USER EXPERIENCE

How to do affinity mapping that doesn’t suck

A no-fluff way to interpret and organize your research findings

When I started my career in UX, I thought affinity mapping (a.k.a. sorting or diagramming) was just grouping solution ideas or research findings by topics. I even created several affinity maps that I now can’t look at without shame. Over time, I realized there’s a lot more behind good affinity maps — and below is my recipe.

Of course, there’s no one-size-fits-all, so I’ll focus on the type of affinity mapping that follows research (as opposed to solution brainstorming) and I’ll illustrate it with examples from a B2B context.

Step 1. Prepare good notes

Most affinity mapping exercises fail because the input material isn’t clean enough. After UX research, you end up with hundreds of information bits — but not all of it is useful.

A direct quote from a participant is rarely a good note for affinity mapping. Yes, you want to include the “voice of the user,” but contrary to popular belief, concise and insightful user quotes are the exception, not the norm. When people speak naturally, they often ramble, use filler words, or take a few tries to get to the point.

Raw data needs interpretation. Those 5 minutes of the user’s thinking aloud or 2–3 paragraphs of interview auto-transcript should be distilled into concise statements:

• Action (who does what) + insightful context of this action (how, why, with what outcome, etc.).
• One full sentence per note — not just keywords.
• Each note should be understandable on its own.
• One note = one finding.

This way, you’ll have far better input for sorting and grouping — and the entire process will move faster.

Also, make sure each note is tagged with the code of the interview it came from. If you’re, like me, working in a B2B or enterprise business where users are professionals from specific companies, you can include company context in the code, too.

For example, if you interviewed 4 representatives of one company and 3 persons from another company, you can put the following codes on corresponding notes: A1, A2, A3, A4; B1, B2, B3.

Below are three examples of transcripts paired with affinity mapping notes. To keep things vivid, I’ll be using realistic data from the safety management domain, which I’ve worked with extensively in recent years. Safety management is about keeping people safe in hazardous workplaces, like mines, chemical plants, or oil refineries. It involves analyzing risks, designing safeguards, and inspecting whether everything is followed correctly so nobody gets hurt.

Example 1

🗣 Interview transcript

Company A, interviewee 1. Liam O’Connell, Health & Safety Manager:

“Um, yeah, so whenever there’s an incident — say, someone trips in the warehouse or, like, something minor but still reportable — I usually get a call first. And then I go into the system to check if anything’s been entered already. Sometimes it has, sometimes not. And if it’s not, I have to, you know, track down the details, like who was involved, what exactly happened, and all that, before I can even log it properly. Oh, forgot to mention, it can also be an email or MS Teams message, not just a call. But usually it’s a call.”

📝 Affinity mapping notes

A1: I get a call or email about a safety incident and log it into the system for the person who contacted me.

A1: I check the incident management system for existing reports before logging a new safety incident to avoid duplicates or missing details.

Example 2

🗣 Interview transcript

Company B, interviewee 1. Anika Sharma, Health & Safety Manager:

“We have this weekly routine where I sit down with EHS supervisors to go over any open risks that haven’t been addressed yet. We usually look at what’s pending from previous audits or stuff that came up from the last safety walkthrough. It helps keep things from slipping through the cracks, you know…”

📝 Affinity mapping note

B1: I review unresolved risks with supervisors weekly to make sure issues from audits or inspections are being followed up on.

Example 3

🗣 Interview transcript

Company A, interviewee 3. Jonas Petrovic, Safety Coordinator:

“I wish the platform would just flag when someone’s forgotten to upload evidence — like photos or documentation. Right now, we sort of assume it’s there, but only later realize something’s missing when we’re reviewing the report more thoroughly. It’s kind of annoying because then we have to go chase people down after the fact.”

📝 Affinity mapping notes

A3: I only realize evidence is missing during incident report review because the system doesn’t flag unsubmitted attachments.

A3: My team has to manually follow up with incident reporters when required evidence is missing, and it is annoying.

It’s not a strict rule, but words like “because,” “that’s why,” “so that,” “in order to,” or “and only after that” often signal that your note contains a real finding — something valuable and worth keeping. On the other hand, words like “we” instead of “I,” or qualifiers such as “usually,” “typically,” and “normally” can indicate generalizations — when the person is speaking hypothetically rather than describing a concrete experience. Something to watch for during affinity sorting.

Step 2. Identify preliminary themes

Now that I’ve finished my rant about note quality, we can finally start organizing them on the canvas. However, the trick is that you’re not ready to build final groups yet. Instead, start with loosely sorting the notes into broad, temporary themes.

If you’re researching workplace safety, your early themes might be:

— “Safety training,”
— “Safety incident reporting,”
— “Investigating incidents,”
— “Safety culture,”
— “Inspections and audits.”

It’s important to remember this is just a preparatory step. These themes will probably shift quite a bit once you begin forming more specific and meaningful groups, so don’t get attached to them too much.

Step 3. Create first-level groups

Once you have those preliminary themes, go through them one by one and start forming groups of notes. But don’t group notes just because they share keywords — they should point to the same finding. For example, they might describe the same user need, reflect a similar work habit, or highlight the same pain point.

Each group should include 2 to 5 notes. Sounds a bit artificial? Maybe at first — but here’s the reasoning behind it:

• If you have more than 5 notes, there’s a good chance you’ve combined two findings into one group — try to split it.
• Fewer than 2 notes? That’s not a real group. One-note groups amplify a single data point, thus increasing bias.

Each group needs a clear, concrete name. And at this stage, that doesn’t mean a broad theme like “Safety Inspections” or “Risk Management.” Instead, go for an actionable statement that captures the essence of the finding. A useful technique is to phrase group names as “I statements” — describing the action from the interviewee’s point of view. For example:

“I verify if evidence is attached to an incident report.”

