Statistical persona

A statistical persona comes from analysing numbers, typically a survey answered by hundreds or thousands of people, and finding clusters of respondents who answer similarly. Each cluster becomes a persona. It is the quantitative counterpart of the qualitative persona.

What it adds

Scale and proportion. Where interviews tell you that some users manage their work in spreadsheets and ignore the product, a survey tells you how many, and whether they are concentrated in one region or one plan tier. That makes statistical personas the right basis for prioritisation: which segment is large enough, valuable enough or underserved enough to design for first.

The catch

The clusters are only as good as the questions. A survey can only reveal differences in things you thought to ask about. This is why the sensible order is qualitative first, quantitative second: interviews tell you which behaviours and attitudes matter, the survey measures them. Skip that order and you risk clusters that are statistically tidy and meaningless, such as groups defined by how they answered a rating scale rather than by anything they do.

In practice

The finished persona usually still carries a short narrative, borrowed from earlier interviews, because a table of percentages does not help a designer picture a person. The numbers go beside the story, not in place of it. See quantitative research for the wider method.

Common mistake

Reporting false precision. A segment that is 23.4 per cent of a convenience sample is not 23.4 per cent of your market.