Affinity Diagram: The Sorting Powerhouse That Turns Team Chaos Into Clear Priorities

When a process improvement team collects dozens: or hundreds: of comments, observations, complaints, and possible causes, the challenge is rarely a shortage of ideas. The real challenge is making sense of them.

An Affinity Diagram converts a large volume of qualitative information into meaningful categories based on natural relationships. It helps a team see patterns that are difficult to identify when ideas remain as an unstructured list.

In the realm of Lean Six Sigma, this makes the Affinity Diagram a valuable bridge between creative thinking and disciplined analysis. It supports the Define, Measure, Analyse, and Improve phases of DMAIC, helping teams organise the Voice of the Customer, potential root causes, measurement concerns, or solution ideas before deciding what deserves deeper investigation.

What Is an Affinity Diagram?

An Affinity Diagram is a visual method for grouping ideas, facts, opinions, or observations according to their natural relationships.

The fundamental purpose is not simply to place similar words next to each other. It is to reveal the underlying structure within a complex set of information.

For example, a delivery improvement team may collect these observations:

• Incorrect customer addresses
• Late warehouse picking
• Carrier capacity constraints
• Incomplete packing instructions
• Delayed order entry
• Driver shortages
• Poor rush-order prioritisation
• Long loading times

Individually, these statements point in different directions. Once grouped, however, they may reveal four broader themes:

1. Carrier and logistics constraints
2. Customer and order data
3. Warehouse operations
4. Information technology and prioritisation rules

Those categories give the team a clearer basis for selecting measures, investigating causes, and defining improvement priorities.

Affinity Diagrams are sometimes associated with the KJ Method, named after Japanese anthropologist Jiro Kawakita. The approach is particularly useful when information is primarily qualitative and cannot yet be analysed effectively using statistical tools.

Where the Affinity Diagram Fits in DMAIC

An Affinity Diagram can support several stages of a Lean Six Sigma project.

Define

During Define, teams can organise:

• Voice of the Customer comments
• Stakeholder expectations
• Project concerns
• Customer complaints
• Initial observations about the process

This helps transform broad feedback into potential Critical to Quality (CTQ) requirements and sharper project boundaries.

Measure

During Measure, an Affinity Diagram can organise:

• Data collection concerns
• Inconsistent operational definitions
• Measurement system risks
• Sources of missing or unreliable data
• Different interpretations of the process

The result can inform a stronger data collection plan and reduce ambiguity before baseline performance is calculated.

Analyse

During Analyse, teams often use the tool to group potential causes of defects, delay, rework, or variation. These categories can then feed into:

• Fishbone diagrams
• Cause-and-effect matrices
• Pareto analysis
• Five Whys
• Hypothesis testing
• Regression or ANOVA

The Affinity Diagram does not prove a root cause. Instead, it helps the team decide which potential causes should be measured and tested.

Improve

During Improve, solution ideas can be grouped into themes such as:

• Technology changes
• Standard work
• Training
• Layout and flow
• Supplier improvements
• Policy changes

This structure makes it easier to compare solution packages using an impact-versus-effort matrix or a Pugh decision matrix.

The Seven-Step Affinity Diagram Method

A disciplined process protects the team from premature judgement and allows natural relationships to emerge.

1. Define a clear focus question

Begin with a specific prompt. Strong examples include:

• “What contributes to late customer deliveries?”
• “What causes extended patient waiting time?”
• “What prevents first-pass approval of invoices?”
• “What customer needs must our service reliably meet?”

A vague question produces vague data. The prompt should relate directly to the project problem statement or improvement objective.

2. Generate and capture the ideas

Ask participants to provide observations, facts, or ideas. Write one idea per card or sticky note.

Each note should be understandable without an explanation from the person who created it. Avoid combining several issues into one statement because that makes later classification less precise.

For example:

• “Orders are released after the carrier cut-off”
• “Packing labels are printed in batches”
• “Address fields are not validated”

These are more useful than a single note stating “Poor order process.”

3. Display the information randomly

Place all notes on a wall, table, or digital collaboration board without pre-sorting them.

This step matters because pre-existing categories can influence how participants interpret the data. Random placement encourages people to respond to the content itself rather than to departmental assumptions.

4. Sort silently into natural relationships

Participants now review the notes and move related items into groups. During the initial sort, silence is valuable.

A silent sort reduces the influence of hierarchy, confidence, and persuasive personalities. Each participant can interpret the information independently and identify relationships that may otherwise be overlooked.

Use the question:

“Is this note similar enough to belong with that group, or is it meaningfully different?”

If a note reasonably belongs in two groups, duplicate it rather than forcing an artificial choice.

5. Create header cards

Once the clusters begin to stabilise, write a concise header for each group.

A strong header captures the theme without prescribing the solution. For example:

• Warehouse release timing
• Customer information quality
• Carrier capacity
• System prioritisation logic

Avoid headers that are too broad, such as “People” or “Technology.” The header should help the team understand what the cluster means and why it matters.

6. Review and refine the clusters

Invite the team to examine whether:

• Every note belongs in a logical group
• Any group contains unrelated ideas
• Two groups should be combined
• One group should be separated
• The headers accurately describe the content
• The categories relate to the project scope

This is the point at which discussion becomes useful. The team should seek shared understanding, not force artificial consensus.

