Startup / LEARNING STUDIO
Many conversations yield only praise and feature requests.
A defined population and a voluntary interview.
Start with diagnostics; skip explanations you can already apply independently.
Ask about the latest event: trigger, sequence, participants and obstacles. Probe records and outcomes, allowing no problem as a valid answer. Understand the process before pitching. AI roleplay can rehearse a guide but is not customer evidence.
Recruit beyond supporters and record sampling. Separate quotations, observations and interpretation, anonymizing notes. Obtain separate agreement for recording, quoting or public display. Small interviews can reveal mechanisms and language, not population proportions.
Worked example · fictional
Fictional: five people say yes to a hypothetical free AI tool.
Investigate the recurring manual workflow, retaining all four other cases and revising the target group.
Explain the product repeatedly until the user agrees.
Answers reflect courtesy and leading.
Use neutral past-event questions and hear the whole sequence.
Watch 00:00–03:33 on testing business hypotheses with customers; the source also links a transcript.
Output: Turn three opinions into questions about observable behavior.
Free public video; use the original clip if the transcript is unavailable. · Source checked 2026-10-26
Read planning, interviews and analyzing findings; distinguish a research question from a product idea.
Output: Map anonymized evidence to interpretations.
Free public guidance. · Source checked 2026-10-26