How to Use an Affinity Diagram to Group Ideas

Published October 14, 2026 · Updated October 14, 2026 · 8 min read · by the Sakura5 team

To use an affinity diagram to group ideas, write every idea on its own note, spread the notes out, move them silently into clusters of things that feel related, and only then give each cluster a name that says what its notes have in common. The order is the whole method. You sort first and label last, so the groups come from the ideas instead of from the categories you walked in with.

That sounds obvious, and it’s exactly what most people get wrong. Put forty sticky notes in front of a team and someone will write “Pricing,” “UX” and “Marketing” on the whiteboard within a minute. From then on, every note gets forced into one of those three boxes, and anything that doesn’t fit quietly disappears.

This guide covers why that happens, a six-step workflow you can run alone or with a group, a worked example with real numbers, and what to do when your clusters won’t settle.

What an affinity diagram is, and why top-down sorting fails

An affinity diagram is a way of organising a large, messy pile of ideas, observations or quotes by grouping them according to natural similarity. It was developed by the Japanese anthropologist Jiro Kawakita in the 1960s, which is why you’ll also see it called the KJ method. Today it’s a standard tool in quality management, UX research and design thinking, and the Nielsen Norman Group has a clear overview of affinity diagramming for research if you want the formal version.

The reason it works is that it reverses the usual direction of sorting. Normally you decide the categories, then file items into them. That’s fine when the categories are well known, like sorting receipts by month. It fails when you’re trying to find out what the categories are, which is the situation you’re in after a brainstorm, a round of interviews or a stack of customer feedback.

Top-down sorting has three predictable failures:

• Categories confirm what you already believed. If you start with “Pricing” as a box, you’ll find pricing problems. You won’t notice that six notes are really about trust.
• Odd ideas get lost. A note that fits no existing box goes into “Other,” and “Other” is where good insights go to die.
• Groups are too broad to act on. A cluster called “UX” with fifteen notes in it tells you nothing. A cluster called “People can’t find the export button” tells you what to fix.

How to use an affinity diagram to group ideas in six steps

1. Get every idea onto its own note

One idea per note, written as a short phrase or full sentence rather than a single word. “Onboarding” is a topic; “I didn’t know I could import my old files” is an idea. Use sticky notes, index cards or a digital board — the medium matters less than the rule that each note can move independently. If you’re starting from a brain dump, our guide to doing a proper brain dump covers how to get everything out first.

2. Spread them out and read them all once

Lay every note out where you can see it. Read through the whole set before moving anything. This takes a few minutes and stops you from building your first cluster around whichever note you happened to read first.

3. Sort silently, in pairs first

Pick up a note, find another that feels related, and put them together. Then keep going. Don’t explain why, don’t debate, and don’t name anything yet. In a group, sorting in silence is the part that matters most: talking lets the most confident person define the groups, which brings you straight back to top-down sorting. If someone disagrees with where a note sits, they just move it. A note that gets moved back and forth repeatedly probably belongs in two places, so write a duplicate.

4. Leave the misfits alone

Some notes won’t fit anywhere. Leave them standing on their own rather than forcing them into the nearest cluster. Loners are often the most interesting notes in the pile, and they sometimes become the seed of a group once a second related idea turns up.

5. Name each cluster last, and name it as a statement

When movement has mostly stopped, read each cluster and write a header that says what its notes have in common. Make the header a sentence, not a noun. “Customers don’t trust the free trial” is a finding you can act on; “Trial” is a label. If you can’t write one sentence that covers every note in a cluster, it’s probably two clusters.

6. Group the groups, then decide

Look at your named clusters and see whether some belong together under a broader theme. Two levels is usually enough. Then do the step people skip: decide what each top-level theme means for you, whether that’s a fix, a question to research or a section of a report. An affinity diagram that ends without a decision is just a tidier pile.

A worked example: sorting 52 pieces of app feedback

Say you’re Priya, the only product person at a small team building a habit-tracking app. You’ve collected 52 pieces of feedback over a month: 31 App Store reviews, 14 support emails and 7 notes from two user calls. Your manager wants to know what to fix next quarter, and you have one afternoon.