“I reassess the risk after safety violations have happened.”

“I investigate sensitive details in one-on-one conversations.”

This kind of naming keeps your clusters grounded in real behavior and makes it easier to spot patterns, needs, and opportunities later on.

You’ll probably come across a few outliers — notes that describe something very specific to a single participant or a unique business rule of their company. That’s fine. Just place such notes aside on the “Outliers” board so they don’t pollute your data.

I haven’t mentioned this yet, but affinity mapping can be both a solo exercise and an engaging workshop where the entire team works together to unpack research results. In fact, the team format works even better: it not only surfaces insights but also helps promote the findings and drive real change. Too often, research is recognized but not acted upon. But when the whole team — designers, engineers, product managers— sorts sticky notes with findings, they subconsciously absorb the material and start to accept new facts.

In a workshop format, you can include short review rounds after each step, letting participants wander around the canvas to make small adjustments — tweaking group names or regrouping notes until everything feels logical and clear.

Step 4. Identify patterns in groups

Once you’ve grouped everything, it’s safe to let go of those early themes. They served their purpose in keeping you oriented among hundreds of notes, but now it’s time to surface the real patterns.

Step back and scan the groups for patterns or recurring observations:

• Are there several groups connected to the same workflow stage?
• Do people behave similarly in comparable situations?
• Do you see the same pain points pop up in more than one group?

Write these notes down plainly, without overthinking; these patterns will evolve as you move forward.

Continuing the safety topic, your patterns might sound like:
— “Non-centralized risk register,”
— “Usage of reports in safety briefings,”
— “Personal conversations to drive safety culture,”
— “Proactive vs. reactive safety.”

(Compare these to much broader themes from Step 2: “Safety incident reporting,” “Inspections and audits,” “Safety culture,” etc.)

The difference is that now your observations are sharper and closer to something actionable. But don’t treat them as final insights yet — think of them as signposts guiding you toward the next stage.

Step 5. Unite groups into clusters

Once you’ve written down a few patterns, use them as beacons to bundle your first-level groups into clusters — adding another layer of hierarchy to your map. And again, aim for 2–5 groups per cluster.

• More than 5 groups in a cluster? You’ve likely merged too much stuff — try rearranging it.
• Fewer than 2 groups? That’s not a real cluster. It overemphasizes a single group and increases bias.

As your clusters take shape, name them using the same “I statement” format, making sure they summarize the essence of all included groups:

Cluster: “I need timely and accurate incident info to make good decisions”

Groups inside:
— “I often receive incomplete information about incidents”
— “I must know who was involved in an incident to follow up effectively”
— “I receive incomplete information, so my investigation takes longer”
— “I lack visual evidence, so I can’t confirm what actually happened”

Once you’ve named your clusters, step back and look at the whole map. Are there overlaps where two clusters could be merged? Are some clusters made up of notes from just one interview code? That might mean they’re outliers or hyper-specific.

Step 6. Create super-clusters

If your research produced so many notes that you now have a dozen or more clusters, you may need a third level of hierarchy. How to get there? Just like in Step 4, take a holistic look at your affinity map — but this time, focus on super-patterns.

Don’t get pulled back into groups or individual notes; at this stage, your lens should be global. This is how super-clusters will naturally emerge — broad categories capturing 2–5 related clusters.

Super-cluster: “People-centered approach to safety”

Clusters inside:
— “I adapt safety communication to match workers’ language”
— “I involve frontline staff in reviewing and improving safety procedures”
— “I share safety knowledge widely in the organization”

As a result, you’ll end up with a full hierarchy — from individual notes to concrete behaviors, mental models, and goals of the audience you studied. You may also notice gaps in your data, which can guide follow-up research. More often, you’ll see the initial research topic expand naturally, branching into related areas you hadn’t anticipated.

For example, if your research began with the goal of better understanding workflows behind workplace safety, you might end up uncovering not just that topic, but also related areas such as safety training, near-miss reporting, and cultural differences in safety practices.

Step 7. Use the affinity map

Seeing your completed affinity diagram is a relief — but it’s the start, not the finish. The map gives you organized findings, not automatically final insights or action plans.

Next, walk the map with your team: review key clusters and super-clusters, discuss what they imply, and brainstorm solutions that map to them.

Besides, you can turn affinity outcomes into additional deliverables. Your affinity map is likely to include all necessary ingredients already:

• Personas — bring your audience segments to life by summarizing their key workflows, tasks, goals, needs, and pain points.
• Journey maps — visualize how people move through processes or experiences, showing where challenges and opportunities lie.
• Service blueprints — layer the user’s journey with backstage processes to see where systems, tools, or teams need improvement.
• Workflow diagrams — visualize complex processes, especially when different roles or departments interact.

Further reading

Affinity mapping is a powerful way to distill customer experience research or solution brainstorming. Its value, however, depends on the quality of individual notes and how thoughtfully you group them. Done well, it drives alignment and decisions; done poorly, it becomes just a wall of sticky notes no one acts on.

Everything I’ve described here builds on work by more experienced UX practitioners. Here are some recommended reads:

• “Rapid Contextual Design,” a book by Karen Holtzblatt, Jessamyn Burns Wendell, and Shelley Wood.
• “Affinity Diagramming for Collaboratively Sorting UX Findings and Design Ideas,” an article by Rachel Krause and Kara Pernice for Nielsen Norman Group.
• “Avoiding 3 Common Pitfalls of Affinity Diagramming,” an article by Maddie Brown for Nielsen Norman Group.
• “Affinity Diagrams: How to Cluster Your Ideas and Reveal Insights,” an article by Rikke Friis Dam and Teo Yu Siang for IxDF.