Affinity Diagrams: Turning Messy Ideas Into Structure
A list of two hundred and seventy defects is not information. It is a record, and it will be worked through item by item by whoever is least able to refuse, at roughly the pace items were added. The same is true of a retrospective that produced sixty observations, or a requirements exercise that collected four hundred statements from thirty people. Volume defeats attention, and the response is usually to prioritise the top ten and quietly lose the rest.
Affinity grouping is the cheapest answer to that problem. Separate items are sorted by natural similarity, without argument, and the clusters that emerge get their names afterwards. What comes out is a small number of themes, each carrying its own evidence, and themes are something a project can act on. Section 5 of the PMBOK® Guide Eighth Edition lists it among the available techniques, and it needs nothing beyond a wall, some cards and a rule about silence.
How it is done
Every item goes on its own card, in the words of whoever raised it. Summarising at this stage destroys the thing that makes the technique work, because the grouping depends on detail that a summary removes. Two hundred cards feels absurd and takes twenty minutes to produce from an existing list.
The group sorts them without discussing them. People move cards next to other cards that feel related, silently, including cards somebody else has just moved. It looks chaotic for the first few minutes and then settles, and the silence is what prevents the exercise turning into a debate about categories.
The clusters get named last, by the group, once they have stopped moving. The name describes what is actually in the pile, which is frequently not what anybody would have predicted. Where a cluster resists naming, it usually contains two ideas and wants splitting.
Why the order matters
Naming first produces the categories you already had. Give a group four headings and they will file everything into them, including the items that fit none of them, and the exercise will confirm the view of whoever wrote the headings. The information sitting in the awkward items is lost precisely because they were awkward.
Discussing during the sort hands the outcome to the most confident voice. Sorting in silence means a junior engineer's sense that two items belong together carries the same weight as a director's, and it usually reflects closer knowledge of the work. The discussion still happens, once the patterns are visible and harder to talk away.
Two hundred and seventy-four snags
A refurbishment of two floors in a 1970s office building reached handover with a snag list of two hundred and seventy-four items. The contractor read it as a quality problem, the client read it as evidence of a poor job, and the project manager was being asked to work through it line by line, which would have taken most of a month.
Instead, the list went onto cards and four people sorted them in an hour: the surveyor, the facilities manager, the contractor's site manager and a joiner who had worked on both floors. Four clusters formed, and the largest by a considerable margin held damage to finished joinery and ironmongery.
The name the group gave that cluster changed the conversation: second-fix work damaged after it had been completed. The items were not defects in workmanship at all. They were the consequence of decorating and mechanical commissioning running after the joinery was in, on a programme that had been compressed at the front. The remedy was a sequencing change for the remaining floors and a protection specification, which prevented most of the second list entirely. Choosing the technique that fits the mess in front of you, and running it so the pattern comes from the evidence rather than from the room, is practical tool work Omega's PMP® Exam Preparation returns to across techniques.
Where else it earns its keep
Retrospective and survey feedback, where the same sentiment arrives in forty different wordings and a manual read produces whatever the reader already believed. Grouping shows how much of the feedback is really one issue, which is often the most useful number in the exercise.
Requirements collected from many people, particularly across departments that use different vocabulary for the same need. Clusters expose duplication and, more valuably, expose the places where two groups are using identical words for different things.
Lessons across a portfolio, where each project's report is reasonable on its own and the pattern only appears when several are cut up together. Organisations that do this once a year tend to find two or three structural issues they have been paying for repeatedly.
For a PMP® candidate, a scenario presenting volume rather than insight is worth spotting early. A description in which a team has gathered extensive feedback and cannot decide what to do with it is describing an analysis gap, and the useful response is a technique that reduces the volume to themes before any prioritisation is attempted.
The moment to reach for this is the first time somebody says there is too much of it to go through. An hour, a wall, the people closest to the work and a rule against talking will tell you what your two hundred items are actually about, and the answer is normally three or four things.
Interested in going further?
Turning a mass of observations into a small number of causes somebody can own is a skill that pays back on every project with more feedback than time. Omega's PMP® Exam Preparation covers the analytical techniques alongside the judgement of when each one fits.
The PMBOK® Guide Eighth Edition sets this technique among the others a project can draw on, which helps in choosing between them.
Ad · Amazon affiliate link.
References
A177: Five Whys vs Fishbone Diagram
A180: Facilitation Techniques for Difficult Project Conversations
A187: Retrospectives That Lead to Change
A178: SWOT Analysis in Project Management
A182: Multicriteria Decision Analysis Explained
PMP and PMBOK are registered marks of the Project Management Institute, Inc.
