• During a workshop I asked 4 SMEs to come up with 4 proto-personas.
  • Each SME came up with a different breakdown of demographic, paint-points and needs (so there was a total of 16 personas).
  • Each persona for each SME shared traits from all other personas, but there was no clear similarity between personas.
  • We affinity diagramed all of their paint points and needs and ended up with 4 clusters.

Provided that we're going to verify our assumptions about the proto-personas are valid after we land on a solution:


  • Would it make sense to make a persona based on each of these clusters, and make them poain-point/need dependent? (i.e. time-constrained Joe or budget-conscious Jake).
  • What are the pros/cons of doing it like this?
  • How could you alternatively build a 3-4 persona set from the 16 divergent personas brainstormed before?

2 Answers 2


To your initial question, "Is it ok to base your personas in affinity diagrams?" the answer is yes. Affinity diagrams are all about grouping information together in clusters, so go for it.

Your other questions:

  1. I would label each cluster by persona. Then add sub labels for pain-point and needs. Keep it simple.
  2. Pros are that you'll quickly see some overlap with each persona that should help prioritise what design features will have the greatest reach. Cons - if you're doing this with paper - this is a big job with 16 personas.
  3. Start by creating a affinity diagram 'cluster' for each of the 16 personas. Then, with the SMEs, start to condense these down into 4 personas. As great as it is to have data on 16 - albeit fictional - personas, you'll be better off building your 'thing' if you only have to focus on 4 personas. You can then present the personas back to your SMEs as a user dossier card, similar to the picture example here: https://www.nngroup.com/articles/persona/.

Also, a fairly new alternative to personas is the Jobs-to-be-done approach: https://www.nngroup.com/articles/personas-jobs-be-done/. The affinity diagram is just one step in your design process, if it helps you understand - and simplify - some accurate data then go for it.


No. Personas should be based on the results of user research, not on requirements gathered from SMEs. They should capture the needs and attitudes of actual users. In my experience, clients, project owners, marketing departments, etc, know what they want users to do and want, but don't know what users actually do and want. And that is what personas need to model.

Here's what Nielsen Norman Group says:

Ideally, the persona-creation process should be a part of the research phase for a product or feature, before the actual design process starts. Field studies, surveys, longitudinal studies, interviews, and other methods of user research should be conducted first to define characteristics of typical users. (Keep in mind that any self-reported data, such as that resulting from focus groups and surveys, is possibly misleading and should be verified through other methods.)

And usability.gov:

Personas should be based on qualitative and some quantitative user research and web analytics. Remember, your personas are only as good as the research behind them.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.