test each one needs.
6. Close the loop. Each theme gets a verdict: roadmap item, needs more evidence, or parked with the reason. Offer to file roadmap items as tickets on a yes.

Roll it up

When the ask is a periodic rollup rather than a single study, read the week's notes, the theme log, and any connected feedback source, and log what you could not reach. Hold the floor: at least one customer conversation per week, and flag a week with none as a gap. Rank the themes, lead each with its two strongest quotes, give each a verdict, and keep a log of briefed themes so the same one is never rolled up twice without saying what is new since. A week with no new feedback is one line saying it was quiet, not a rollup.

Output

The interview plan or the dated theme brief, with the roadmap verdict on each theme and the notes saved beside it. Never pad to look busy.

When the sample is thin

Say the sample is thin and refuse the verdict; name how many more conversations would settle it. Never invent a quote, a user, or a count, and never brief one interview as a finding.