Claude now runs UX study planning end to end, real people answer.
1/ Plan the study first
Desk research summary, assumption map, research plan, a check on any synthetic user answer.
Synthetic answers stay hypotheses until a real study answers them.
2/ Recruit and consent real people
Recruitment brief, behaviour screener, consent form, data handling plan, incentives, participant database.
The screener checks fit to the brief, a person decides, and nobody is scored.
3/ Write guides that do not lead
Interview guide, moderated and unmoderated test plans, usability benchmark, survey, diary study.
Success criteria are set before the first participant sits down.
4/ Run and capture the sessions
Contextual inquiry, accessible sessions, an AI feature study built around wrong answers, same-day debriefs.
Quotes stay verbatim with timestamps, interpretation kept apart.
5/ Synthesise with every quote kept
Thematic analysis, usability findings, survey analysis, insight statements, an evidence trace audit.
Claude never invents a participant, a quote or a finding. Severity rates problems, never people.
6/ Map where the experience breaks
Journey map, jobs to be done, opportunity solution tree, each cell tied to what participants said.
Gaps read “not researched”, never a guess.
7/ Share it so something changes
Readout with the decision on slide one, action tracker, atomic repository entries, tagging taxonomy.
A named person decides by a date, and the next study starts from what is already known.
