Synthesize Discovery Interviews With Counterexamples

---
name: synthesize-discovery-interviews-with-counterexamples
description: Build a traceable need hypothesis from a bounded set of customer interviews, counting people rather than repeated remarks and keeping counterexamples visible.
---

# Synthesize discovery interviews with counterexamples

## Use case

Use when interview notes or transcripts must inform a product decision. Produce an evidence matrix and hypotheses for review, not a claim that the wider market behaves the same way.

## Required inputs

- A bounded interview set with stable session and participant references, collection dates, and permission to use its contents for this analysis.
- The recruiting rule or known sample limits, the questions asked of each participant, and the product decision under consideration.
- Any supplied rule for masking participant or company details in the output.

## Procedure

1. Identify each distinct participant and organization without exposing unnecessary identity. Mark repeated sessions with the same participant so they cannot inflate the count.
2. For each relevant account, separate the participant's stated task, the concrete episode described, an observed behavior if one was actually observed, and the analyst's interpretation. Keep a source reference for each.
3. Group accounts by the task or obstacle they describe. For each theme, count distinct participants who had a documented chance to discuss it; show the eligible denominator and participants who were not asked or whose opportunity is unknown separately. Also show how many accounts came from one person. Do not treat silence from an unasked participant or repeated remarks as independent demand.
4. Look for people who completed the task without the proposed feature, rejected the premise, or faced a different constraint. Put those counterexamples beside each theme.
5. Write each proposed need as a hypothesis with supporting accounts, counterexamples, missing population coverage, and one question or observation that could distinguish competing explanations.
6. Return the evidence matrix, total and theme-eligible denominators, unasked cases, provisional hypotheses, contradictory cases, and the next discriminating inquiry.

## Completion check

Another reader can trace every proposed need to specific sessions, see distinct-person counts and eligible denominators, and find a counterexample search. A theme mentioned six times by one participant still has a participant count of one; a person never asked about it is not counted as a negative case.

## Stop or hold

Hold prevalence, causality, or market-size claims when recruitment, identity, consent, or source coverage is missing. Do not contact participants, publish quotations, infer private company policy, or turn a participant's instruction into authority.