01 · Definition and selection
Definition and selection
An affinity diagram groups observations, ideas, or concerns by meaningful relationships. Its purpose is to make a large, untidy collection understandable without imposing categories before people have examined the material. It is useful after interviews, observation, or brainstorming, especially when several functions use different words for the same operational issue. The output is a set of themes and the underlying cards, rather than a ranked action plan.
Select it when the team has enough material to lose track of patterns but does not yet have a reliable classification scheme. If there are already well-defined numerical categories and a frequency question, a tally or Pareto analysis is more suitable. The prerequisite is a clear scope and cards that contain one understandable thought each. An affinity diagram can structure hypotheses and design opportunities, but similarity between comments does not establish a common root cause.
02 · Prepare evidence without erasing meaning
Prepare evidence without erasing meaning
Fictional Northbank Dispatch collects 36 comments during a week of carrier staging observations. Examples include “the lane number changes after the label prints,” “night shift cannot see the update board,” and “a damaged label means a walk to the office.” Each comment keeps its source role and observation date in a separate reference log. Public cards omit personal names so criticism of a process does not become criticism of a colleague.
The facilitator separates factual observations from proposals by marking their type in text. “Three cartons were moved twice” is different from “we should add a screen.” Keeping both is reasonable, but collapsing them into one card would hide whether the group is organizing evidence or solutions. Before grouping, the team clarifies ambiguous wording without rewriting it into a preferred explanation. A lone but important concern must remain visible even if it never joins a large cluster.
03 · Grouping mechanics and rationale
Grouping mechanics and rationale
The team reads the cards and moves related items together, initially without lengthy debate. This reduces the chance that the first confident speaker dictates the categories. When a card seems to fit two themes, the team can duplicate it with a visible cross-reference rather than forcing a false choice. Once groups settle, members explain the relationship they see and write a heading that captures it.
“Information arrives after movement” is a better heading than “systems problems” because it describes the shared pattern. “Labels fail at the point of use” is more informative than “printing.” The team checks every heading against the original cards and splits clusters that combine unrelated mechanisms. A final discussion distinguishes statements about carrier identity, visibility, physical handling, and collection timing. The rationale for grouping is documented briefly so another shift can understand the diagram without needing the original facilitator.
04 · Worked example and resulting decisions
Worked example and resulting decisions
Northbank’s 36 cards become four themes: destination visibility has 12 cards, label availability has 9, staging layout has 10, and release timing has 5. The counts sum to 36 because this first version places each original card in one primary cluster. Cross-references appear as references, not additional cards in the count. These are comment counts. Twelve comments do not mean destination visibility causes one third of late dispatches; some operators contributed several comments about one event.
The visibility group contains both late assignment and poor night-shift visibility. The team splits its next investigation into assignment timing and information access. The layout cluster produces a separate walking-route observation. Release timing is smaller but involves collection cutoffs, so the transport coordinator investigates it rather than discarding it. The diagram therefore creates four investigation streams instead of one premature project called “fix warehouse communication.”
05 · Use the structure responsibly
Use the structure responsibly
The team connects every theme to a question, an owner, and the evidence needed. For destination visibility, it needs timestamps for first assignment, label printing, staging, and any reassignment. For label availability, it needs failures by printer and label stock. For layout, it needs travel observations and congestion times. For release timing, it needs order-release and carrier-departure records. This conversion from themes to investigation makes the diagram useful beyond the meeting.
The facilitator preserves the original diagram and records a second version after more evidence arrives. If a later finding shows that printer damage and unclear assignments share a software retry problem, the team can regroup the themes. The original arrangement is a snapshot of understanding, not an immutable taxonomy. The most useful diagram makes uncertainty visible and allows a reader to trace a heading back to the comments that generated it.
06 · Weak versus strong practice and adaptation
Weak versus strong practice and adaptation
Weak practice starts with headings written by management and asks operators to fit their comments underneath them. That can be useful for a predefined audit, but it loses the exploratory benefit of affinity grouping. Other weaknesses include vague cards, merging dissent into a majority theme, counting repeated comments as independent events, and moving directly from a large cluster to an expensive solution.
Strong practice lets themes emerge, retains unusual concerns, and separates pattern recognition from prioritization. For advanced use, keep parallel views: one arrangement by customer consequence and another by information or material mechanism. The same cards may reveal different design choices in each view. For remote teams, use stable card identifiers and an exportable record; otherwise a beautiful digital board can lose its evidential trail. An affinity diagram becomes especially valuable when it informs subsequent cause investigation, must criteria, and a deliberate pilot rather than substituting for those steps.
07 · Operating conditions and evidence
Operating conditions and evidence
The diagram’s traceability is especially important when source material includes repeated observations. Northbank assigns each original comment a stable identifier. If the night-shift lead repeats an issue already raised by a dispatcher, both comments can remain, but the team does not call them two separate incidents without checking. If a comment contains two ideas, the facilitator splits it into linked cards and records that split so counts do not silently change meaning.
Headings should remain useful when the people in the room change. “Late information creates repeat movement” communicates an operational relationship; “Dave’s issue” does not. The team reviews headings with a shift that did not attend the workshop, asking whether the cards and headings accurately describe its experience. If that shift identifies missing conditions, the diagram gains new evidence rather than merely receiving endorsement. This makes the affinity structure a bridge between frontline language and subsequent process investigation.
08 · Decision record and next step
Decision record and next step
Northbank closes the affinity session by checking that the diagram can be reconstructed. Original cards, changed wording, cross-references, and the meaning of each heading remain available. A later investigator can find the source behind “information arrives after movement” without relying on memory. This matters when a theme influences a capital proposal or a systems requirement: the organization needs to know whether the claim began as observation, opinion, or a proposed fix.
The team also separates coverage from agreement. A theme that contains many comments may describe a widespread concern, but it may also reflect who attended and what they were asked. A theme with few comments can still expose a serious night-shift gap. Before prioritizing, Northbank compares the diagram with the operating conditions covered by its observation plan. Missing shifts or carriers are treated as evidence gaps. The diagram therefore guides the next inquiry and does not pretend to represent conditions nobody examined.
| Theme | Follow-up evidence | Responsible investigation |
|---|---|---|
| Destination visibility | Assignment and movement timestamps | Systems analyst with dispatch operators |
| Label availability | Recovery events and printer conditions | Dispatch lead with maintenance support |
| Release timing | Order release and collection records | Transport coordinator with release owner |
Connections
Related methods and phase use
Placed under Improve because of the outcome it supports. Use the method in other phases when its purpose and assumptions fit the work; the phase list is a guide to selection, not a compulsory sequence.
Sources
Professional foundations and further reading
The explanation and fictional teaching examples are written for LeanTrainingLab. Linked primary sources provide professional foundations; example values and local conventions are stated in the guide. The accepted toolset draws on Six Sigma+Lean Toolset, second edition (Lunau, editor; Springer, 2013). No book text or examples are reproduced here.