5 Common Mistakes When Creating User Personas | Pehloo
UX ResearchCommon MistakesBeginner

5 Common Mistakes When Creating User Personas

AUTHOR: Ribhu Lahiri||8 min read

TL;DR: Most user personas fail for five avoidable reasons — they describe demographics instead of behaviour, there are too many of them to prioritise against, they model an idealised user, nobody updates them, and they live somewhere the rest of the team never looks. Fix those five and the same research becomes a decision-making tool instead of a PDF.

Why do user personas fail?

Personas rarely fail because the research behind them is wrong. They fail because the document cannot be used at the moment a decision gets made, so the team falls back on whoever argues most confidently in the room.

The pattern has a name: the forgotten PDF. Someone spends a fortnight on interviews, produces a polished deck with stock photography, presents it once, and it is never opened again. Meanwhile the roadmap is settled by whoever has the loudest opinion about what users want.

A persona that earns its place does at least one of three things: it settles an argument about scope, it tells a recruiter who to book for interviews next week, or it stops a feature from being built. If yours does none of those, the problem is the format rather than the research.

Mistake 1: Are you describing demographics instead of behaviour?

Demographics are a filter; behaviour is the signal. Knowing that "Startup Steve" is 32, drives a Honda and owns a golden retriever tells an engineering team nothing about how to build a data export — which is the decision the persona is supposed to inform.

Behaviour-driven personas sound like this: checks dashboards in twenty-minute gaps between meetings, abandons any setup flow longer than two screens, and re-uses whatever answer a colleague gave last quarter. Every clause in that sentence changes a design decision. Age, car and dog do not.

Demographics still matter in exactly two places: recruiting (you have to find people who match) and market sizing (you have to know the segment is big enough). Keep them, but put them where they belong — in the recruitment block at the bottom, not in the headline of the persona.

Mistake 2: Do you have more than five personas?

Fifteen personas means zero personas. You cannot prioritise a roadmap against fifteen equal claims on the same sprint, so in practice the team ignores all of them and optimises for the biggest customer in the room.

The usual cause is that every interview produced a persona, so every edge case got promoted to a first-class user. The fix is a hierarchy: two or three primary personas that get a vote on scope, and a longer list of secondary personas that are documented but explicitly do not vote. Naming the hierarchy is what makes prioritisation possible — when two primaries disagree, you have a genuine product strategy question instead of a spreadsheet of everyone who might conceivably sign up.

Persona Forge is built around that split: it generates two primary personas plus supporting secondary personas, on the assumption that a set you can argue about is more useful than a set you can't.

Mistake 3: Did you describe the ideal user instead of a real one?

Teams write personas who are patient, rational, and delighted to spend an afternoon in onboarding. Real users are distracted, mildly resentful of the time your product asks for, and ready to use whatever they already know instead of learning something new.

The most valuable sentence in any persona is the one that explains why this person would abandon your product — what has to go wrong for them to close the tab and go back to the spreadsheet they were using before. If no such sentence exists, the persona is a wish list.

Write the friction in deliberately: what they are trying to finish before they got interrupted, which step they will skip, and what they will complain about to a colleague. A persona that is argued with is being used; a persona that is admired is decoration.

Mistake 4: When did you last update your personas?

Personas have a half-life. Once the product ships, the market shifts, or a competitor changes what "good" costs, an un-updated persona stops describing a user and starts describing your memory of one.

A stale persona is worse than no persona, because it still carries the authority of research. Someone in a planning meeting will quote it accurately and reach the wrong conclusion.

The cadence does not need to be heavy: re-read the set at each planning cycle and change the one field the last release invalidated. Keep the date visible on the persona, so nobody has to guess how old the assumption is.

Mistake 5: Can anyone outside the UX team use them?

If the only way to see a persona is to open a Figma file that lives in a designer's drafts, then for engineers, support and marketing the persona effectively does not exist.

The practical test: can a product manager quote a specific frustration from a persona in a sprint review without opening a file? Can support read the same sentence when a ticket contradicts it? Personas change decisions only when they are as cheap to reference as an opinion.

That means structured, searchable and shareable — a permanent link per persona, plain text you can paste into a doc, and an export you can attach to a brief without re-typing it.

What does a usable persona look like?

The difference between a persona that changes decisions and one that gathers dust is not the amount of research; it is whether each field is specific enough to argue with.

FieldWeakStrong
GoalWants to be more productiveNeeds the weekly report out before the 9am stand-up, on the first attempt
Demographics32, urban, degree-educatedOperations lead, 4-10 person team, no dedicated analyst
FrustrationDislikes manual workRe-exports the same CSV every Monday because the saved view loses a filter
BehaviourEarly adopterTries the free tier, abandons setup after two screens, returns if a colleague sends a link
EvidenceAssumed from the marketSix support tickets and two interviews in the last quarter
Recruitment"Ops people"Owns weekly reporting, has tried and dropped a comparable tool in the last year

How do you avoid all five without weeks of research?

You can remove the format failures before you have any research at all. The five mistakes above are structural: how many personas you have, what fields they carry, where they live and whether anyone maintains them. None of those require a research budget to get right — they require a starting set that is structured, editable and dated.

That is what Persona Forge is for. Describe your product, and it produces two primary personas plus secondary personas with behaviour, goals, frustrations, tech comfort and recruitment criteria already in the right shape. They are explicitly hypothetical assumptions — a brief for the interviews you are about to run, not a replacement for them — which is precisely why they are worth arguing with. You can edit any field, question a persona directly in a chat to find the weak assumption, and export the set as a PDF when a brief needs one.

If you want the method rather than the tool, read how to create data-driven user personas for the process that works when you have data, or what user personas are and why they matter if you are starting from scratch. And if your buyer is a different person from your user, the difference between user and buyer personas is usually the fifth mistake in disguise. Plans and limits are on the pricing page.

Frequently asked questions

How many user personas should a product have?

Two or three primary personas work for most products, with as many secondary personas as you need documented but excluded from scope decisions. The limit is not a research finding; it is the number of competing goals a team can hold in mind while prioritising.

Are AI-generated personas accurate?

They are hypotheses, not findings. A generated persona is a structured guess about who would use a product and why, which makes it a good recruiting brief and a good starting argument — but it is validated by interviews, not by the model that wrote it.

How often should personas be updated?

At least once per planning cycle. Change the specific field that the last release invalidated, and keep the date visible so nobody has to guess how old the assumption is.

What is the difference between a user persona and a buyer persona?

A user persona describes the person who uses the product; a buyer persona describes the person who chooses and pays for it. Selling to the buyer while building for the user is the whole point of keeping them separate.

Can a persona be too detailed?

Yes. Detail that does not change a decision is noise: it makes the persona expensive to read and easy to dismiss. Every field should be able to answer "what would we build differently if this were false?"

Stop Guessing. Start Building.

Don't let bad assumptions ruin good products. Generate perfectly calibrated, data-driven user personas in 10 minutes.