7. Link the categories to DMAIC actions

Finally, convert the diagram into a project decision tool.

For each cluster, identify:

• What evidence is already available?
• What data must be collected?
• Which process step is involved?
• Is the factor controllable?
• Which customer or business requirement does it affect?
• What tool should be used next?

The completed diagram should lead to action, such as a stratified data collection plan, a cause-and-effect matrix, or a focused root-cause investigation.

Worked Case Study: Reducing Late Deliveries

Consider a hypothetical e-commerce operation processing 12,000 customer orders per month. Its baseline late-delivery rate is 18.4%, meaning approximately 2,208 orders miss the promised delivery date each month.

The project goal is to reduce late deliveries to below 10% within four months, while maintaining order accuracy above 99%.

Define: Collecting the initial ideas

A cross-functional team of eight participants from customer service, warehouse operations, transport, IT, and planning generates 48 observations about late deliveries.

Examples include:

• Orders released after the daily carrier cut-off
• Driver shortages during peak periods
• Incorrect customer addresses
• Rush orders not clearly identified
• Packing documents completed manually
• Warehouse pick lists printed in batches
• Long loading times at the dispatch dock
• System downtime during order entry
• Carrier capacity constraints
• Customers unavailable at the first delivery attempt

After a silent sort, the 48 notes form four natural clusters:

The Affinity Diagram immediately improves the team’s focus. Instead of investigating 48 separate ideas, the team can structure its next analysis around four broader areas.

Measure: Connecting categories to data

The team then creates operational definitions and collects six weeks of data covering 18,000 orders.

The results show:

• Warehouse release delays: 1,026 orders
• Customer or order information issues: 438 orders
• Carrier-related delays: 684 orders
• IT and prioritisation issues: 522 orders

Because some orders involve more than one contributing condition, the categories are not treated as mutually exclusive defect counts. The team uses the Affinity Diagram as a classification framework, then applies process data to understand frequency and interaction.

Further stratification reveals that 71% of warehouse-related late orders occur when orders are released after 2:00 p.m. It also shows that rush orders without a visible priority flag are 3.2 times more likely to miss the carrier cut-off.

Analyse: Testing the priorities

The team does not assume that the largest cluster is automatically the root cause. Instead, it tests the most important relationships.

The analysis includes:

• A Pareto chart of delay categories
• A time-series review of order release times
• A comparison of late-delivery rates for prioritised and non-prioritised orders
• A process walk through warehouse release and loading
• A review of system downtime and order-entry records

The evidence validates two key contributors:

1. Orders released after the carrier cut-off have a late-delivery rate of 34.8%, compared with 8.6% for orders released before the cut-off.
2. Rush orders without a system priority flag have a late-delivery rate of 29.4%, compared with 9.1% for correctly flagged rush orders.

The Affinity Diagram did not establish these relationships by itself. Its value was in organising the original qualitative information so the team could select focused, testable hypotheses.

Improve and Control: Converting insight into results

The team pilots three changes:

• Automatic priority flags for rush orders
• A visual dispatch cut-off board at the warehouse
• A twice-daily release review between planning and dispatch

During the four-week pilot, the late-delivery rate falls from 18.4% to 7.1%, a relative reduction of approximately 61%. Monthly late deliveries decline from an estimated 2,208 to 852, based on the same 12,000-order volume.

The team also tracks supporting measures:

• Order release after cut-off: 22.0% to 6.4%
• Rush-order prioritisation accuracy: 78% to 98.6%
• Dispatch loading time: 42 minutes to 29 minutes
• Customer order accuracy: 98.7% to 99.3%

The new process is documented in a control plan, with daily review of cut-off adherence and weekly monitoring of late-delivery performance.

Practical Rules for Better Affinity Diagrams

Use these rules to improve the quality of your next session:

• Keep the focus question specific.
• Use one idea per note.
• Separate sorting from evaluation.
• Allow duplicate notes when relationships overlap.
• Name clusters after the pattern becomes visible.
• Use descriptive, neutral headers.
• Link every important cluster to a measure or DMAIC decision.
• Treat the diagram as a starting point for analysis, not statistical proof.

For project teams that need a structured record of this work, the Lean Six Sigma Project Storyboard Toolkit includes an Affinity Diagram within its Define-phase sequence, alongside the project charter, SIPOC, Voice of the Customer, CTQ tree, and stakeholder analysis.

Turn Organised Ideas Into Measurable Improvement

An Affinity Diagram gives a team a practical way to move from unstructured information to shared understanding. It respects qualitative insight while preparing that insight for quantitative investigation.

To fully appreciate its value, remember the sequence:

1. Capture the information.
2. Sort by natural relationships.
3. Name the emerging themes.
4. Connect those themes to data.
5. Test the most important causes.
6. Implement and control the resulting improvement.

That is the discipline of DMAIC in action.

If you want to build the broader capability to lead structured improvement projects, explore Lean Six Sigma Green Belt Online Training. The CSSC-accredited, self-paced course covers practical tools across project definition, data collection, root-cause identification, hypothesis testing, solution selection, piloting, and control.

Enrol in Lean Six Sigma training or pursue your next professional certification today, and learn how to turn team insight into validated process improvement.

Kaizen. Kai-Care. Kai-Done. ( Lean Six Sigma)