• Transform data into themes.


• Discover hidden patterns.


• Build shared understanding.


• Move from observations to insights.

• Organized affinity diagram.


• Clustered themes.


• Foundation for synthesis.

Data quality is key. Vague notes lead to meaningless groupings. Push for specific observations. Each note should be understandable without context. Good: "Clicked back button 3 times trying to undo form entry." Bad: "Navigation confusion."

The silent sorting phase is critical. Talking shapes the groupings and you lose diversity. Redirect firmly: "Let's keep working silently."

If two people keep moving the same note, it's revealing something. Set it aside. Split it if it contains two distinct ideas.

Groups of 1-2 items are orphans. Groups of 3-7 are ideal. Groups of 8-12 are manageable. Groups of 15+ need splitting. A giant "misc" group means your other groups are too narrow.

Resist creating groups based on existing frameworks. Let categories emerge from the data.

Bad headers: "Problems," "Opportunities," "Feedback." Good headers: "Users expect immediate confirmation of actions."

If nothing is clustering after 20 minutes, your data might be too diverse or homogeneous. Stop and reassess.

Physical is better. But digital works. Use a large canvas. Make notes moveable. Don't create pre-made buckets.