“You have 8 user interviews that all say something different - how do you synthesize without guessing?”
Turn scattered interviews into insight without making things up
Recommended AI Agent Skills
Evidence-grounded user interview synthesis, verbatim quotation tagging, and friction point discovery.
Multi-source deep qualitative research synthesis without AI hallucinations or speculative extrapolations.
A Discovery-stage interview synthesis workflow that keeps raw customer quotes clearly separate from the PM's own interpretation.
Action framework
- Pull direct user quotes out of your notes separately - keep "what the user said" clearly apart from "what I think it means".
- Count how many interviews raised the same issue - one person saying it isn't an insight, 5 out of 8 saying the same thing is.
- Attach direct evidence to every conclusion (a quote or a repeat count) - if you can't attach evidence, it's a hypothesis, not a conclusion.
- Note contradictions between interviews instead of smoothing them over - a contradiction is information too.
- Before sending the report, reread it and ask: where in my notes does this line come from? If you can't point to it, cut it.
Before
"Users generally don't like this feature, we should redo the whole thing."
After
"5 of 8 users struggled at the confirmation step (quote: 'I couldn't tell if it went through' - P3, P5, P6). 2 of 8 mentioned load speed. 1 of 8 had no issues. → The problem is concentrated in the confirmation step, not enough evidence to redo the whole flow."