What is the affinity-diagram skill?
The affinity-diagram skill is a SKILL.md in Owl-Listener/designer-skills for clustering many qualitative data points into themes and insight statements. Your AI coding agent works bottom-up and tags every observation with a participant ID so no single interview dominates the themes.
The affinity-diagram skill clusters many qualitative data points into themes and insight statements. Use it when synthesizing across multiple sessions or sources, with interview notes, observation data or survey responses as input.
The agent acts as a UX researcher and reads any files you provide first. It presents the result as a structured hierarchy of themes with insight statements and supporting evidence.
The eight instructions
- Extract individual observations, quotes and notes from the raw data.
- Cluster bottom-up into natural groups, without starting from predefined categories.
- Name each cluster with a descriptive theme label.
- Organize clusters into higher-level themes, typically 3 to 5 at the top level.
- Write an insight statement for each theme that captures the so what.
- Identify patterns: frequency, intensity and connections between themes.
- Prioritize insights by impact on design decisions.
- Present the diagram as a hierarchy with insight statements and supporting evidence.
The cross-interview sampling principle
The skill says to index evenly across all participants. Process each transcript fully before clustering, tag every observation with an ID such as P1, P2 and P3, and treat each participant as an equal source of signal.
After clustering, check that each participant appears at least once. Patterns found in only one interview are flagged as single-source, not discarded. This prevents building themes from the first transcripts and fitting the rest afterward.
Who the affinity-diagram skill is for
Best for
- Synthesizing across several interviews, sessions or sources.
- Turning raw notes into themes that feed a design brief.
- Ranking insights by their impact on design decisions.
- Research themes that inform a UX audit of an existing product.
Not for
- A single transcript. The SKILL.md points to summarize-interview for that.
- One segment's inner state. It points to empathy-map for that.
How to install Affinity Diagram Skill: Cluster Research Into Insights
/plugin marketplace add Owl-Listener/designer-skills
/plugin install design-research@designer-skillsAdd the marketplace
Run /plugin marketplace add Owl-Listener/designer-skills in Claude Code. Nothing installs yet; it only tells Claude where the skills live.
Install the plugin
Run /plugin install design-research@designer-skills. The skill sits at design-research/skills/affinity-diagram.
Ask for a synthesis
Point the agent at your interview notes and ask it to use affinity-diagram, with participant IDs on every observation.
Check that it loaded
Ask for a synthesis of several interviews. Expect themes with insight statements, supporting evidence, and every participant appearing at least once.
Key features
- Extract data pointsPulls individual observations, quotes and notes out of interview notes, observation data or survey responses.
- Bottom-up clusteringGroups related points into natural clusters and does not start with predefined categories.
- Named clustersGives each group a descriptive theme label that captures the essence of the group.
- Theme hierarchyOrganizes clusters into higher-level themes, typically three to five at the top level.
- Insight statementsWrites a clear insight statement for each theme that captures the so what.
- Pattern analysisNotes frequency, intensity and connections between themes.
- Impact rankingRanks insights by their impact on design decisions.
- Participant taggingTags every observation with a participant ID such as P1, P2 and P3 before grouping, so each participant is an equal source of signal.
- Single-source flagsPatterns that appear in only one interview are flagged as single-source and not discarded.
Frequently asked questions
What is the affinity-diagram skill?
It is a SKILL.md in the Owl-Listener/designer-skills repository, inside the design-research plugin. It tells an AI coding agent to cluster many qualitative data points into themes and insight statements, for use when synthesizing research across multiple sessions or sources instead of one.
How do I install affinity-diagram?
Run /plugin marketplace add Owl-Listener/designer-skills in Claude Code, then /plugin install design-research@designer-skills. The skill sits in the design-research plugin at design-research/skills/affinity-diagram, so installing that one plugin is enough to get it. You do not need the other plugins in the collection, and you can add more later by name.
How does affinity-diagram cluster data?
It clusters bottom-up. The agent extracts individual observations, groups related points into natural clusters without predefined categories, names each cluster, organizes clusters into higher-level themes, writes an insight statement for each theme, notes patterns and ranks insights by impact on design decisions.
What is the cross-interview sampling principle?
It says to index evenly across all participants. Process each transcript fully before clustering, tag each observation with a participant ID such as P1 or P2, and after clustering check that every participant appears at least once. If any are absent, go back and revisit the data.
What happens to patterns from only one interview?
They are flagged as single-source, not discarded. The skill treats each participant as an equal source of signal, so a pattern from one person stays visible in the output while the reader can still see that it has only one source behind it.
Should I use affinity-diagram for one interview?
No. Its description says to use it when synthesizing across multiple sessions or sources. For a single transcript it points to summarize-interview, and for one segment's inner state it points to empathy-map, so choose by how much data you have.
What does the affinity-diagram output look like?
The agent presents a structured hierarchy: top-level themes, usually three to five, with named clusters beneath them. Each theme has an insight statement and supporting evidence, and insights are ranked by their impact on design decisions. The agent can read your files first.
What license does affinity-diagram use?
It is MIT licensed. The skill ships in the Owl-Listener/designer-skills repository, which GitHub lists under the MIT License, and it lives at design-research/skills/affinity-diagram in that repository. Use and redistribution follow the terms of that license.