research-intake

Research Intake — Processing Raw User Data

Transform messy real-world data into structured insights.

Accepted Data Types

This skill processes ANY user data the user provides. Common types:

• Interview transcripts (pasted text, uploaded files)
• Survey results (CSV, pasted responses, summaries)
• Support tickets / bug reports
• App store reviews / NPS verbatims
• Analytics data (pasted numbers, screenshots described)
• Usability test notes
• Sales call notes
• Forum/community posts
• Customer emails or chat logs
• Competitive analysis notes

Processing Flow

Step 1: Receive Data

When the user provides data, acknowledge what you received and classify it:

"I see you've shared [type of data]. Let me analyze this for patterns. Is there anything specific you want me to look for, or should I do an open analysis?"

Step 2: Extract Signals

For each data source, extract:

## Data Source: [type and brief description]

### Key Themes (frequency-ranked)
1. **[Theme]** — mentioned [N] times
   - Supporting evidence: [direct observations, not quotes]
   
### User Pain Points
- [specific, actionable pain points with severity]

### Unmet Needs
- [things users want but don't have]

### Positive Signals
- [things users value about current solutions — preserve these]

### Behavioral Patterns
- [how users actually behave, especially if different from expectations]

### Surprising Findings
- [anything unexpected or counter-intuitive]

Step 3: Cross-Reference

If multiple data sources are provided, identify:

• Converging signals: Themes that appear across multiple sources (high confidence)
• Diverging signals: Contradictions between sources (needs investigation)
• Gaps: Important questions that NO data source answers

Step 4: Confidence Assessment

## Research Confidence Assessment

| Insight | Data Sources | Sample Size | Confidence | Action |
|---------|-------------|-------------|------------|--------|
| [insight] | [which sources] | [how much data] | High/Med/Low | [use as-is / validate further / treat as hypothesis] |

Step 5: Feed Forward

Present the synthesis and ask:

"Here's what I'm seeing in the data. Should I use these insights to build personas, or do you want to adjust anything first?"

Structured insights from this skill feed directly into:

• persona-builder — behavioral patterns become persona foundations
• jobs-to-be-done — unmet needs become job statements
• user-journey — pain points map to journey phases
• telemetry-designer — themes suggest what to measure

Rules

• Don't invent data — if the sample is small, say so. Don't over-generalize from 3 data points.
• Distinguish observation from interpretation — "5 users said X" is observation. "Users hate X" is interpretation. Label both.
• Preserve nuance — don't flatten contradictions. Surface them.
• Ask about context — "When was this data collected? Has anything changed since?"
• Respect privacy — if data contains PII, note it and suggest anonymization