Personas describe who, not what they will do

A persona is a composite portrait of a customer group: their unmet needs, goals, knowledge, and likely behaviors. Used well, it keeps a team focused on who they serve. But a persona is a summary, not a simulator. It cannot tell you what a customer thinks at step three of your checkout, what they feel when a process stalls, or what they actually do when something goes wrong. Teams that walk personas through journey maps in workshops, assigning thoughts and feelings by committee, are writing fiction and calling it insight.

Why it matters to the business

Invented behavior misallocates real money: features get funded, flows get redesigned, and journeys get validated against evidence nobody collected. Even real customers cannot reliably report their own actions: a behaviorally verified sampling study by 84.51 found 75% of respondents misstated their purchase behavior compared with their actual loyalty-card data. If actual customers get their own behavior wrong, a composite invented in a workshop has no chance. The honest alternative is cheap: Jakob Nielsen's Nielsen Norman Group research shows five users uncover about 85% of usability problems in qualitative testing.

How to use it

• Use personas to frame the audience and the problem space: who these customers are, what they need, what they already know.
• When someone asks what the persona would do at a given step, log it as a research question and schedule observation.
• Replace persona role-play workshops with small fast studies: five users per round, then test, fix, and retest.
• Validate every likely-behavior claim in a persona against behavioral data: analytics, support intents, purchase records.
• Put a review date on each persona. A stale persona is quietly fictional too.

Where teams get it wrong

The role-play workshop feels like research: sticky notes, empathy, a persona named Sarah walked through onboarding while the room assigns her emotions. The output looks like data and gets pasted into journey maps and decks, where it acquires false authority. Bias goes in, invented behavior comes out, and decisions borrow confidence from a document that never met a customer.

Ask your team

• Which claims in our journey maps came from watching customers, and which from role-playing personas?
• When did we last check a persona's likely behaviors against actual behavioral data?
• If I deleted our personas tomorrow, what observed evidence about customer behavior would remain?

Simulation belongs to research, not invention.