Direct answer
Personas improve marketing only when they change a decision: who to turn away, what to say, where to show up. They do when built from recent real buyers and lost deals, and fail when built from invented demographics in a workshop, kept to ten profiles or more and never consulted again. A good test: take a decision from last quarter and ask whether the persona would have changed it. If not, it is decoration.
Buyer personas are standard equipment. A workshop, a laminated profile with a first name and a stock photo, a slide in the strategy deck. Whether they improve marketing is rarely asked, because asking would mean admitting what the workshop cost.
The honest answer is that they sometimes do and often don't, and the difference is easy to see from outside. A persona earns its place by changing a decision. If it has not changed one, it was a workshop, whatever it cost.
When personas work
Personas work when three things are true. They are built from real, recent buyers, not from a room of people guessing what customers think. They are few: one or two, because a business that has ten primary buyers has not decided who it is for. And they are used, which means someone can point to a decision they changed: a segment dropped, an offer reworded, a channel stopped.
The evidence base is plain. Interview six to ten customers who bought in the last year and a few who nearly did and didn't. Ask what triggered the search, who else was involved, what else they considered and what almost stopped them. Write each up as a buyer record. A pattern across them is a persona with evidence behind it, and it tells you things a workshop cannot, such as which objections actually surfaced and who the real approver was.
When they are decoration
| Persona that works | Persona that is decoration |
|---|---|
| Built from recent real buyers and lost deals | Built in a room from what the team believes |
| One or two, with a clear reason for each | Eight or ten, so nobody is excluded |
| Describes the trigger, the room and the doubt | Describes age, hobbies and a favourite podcast |
| Says who the business will not serve | Describes everyone who might conceivably buy |
| Has changed a named decision | Has not been opened since the workshop |
The decorative persona has a recognisable look. A first name, an alliterative job title, a photo and a biography invented to make the profile feel human. It is a pleasant way to avoid the harder question of who the business is not for. A profile that excludes no one is a description of the market, not a decision about it.
A persona that excludes no one is a description of the market, not a decision about it.
The serious objection
The strongest case against personas comes from marketing science, and it deserves a fair hearing. Professor John Dawes of the Ehrenberg-Bass Institute coined the 95:5 rule for business markets: only about 5% of target customers are actively looking to buy at any given time, and the other 95% are not yet in the market. That statistic is specific to business-to-business buying, but the logic holds more widely where buyers think carefully before they buy.
The implication is awkward for a lot of persona work. A persona built from people who bought describes the buyer at the moment of purchase, the small slice who happened to be in the market. Marketing aimed only at that slice misses the larger group who will buy later and who need to know you exist and what you stand for before that day arrives. So a persona is a poor guide to reach. It is a good guide to message, because it tells you what a buyer needed to hear when it finally mattered.
A test for your own personas
Working test, from experience — not a published study: take one real decision from last quarter, such as a campaign brief, a homepage headline or a segment you chose to chase, and ask whether the persona would have changed it. If you cannot find a single decision that it did change, or would have, the persona is not working. It may still be accurate. It simply is not being used.
Personas also mask a disagreement. When sales and marketing describe a different customer, a shared persona can paper over the gap for a quarter while the actual disagreement sits unresolved. Building personas from buyer records puts the disagreement on the table, where it can be settled with evidence rather than by whoever speaks last.
Where a persona is a deliverable rather than a decision, skip it and write the audience decision instead: who the business is for, who it is not for, and how you know. If you are not sure that decision exists, the Free 5-Question Check is a quick way to find out. The wider case for deciding before spending is in Reports, Not Revenue.
- Pull six to ten buyers from the last year, plus a few lost or near-lost deals.
- For each, write down the trigger, who was in the room, the alternatives considered and what almost stopped it.
- Look for the pattern across them, and write one or two personas from that pattern only.
- Add a line for who the business will not serve, and why.
- Name one decision the persona changed. If you cannot, treat it as decoration.
