Skip to content
KatafactsBeta

Voice of the customer

Affinity diagram

Affinity diagram is licensed CC BY 4.0. Attribution: Katafacts (katafacts.com).

Customise with AISkill file ↓

1 · What it is

What it is

An affinity diagram groups a set of individually-true observations into themes — the classic version is sticky notes moved around a wall until patterns emerge. This does the same synthesis from a curated set of quotes and observations gathered ahead of time: a theme gets drafted for every note, then it's yours to reassign until the clustering feels right, with the tally updating live as you edit. Where a VOC evidence log is a running capture tool you add to over time, an affinity diagram is a discrete session on notes you've already pulled together — same underlying method, different moment in the process.

2 · When to use it

When to use it — and when not to

Use it when

  • You've already gathered a set of quotes or observations (from interviews, tickets, surveys, or gemba notes) and want to synthesize them into a small number of themes in one sitting.
  • You're preparing for or running a workshop and want a first-pass clustering to react to and adjust, rather than starting from a blank wall.
  • You want the vital-few themes named with a defensible tally, not just a gut sense of 'a few people mentioned pricing.'

Not when

  • You're capturing evidence as it comes in over time rather than synthesizing a curated set in one session — that's the VOC evidence log itself.
  • You only have two or three notes total — there's not enough material for a clustering exercise to reveal a real pattern.
  • You already know the themes and just want confirmation — genuine synthesis means letting the notes suggest the clusters, not the other way around.

3 · How to fill it in

How to fill it in

Scope
Which accounts, segment, or population does this curated note set cover, and what prompted pulling it together for a synthesis session?
Evidence entries
Enter each quote or observation you've already gathered, with its source, source type, and date. This is the curated set for the session, not everything you've ever collected.
Themes
Drafted from your notes, then genuinely yours to edit — reassign any note's theme during the session and the clustering updates immediately, the same way moving a sticky note to a different pile would.
Narrative
Drafted from the top themes — what the clustering, taken together, actually suggests.
Next steps
What will the team do with this synthesis, ranked by impact and effort, owned and dated? A next step not tied to a named theme is a coaching failure mode, not a real action.

4 · What good looks like

What good looks like

The example below is a workshop-style synthesis of nine at-risk-account notes into four themes: two tied for the largest cluster (pricing confusion and missing integrations), a third notable theme (slow support), and one single-note theme correctly left out of the vital-few list rather than forced into significance.

Same example, as a downloadable xlsx workbook.

Download .xlsx

VOC evidence log

Q1 at-risk accounts — affinity diagram

Dana Whitfield, Product Lead · 2026-03-12

Team: Marcus Lee, Support, Priya Nair, Customer Success · Sponsor: Alex Romero, VP Product

Scope

Covers accounts flagged at-risk in Q1 2026 — a curated set of notes pulled together for a single workshop session, not the running capture log itself. Four accounts, gathered across interviews, support tickets, a survey, a win-loss call, and one gemba session.

Evidence entries

Confusing pricing tiersTop theme

  • I still don't understand what separates the Growth plan from Scale — the pricing page doesn't make it obvious.

    Palisade Ventures · interview

  • We got moved to a higher tier automatically and nobody told us why.

    Thornbury Group · support-ticket

  • Our renewal quote didn't match what I thought we'd signed up for.

    Calder Industries · win-loss

Missing integrationsTop theme

  • We can't connect this to our data warehouse without a manual export step.

    Palisade Ventures · interview

  • No native Salesforce integration is a dealbreaker for our team.

    Meadowlane Co · survey

  • Our engineers built a workaround because the API doesn't support what we need.

    Thornbury Group · gemba

Slow support turnaroundTop theme

  • We waited four days for a response on a billing question.

    Calder Industries · support-ticket

  • Support tickets seem to sit longer than they used to.

    Meadowlane Co · survey

Reporting gaps

  • The dashboards don't roll up to what our finance team actually needs.

    Calder Industries · interview

Themes

Top themes: Confusing pricing tiers, Missing integrations, Slow support turnaround

  • Confusing pricing tiers3 · 33.3%
  • Missing integrations3 · 33.3%
  • Slow support turnaround2 · 22.2%
  • Reporting gaps1 · 11.1%

Narrative

Confusing pricing tiers and Missing integrations are the two largest clusters, tied at the top — one about not understanding what's included at each plan level, the other about specific technical gaps (data warehouse, Salesforce, a custom API workaround). Slow support turnaround is the third theme worth acting on, distinct from either. Reporting gaps has only one entry in this set — worth watching for a pattern in future sessions, not yet a vital-few finding on its own.

Next steps

  • Add an explicit plan-comparison table to the pricing page naming exactly what changes between tiers, and require a human confirmation step before any account is auto-moved to a higher tier.

    Linked finding: Confusing pricing tiers — tied for the largest cluster, three separate accounts

    Impact: high · Effort: medium · Owner: Dana Whitfield · Due: 2026-04-02

  • Scope a native Salesforce integration as the first of the missing-integration requests, since it's the one named as a dealbreaker rather than a workaround.

    Linked finding: Missing integrations — tied for the largest cluster, three separate accounts

    Impact: high · Effort: high · Owner: Marcus Lee · Due: 2026-05-01

5 · Common mistakes

Common mistakes

  • Running the synthesis with too few notes to reveal a real pattern.

    A handful of notes can't tell the difference between a real cluster and coincidence — affinity diagramming needs enough raw material for a theme to mean something beyond one person's opinion.

  • Forcing every note into a theme, including a single outlier note that doesn't fit anywhere.

    Same vital-few discipline as every other artifact in this catalogue — a genuine one-off observation is worth noting, not stretched into a cluster it doesn't belong to.

  • Treating the first draft of theme labels as final instead of adjusting them during the session.

    The whole value of this exercise is the team reacting to and correcting the first-pass clustering together — accepting the draft without discussion skips the actual synthesis.

6 · What it connects to

What it connects to

upstream

  • VOC evidence log

    The curated notes an affinity-diagramming session works from typically come from an ongoing evidence log — pull the relevant subset in rather than starting from nothing.

downstream

  • CTQ tree

    Once a theme is named with real weight behind it, a CTQ tree is where you'd translate it into a specific, measurable requirement.

7 · Where AI helps

Where AI helps

Judgement — stays yours

  • Deciding whether a drafted cluster is actually right, or moving a note to a different theme during the session
  • Deciding which theme is worth acting on first

Analysis — AI helps

  • Drafting a first-pass theme label per note from its actual content
  • Drafting the narrative from the top themes

Drudgery — automated

  • Tallying notes per theme and computing each theme's share of the total, with the arithmetic shown
  • Identifying the vital-few top themes
  • Exporting to xlsx in the house format

9 · Rate this kata

Rate this kata