agentleFS
nisarthnisarth/claude-skills-store/skills/product-management/synthesize-research/SKILL.md

synthesize-research

Skill1 starsChanged 7 months agoMITView on GitHub

Synthesize user research from interviews, surveys, and feedback into structured insights. Use when you have a pile of interview notes, survey responses, or support tickets to make sense of, need to extract themes and rank findings by frequency and impact, or want to present research to stakeholders. Trigger phrases: "synthesize research", "analyze interviews", "summarize user feedback", "research findings", "what did users say", "theme analysis", "synthesize feedback", "research synthesis", "interview analysis", "survey results"

4 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Preserve the user's voice: Use direct quotes to bring findings to life.

    line 145
  2. Distinctive:

    Include counter-evidence: If 8 of 10 users had a problem but 2 did not, say so.

    line 151
  3. Distinctive:

    Connect to action: Every insight should lead to a recommendation.

    line 152

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

User Research Synthesis

Synthesize raw user research data -- interviews, surveys, feedback, support tickets -- into structured insights with themes, patterns, and actionable recommendations.

Workflow

Step 1: Inventory the Data

Catalog what is available:

SourceVolumeFormatDate Range
User interviews[N]Transcripts/notes[Range]
Survey responses[N]Quantitative + open-ended[Range]
Support tickets[N]Text[Range]
NPS/CSAT comments[N]Text + score[Range]
App store reviews[N]Text + rating[Range]
Sales call notes[N]Notes[Range]

Note any gaps: are certain user segments, time periods, or topics underrepresented?

Step 2: Code the Data

Extract discrete observations from the raw data:

  1. Read through all sources and tag each meaningful statement with:
    • Topic: What is this about? (onboarding, pricing, performance, etc.)
    • Sentiment: Positive, negative, neutral, mixed
    • Intensity: How strongly felt? (mild frustration vs. deal-breaker)
    • User segment: Who said this? (role, plan tier, tenure, company size)
  2. Create observation cards:
Observation: [Direct quote or paraphrased insight]
Source: [Interview 3, Survey Q5, Ticket #1234]
Topic: [Category]
Sentiment: [Positive/Negative/Neutral]
Segment: [User type]
Frequency: [How many sources mentioned this]

Step 3: Identify Themes

Group observations into themes:

### Theme: [Name]
**Frequency**: [X of Y sources mentioned this]
**Segments affected**: [Which user types]
**Severity**: [Critical / Important / Minor]

**Summary**: [2-3 sentences describing the pattern]

**Supporting observations**:
- [Observation 1] (Source: [X])
- [Observation 2] (Source: [X])
- [Observation 3] (Source: [X])

**Counter-evidence**: [Any data that contradicts this theme]

Rank themes by a combination of frequency (how often it appears) and impact (how much it matters to users who mention it).

Step 4: Extract Insights

Transform themes into actionable insights:

### Insight: [Actionable statement]
**Based on**: [Theme(s)]
**Confidence**: [High/Medium/Low]
**Evidence strength**: [Strong/Moderate/Weak]

**What we learned**: [The finding]
**Why it matters**: [Business or product impact]
**What to do about it**: [Recommendation]

Confidence levels:

  • High: Consistent across multiple sources and segments, supported by quantitative data
  • Medium: Mentioned by several sources but not all segments, or qualitative only
  • Low: Mentioned by few sources, may be edge case, needs more data

Step 5: Compile the Synthesis

## Research Synthesis: [Topic/Study Name]
Date: [Date] | Researcher: [Name]

### Study Overview
- **Objective**: [What we set out to learn]
- **Methods**: [Interviews, surveys, etc.]
- **Participants**: [N] participants across [segments]
- **Date range**: [When research was conducted]

### Executive Summary
[3-5 sentences: key findings and their implications]

### Top Insights (Ranked by Impact)
1. [Insight]: [One-line summary]
2. [Insight]: [One-line summary]
3. [Insight]: [One-line summary]

### Detailed Findings
[Theme sections with supporting evidence]

### Recommendations
| Priority | Recommendation | Insight | Effort |
|----------|---------------|---------|--------|
| P0 | [Action] | Insight 1 | S/M/L |
| P1 | [Action] | Insight 2 | S/M/L |

### Open Questions
- [Question that the research raised but did not answer]

### Methodology Notes
[Limitations, biases, and caveats]

Examples

Example 1: Interview Synthesis

User: "I have notes from 12 user interviews about our onboarding flow." Response: Code each interview for observations about onboarding. Group into themes (e.g., "users do not understand the value before being asked to configure," "setup requires information users do not have handy"). Rank by frequency and severity. Recommend specific changes.

Example 2: Survey Analysis

User: "Here are our quarterly survey results with open-ended responses." Response: Segment quantitative scores by user cohort. Code open-ended responses into themes. Cross-reference themes with satisfaction scores to identify what drives satisfaction and dissatisfaction. Highlight where quantitative and qualitative data agree or conflict.

Example 3: Mixed Sources

User: "Combine insights from support tickets, NPS comments, and 5 interviews." Response: Normalize across sources (support tickets skew negative, interviews are deeper). Weight insights by source diversity -- a theme that appears across all three sources is stronger than one from a single source. Note where sources tell different stories.

Best Practices

  1. Preserve the user's voice. Use direct quotes to bring findings to life. Paraphrased insights lose emotional weight.
  2. Count carefully. "Users said X" could mean 2 or 20. Always include frequency: "7 of 12 participants mentioned X."
  3. Separate observation from interpretation. "Users clicked the wrong button" is an observation. "Users found the UI confusing" is an interpretation. Present both but label them clearly.
  4. Acknowledge bias. Who was not in the study? Interviewees are often more engaged than average users. Survey respondents self-select. Note these limitations.
  5. Triangulate. A finding supported by multiple methods (interview + survey + analytics) is stronger than one from a single source.
  6. Prioritize ruthlessly. Present 3-5 key insights, not 20 observations. Stakeholders act on focused recommendations, not comprehensive reports.
  7. Include counter-evidence. If 8 of 10 users had a problem but 2 did not, say so. Understanding the exception often reveals the root cause.
  8. Connect to action. Every insight should lead to a recommendation. Research without action is academic, not product work.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.