First 20 minutes. You write each piece of feedback as a one-line note in the user’s own words where possible. Long emails become two or three notes, so 52 sources turn into 64 notes.

Next 30 minutes. You sort without naming anything. Pairs form fast: two notes about reminders arriving at the wrong time, three about losing a streak after a phone change. After half an hour you have 11 clusters and 6 loners.

Next 15 minutes. You name each cluster as a statement. A few examples: “Reminders ignore time zones when people travel” (7 notes), “Streaks are lost when people switch phones” (9 notes), “People want to pause a habit without breaking the streak” (8 notes). Two clusters can’t take a single sentence, so you split them, giving you 13.

Next 15 minutes. You group the groups. Three top-level themes emerge: Streaks feel fragile (4 clusters, 24 notes), Reminders feel dumb (3 clusters, 15 notes) and Setup is confusing (3 clusters, 13 notes). The remaining clusters are small feature requests.

Here’s the result top-down sorting would have hidden. If you had started with boxes like “Bugs,” “Features” and “Design,” the streak problems would have been split across all three: phone-change data loss under bugs, pausing under features, and the anxious tone of the streak counter under design. Grouped by affinity, they add up to the single biggest theme in the feedback, and that’s what goes to your manager.

If you’re doing this on your own rather than with a wall of sticky notes, Mindmap AI can speed up the first pass. Paste the 64 lines in as text and it builds an organised first draft of the map on an infinite canvas, with related lines already grouped under branches. Treat that draft the way you’d treat a colleague’s first attempt: drag notes between branches when the grouping is wrong, and rename every branch as a statement in your own words.

Affinity diagram or mind map: which one to use

The two get confused because both end up as grouped ideas. The difference is direction. A mind map usually grows outward from a known centre: you start with a topic and branch into what belongs under it. An affinity diagram grows upward from scattered pieces: you start with the pieces and discover the structure.

In practice they work best together. Use an affinity diagram when you have a pile of raw material and don’t yet know its shape. Once the clusters are named, turn them into a mind map: the main question goes in the centre, top-level themes become main branches and clusters become sub-branches. That map is then much easier to extend, share and turn into a plan. Our guide to organising ideas for a group project shows this hand-off in a team setting.

In Mindmap AI, outline view turns the finished map into a nested list you can paste into a report or a planning doc, which is often the format your manager or tutor actually wants.

When your clusters won’t settle

One cluster keeps growing. A group with more than about a dozen notes is usually hiding two or three ideas. Read it again and ask what the notes disagree about. That disagreement is where to split it.

Everything ends up in three big groups. You’re sorting by topic, not by meaning. Break each big group back into pairs and rebuild it, asking “what is this note actually saying?” rather than “what is it about?”

Too many loners. If more than a fifth of your notes stand alone, the notes may be too vague to compare. Rewrite the loners more specifically and try again.

The group can’t agree. Go back to silent sorting for five minutes. If a note still won’t settle, duplicate it into both clusters and move on. The goal is a useful picture, not a perfect taxonomy.

FAQ

What is an affinity diagram used for?

An affinity diagram is used to organise a large set of ideas, observations or feedback into groups based on natural similarity. It’s common after brainstorming sessions, user interviews, surveys and retrospectives, whenever you have lots of raw material and need to discover the main themes rather than impose categories you already had.

How many ideas do you need for an affinity diagram?

There’s no strict minimum, but the method earns its keep once you have more items than you can hold in your head, roughly twenty or more. With fewer, a simple list usually works. With a hundred or more, sort in batches and merge the resulting clusters at the end.

Should you name the groups before or after sorting?

After. Naming groups first turns the exercise into ordinary top-down filing, and you’ll mostly find the themes you expected. Sort the notes into clusters by feel, then write a header for each cluster that states what its notes have in common, ideally as a full sentence rather than a single word.

Can you make an affinity diagram on your own?

Yes. The silent-sorting rule exists to stop group dynamics from shaping the clusters, but the core method works just as well solo. Use sticky notes, index cards or a digital canvas, sort first and label last, and take a short break before naming clusters so you read them with fresh eyes.

← All articles