Turning discovery notes into product decisions

Founders can end a discovery round with many quotes and no agreement about what to build. The problem is often that findings, assumptions and proposed responses have been mixed together.

Organise evidence around the question

For each research question, assemble supporting and conflicting observations. Preserve enough context to understand the participant’s role and circumstances without exposing unnecessary personal information.

GOV.UK’s guidance on planning research includes sharing findings with the team. A useful founder adaptation is to attach a decision to each material finding: proceed, revise, stop or gather specific missing evidence.

Do not select quotations solely because they express the founder’s preferred narrative. A vivid comment may explain one case while saying little about the wider segment.

Compare possible responses

A recurring reporting delay might be addressed by a new interface, better integration, clearer ownership or a service change. The product team should compare those options before assuming the answer is another feature.

For an illustrative Thai logistics startup, interview notes may show that customers cannot interpret exception codes. Rewriting messages and adding an escalation contact could be more useful than redesigning the entire dashboard.

State why the chosen response is expected to help and what evidence would show it did. This preserves the reasoning when new team members join or the initial plan fails.

Make uncertainty actionable

“More research needed” is too vague to guide work. Specify the missing participant group, transaction example or observation. Assign an owner and decide whether development can continue safely without the answer.

After the change, return to the original user problem. Did the relevant task become easier or did users simply learn a new workaround? Keep new evidence linked to the decision it tests.

Keep the original question beside the product decision in the team backlog. When a proposed feature changes, the record should still explain the user problem it was intended to address. This prevents research from becoming detached evidence reused to justify several unrelated roadmap choices after its context has been forgotten.

GAT’s pilot-to-revenue guide makes the commercial decision explicit. Discovery deserves the same clarity. A short decision record showing what changed, why and what remains uncertain is more useful than a large presentation that leaves the roadmap untouched.