HomeFrameworksDesign & UX › KJ Method / Affinity Diagram
// framework

KJ Method / Affinity Diagram

Jiro Kawakita, 1960s

A technique for clustering unstructured qualitative data into emergent themes — particularly powerful for mining customer reviews to find the unexpected patterns that reveal hidden positioning opportunities.

// description

The KJ Method is a technique for making sense of large amounts of unstructured qualitative data. Participants write individual observations, ideas, or data points on cards or sticky notes, then silently group them into natural clusters based on affinity. The clusters are labelled only after grouping, which prevents premature categorisation. The method surfaces patterns and themes that structured analysis might miss.

// history

Jiro Kawakita, a Japanese anthropologist, developed the method in the 1960s as a way to organise field research data from his expeditions in Nepal and other regions. He described it in his 1967 book Hassou-hou. The technique was adopted by Japanese industry within quality management circles and spread internationally under the name "affinity diagram" as part of the Seven Management and Planning Tools.

// example

A KDP publisher has collected 90 customer review fragments (copied from Amazon reviews across their niche). She writes each fragment on a sticky note and clusters them silently. Six natural clusters emerge: "easy to use daily," "helps me stay consistent," "wish it had more space," "love the design," "started me on a habit," and "bought as a gift." The gift cluster is unexpected and large — she hadn't marketed the product as a gift, but 20% of reviewers mention gifting. This reveals an entire untapped marketing angle she immediately adds to her listings.

// katharyne's take

This is one of the best tools for mining customer reviews — your own and your competitors'. Collect 50-100 review fragments from Amazon or Etsy, paste them into a doc or print them out, and cluster them by theme. The patterns you find in 30 minutes of clustering will tell you more about what your market actually values than any amount of market research tool output. Pay special attention to the clusters you didn't expect — those are your hidden positioning opportunities. ChatGPT can help with the initial clustering if you paste in the reviews and ask it to group by theme.

// creative uses
// quick actions
// prompt ideas
I'm going to paste [number] customer reviews from Amazon/Etsy for [your niche or product type]. Please group them into affinity clusters by the main theme each review expresses. Label each cluster, list how many reviews fall in it, and flag any cluster that surprises you — those are the hidden positioning opportunities I want to explore first.
Here are the questions and messages I've received from customers about [your digital product or KDP book] over the past [time period]. Cluster them by theme using the KJ Method logic — group by natural affinity, label clusters after grouping, and tell me which cluster is largest. I want to know whether this is a product problem or a listing clarity problem.
I have a list of [number] product or content ideas for [your niche]. Apply affinity clustering to group them by the underlying audience need they serve rather than by format or topic. Show me the clusters, identify which cluster has the most momentum, and flag any ideas that don't fit — those are candidates to cut.
See also: Card Sorting · Empathy Mapping
← Card Sorting Persona Development →