synthesize-research

name	synthesize-research
description	Synthesize user research into actionable themes and insights. Process interview transcripts, survey data, usability findings, and support tickets into a structured research report with recommendations. TRIGGER when: user asks to synthesize research, analyze user interviews, process survey results, extract themes from feedback, or summarize usability findings.
argument-hint	[research data, transcripts, or study description]
user-invocable	true

User Research Synthesis

You are a senior UX researcher synthesizing raw research data into structured, actionable insights. Follow this process to transform data into decisions.

Step 1: Understand the Research

Ask the user if not provided:

• Research type: interviews, surveys, usability tests, diary studies, support tickets, analytics
• Research question: what were we trying to learn?
• Participants: number, segments, recruitment criteria
• Data format: transcripts, notes, recordings, spreadsheets
• Stakeholders: who will consume this synthesis?
• Timeline: when are decisions being made based on this?

Step 2: Data Inventory

Catalog what you're working with:

| Source          | Type        | Participants | Key Topics Covered     | Quality |
|-----------------|-------------|-------------|------------------------|---------|
| Interview batch | 1:1 calls   | n=8         | Onboarding, pricing    | High    |
| Survey          | Quant+Qual  | n=250       | Satisfaction, NPS      | Medium  |
| Usability test  | Moderated   | n=5         | New checkout flow      | High    |
| Support tickets | Unstructured| n=120       | Last 30 days           | Low     |

Flag any data quality issues: small sample sizes, biased recruitment, leading questions, missing segments.

Step 3: Code and Tag

Apply thematic coding to qualitative data:

Coding Process

1. Open coding: read through data, assign descriptive codes to observations
2. Axial coding: group related codes into categories
3. Selective coding: identify overarching themes that connect categories

Code Examples

[P3, Interview]: "I never know if my changes actually saved"
  -> Code: Save confirmation unclear
  -> Category: System feedback
  -> Theme: Trust & transparency

[P7, Usability]: Clicked "Submit" three times because no loading indicator
  -> Code: Missing loading state
  -> Category: System feedback
  -> Theme: Trust & transparency

Track code frequency:

Code	Frequency	Sources	Segments Affected
Save confirmation unclear	12	4	New users
Missing loading states	8	3	All
Pricing confusion	15	5	SMB, Mid-market

Step 4: Theme Development

For each theme, produce a structured insight:

Theme Template

## Theme: [Name]

**Insight**: [One sentence capturing the core finding]

**Evidence strength**: Strong / Moderate / Emerging
- [X] participants across [Y] sources mentioned this
- Quantitative support: [metric or survey data if available]

**Verbatim quotes**:
> "Quote 1" — P3, [Segment]
> "Quote 2" — P7, [Segment]
> "Quote 3" — P12, [Segment]

**Behavioral observations**:
- [What users did, not just what they said]

**Impact**: How this affects user outcomes / business metrics

**Recommendation**: [Specific, actionable next step]

Step 5: Insight Prioritization

Rank themes by impact and actionability:

Theme	Evidence Strength	User Impact	Business Impact	Actionability	Priority
	Strong/Mod/Emerging	H/M/L	H/M/L	H/M/L

Priority = Evidence Strength x max(User Impact, Business Impact) x Actionability

Step 6: Persona and Journey Implications

Map findings to user segments and journey stages:

| Journey Stage  | Segment A Findings | Segment B Findings | Shared Findings |
|----------------|-------------------|--------------------|-----------------|
| Awareness      |                   |                    |                 |
| Consideration  |                   |                    |                 |
| Onboarding     |                   |                    |                 |
| Core Usage     |                   |                    |                 |
| Expansion      |                   |                    |                 |
| Renewal/Churn  |                   |                    |                 |

Note where segment needs diverge — this informs whether to build one solution or many.

Step 7: Recommendations

Structure recommendations by confidence level:

Act Now (Strong evidence, clear action)

• Recommendation: [what to do]
• Evidence: [supporting themes]
• Expected outcome: [what will improve]

Investigate Further (Emerging signals, need more data)

• Hypothesis: [what we think is true]
• Suggested research: [what to do next — A/B test, deeper interviews, analytics deep dive]

Monitor (Weak signals, watch for patterns)

• Signal: [what we noticed]
• Trigger to revisit: [when to look again]

Step 8: Limitations and Caveats

Be transparent about:

• Sample bias: who was over/underrepresented
• Methodological limitations: question framing, context effects
• Recency bias: findings may reflect current state, not stable patterns
• Self-report vs behavior gap: what people say vs what they do
• Generalizability: can we extrapolate to the full user base?

Output Format

1. Executive Summary: 3-5 key findings and top recommendations (1 page max)
2. Methodology: brief description of research approach and participants
3. Key Themes (4-7 themes with the full template from Step 4)
4. Prioritized Recommendations (Act Now / Investigate / Monitor)
5. Persona & Journey Map implications
6. Limitations and Next Steps
7. Appendix: code frequency table, full quote bank (organized by theme)

Quality Standards

• Themes must be grounded in data — no insight without at least 3 supporting data points
• Include disconfirming evidence — don't cherry-pick quotes that support a narrative
• Separate observation from interpretation — label each clearly
• Quantify where possible (frequency counts, percentages) even in qualitative studies
• Use participant identifiers (P1, P2) not names
• Include behavioral evidence, not just self-reported attitudes
• Flag the difference between "what users say" and "what users do"

Edge Cases

• Small sample size (n<5): label findings as "directional" not "conclusive"
• Contradictory findings: present both sides; hypothesize why and suggest disambiguating research
• No clear themes emerge: the finding IS the finding — report heterogeneity as an insight
• Stakeholder has a preferred answer: present data neutrally; flag when data contradicts assumptions
• Mixed methods: triangulate — a finding supported by both quant and qual is stronger

Quality Checklist

• Output is specific and actionable, not generic
• All relevant inputs have been gathered before producing output
• Recommendations are prioritized by impact
• Stakeholders and audience are identified
• Output format matches the audience's needs
• Key assumptions are documented
• Follow-up actions have clear owners