Affinity Diagrams
Affinity Diagrams take a pile of loose ideas or notes, written one per card, and group them by what actually belongs together, so the underlying themes surface instead of staying buried in a long list.
Cards move left to right across five steps, from a scattered pile to labelled clusters.
Reach for this when…
- A brainstorm produced eighty sticky notes and nobody can see the shape of them.
- Interview or survey feedback is a wall of quotes with no structure.
- Two people keep describing the same issue in different words and the team hasn't noticed.
How to run it
- Write each idea, issue or data point on its own card - one thought per card.
- Spread every card out where the group can see all of them at once.
- In silence, move cards next to others they seem to belong with.
- Once clusters settle, name each group with a short header that describes it.
- Stand back and look at which groups are large, which are thin, and what's missing.
A worked example
Situation. Amina Chebet ran a garment export business in Nairobi, Kenya, and had sixty pieces of buyer feedback sitting in a spreadsheet nobody read.
Applied. She printed each comment onto a card and had her merchandising team sort them in silence for twenty minutes. Comments about slow sample turnaround, wrong trims, and late size-run confirmations kept landing in the same pile.
Result. The pile was not three problems, it was one: the sampling handoff between design and production. Fixing that single handoff addressed most of the sixty complaints at once.
The catch
The method surfaces groupings, not causes or priorities - a big cluster isn't automatically the most important one. It also depends on who's in the room: a differently staffed group will cluster the same cards differently. Follow it with a step that ranks or tests the groups, don't stop at the wall.
If one person is doing the grouping instead of the team, you've made a list with extra steps, not an affinity diagram.
Origin: Kawakita Jiro (KJ Method)