Study insights

# Study insights

Use once a study is complete, to find the patterns that run across its interviews rather than the detail inside any one of them.

## Definitions

Use these meanings consistently in every step.

- **Study:** The study record being analyzed, or the study whose record link the user pasted into the chat
- **Pattern:** A need, blocker, or workaround that appears in more than one interview. A single mention is an observation, not a pattern.
- **Segment:** A group of participants sharing a role, company size, or use case, used to check whether a pattern holds across the whole study or only part of it
- **Strength:** How well supported a pattern is, based on how many participants raised it and how consistently they described it

## Workflow

1. Read every interview in the study. Confirm the scope first when the user wants a subset, such as a date range or a single segment.
2. Record how many interviews were analyzed and who participated. Every later claim is bounded by that set.
3. Extract the points of feedback from each interview, keeping the participant's own words attached
4. Group them into patterns. Name each pattern by the need underneath it, not by the feature customers asked for.
5. Rate strength for each pattern and cite how many participants raised it and how consistently
6. Check each pattern across segments. Say plainly when a pattern holds for one segment and not another. That distinction usually matters more than the pattern itself.
7. Note contradictions between participants rather than resolving them. A contradiction is a finding.
8. Search the workspace for existing ideas that match each pattern, and suggest new ideas only where nothing covers the need
9. Recommend what to do next, separating what the evidence supports from what would need more research

## Rules

- If the study is missing or has no completed interviews, ask short follow-up questions before analyzing
- State the number of interviews analyzed. Every claim is bounded by that sample.
- A pattern raised by one participant is labeled an observation, not a pattern
- Quote participants directly when citing evidence
- Reference studies, interviews, and ideas by reference number and name
- Do not create ideas or update records. Present the analysis and let the user apply it.
- Headers use sentence case. No em dashes, en dashes, or curly quotes.

## Output

```
## Scope  
<interviews analyzed, participants, and any filter applied>  
  
## Patterns  
| Pattern | Underlying need | Participants | Strength | Representative quote |  
  
## Segment differences  
- <pattern, the segment it holds for, and where it does not>  
  
## Contradictions  
- <what participants disagreed about>  
  
## Single observations  
- <raised once, worth watching>  
  
## Matches existing ideas  
| Pattern | Existing idea |  
  
## Recommendations  
1. <action the evidence supports>  
  
### Needs more research  
- <question, and what would answer it>
```

## Quality checks

- The scope states how many interviews were analyzed
- Every pattern names a participant count and carries a quotation
- Single mentions are separated from patterns
- Segment differences are stated rather than averaged away
- Recommendations distinguish what the evidence supports from what needs more research