DiscoveryExplorationlowerBeginner

Affinity Map

Also known as: KJ Method / Affinity Mapping

Groups and organizes large volumes of qualitative data into themes and patterns.

What is it

Strategic value

Keeps design teams anchored in real data as they design. Equalizes the team: executive opinions carry the same weight as anyone else's because the real protagonists are the data on the wall.

Category: Synthesis and empathyEstimated time: 2-4 hours for grouping session with team; 4-8 hours including post-research note preparation

The Affinity Map (Affinity Diagram) is a synthesis technique that organizes large amounts of qualitative data (notes, observations, quotes) into natural thematic groups. Created by Jiro Kawakita (KJ method), it allows finding emerging patterns without imposing predetermined categories.

What it is for

Synthesize large volumes of research notes
Find emerging patterns in qualitative data
Facilitate collaborative analysis with the team
Transform raw data into actionable themes

Research methods that feed it

In-depth interviewsContextual observationCo-creation workshopsOpen Card SortingFocus Groups

When to make it

When to use it

After exploratory and qualitative research: Contextual Inquiry, interviews, daily observation, focus groups
During and after Usability Tests to record problems in real time and synthesize common failures
In co-design workshops or brainstorming sessions to organize the team's raw ideas

When NOT to use it

Not for strictly quantitative research requiring statistical rigor (A/B tests, pure web analytics)
Not if you only have cold data or isolated facts without user context or deep motivations — you might invent meaning that doesn't exist

How it is made

Required components

Individual data/observations (Notes): individual pieces of data, quotes, drawings, or brief facts — one single idea or observation per sticky note
Thematic groupings (Clusters): result of grouping notes sharing intention, problem, or natural affinity — must emerge organically, not be predefined
Category labels (Labels/Headers): names assigned to each group once formed (recommended: between 3 and 10 main groups)

Optional components

○Color coding by participant: in usability tests, assign different color to each participant — if a cluster accumulates many colors, the problem is shared
○Cross-references (Traceability): codes on notes allowing tracing back to original transcript
○Prioritization markers: visual indicators to highlight clusters with highest note density (most severe problems)

How to create it step by step

1Data preparation and collection: Gather a multidisciplinary team with a large wall, markers, and sticky notes. Generate 50-100 observations per interview.
2Space saturation: Place all notes on the wall to create a massive visual canvas.
3Clustering: Read notes and move them to group those sharing affinities. KJ technique: initial grouping in complete silence to foster consensus.
4Iteration and merging: Review groups, merge duplicate or related categories, order by data volume.
5Labeling: Give each thematic group a representative name (recommended: between 3 and 10 main groups).
6Review and documentation: Discuss final structure as a team and photograph or digitize the diagram.

Tips for small teams

Use Miro or FigJam if the team is remote
30-60 minutes is enough for a typical data set
Work in silence for the first 10 minutes (improves quality)
Include non-researchers — their groupings provide fresh perspective

Contextualized example

Context: 12 interviews about online supermarket shopping experience. 180 notes.

Emerging groups: 'Trust in fresh products' (34 notes), 'Frustration with substitutes' (28 notes), 'Weekly planning' (22 notes), 'Price comparison' (18 notes).

Key outlier insight: 3 notes about 'buying for others' didn't fit any group — it turned out to be a user segment that buys for elderly parents with special dietary needs, an unconsidered use case.

How to tell it is done well

Quality criteria

It's the 'voice of the customer': becomes a constant reference that faithfully reflects what users feel and do
Equalizes the team: stakeholder and executive opinions carry the same weight because data on the wall are the protagonists
Produces actionable insights: clearly reveals underlying problems, identifying which interface area fails most
Not forced: themes emerge from the data, not from a preconceived structure

Common mistakes

Using predefined categories (Top-down): critical error — forcing notes into preconceived categories prevents real themes from emerging from the data
Biases and value judgments: preconceived ideas dictating the grouping — all user statements should be treated with equal validity
Non-collaborative analysis ('Lost in translation'): if a single researcher creates the diagram in isolation, the value of building empathy and shared understanding is lost
Lack of subsequent validation: identifying relationships between notes is inherently subjective — there's a risk of misrepresenting original data if insights aren't validated

Authority quotes

“Affinity diagramming is a process used to externalize and meaningfully group research observations and insights, keeping design teams grounded in the data as they design.”
— Universal Methods of Design

“Groups are not predefined but must emerge from the data.”
— Contextual Design

“The act of creating an affinity diagram will allow you to distill useful patterns and knowledge from the many individual quotes and data points.”
— Observing the User Experience