synthesize-research

Paste your interview notes and state your research question. The skill codes each observation, groups them into themes via affinity mapping, counts frequency across participants, and outputs findings ranked by frequency and impact with supporting quotes.

Use qualitative data to explain what and why, and quantitative data to validate how many. The skill triangulates across sources, flags disagreements between stated preferences and behavioral data, and presents combined evidence for each finding.

It accepts pasted text such as interview transcripts, survey responses, and support tickets, plus uploaded research documents. It can also pull from connected knowledge bases, user feedback tools, product analytics, and meeting transcription services when available.

Aim for 5-8 strong findings rather than many weak ones. Each finding includes an evidence base, frequency count, impact assessment, and a confidence level of high, medium, or low depending on source strength.

Common mistakes include reporting averages without distributions, ignoring non-response bias, over-interpreting small differences, and treating Likert scales as interval data. The skill checks for these issues when interpreting survey results.

Findings from very small samples, such as two interviews, are hypotheses rather than conclusions. Low-confidence findings need follow-up research, and the skill explicitly marks confidence levels and open questions requiring further investigation.