Retail media buyer personas as evidence cards, with the stereotype fields removed and a worked example

Who should get a card?

Give a separate card to each evidenced decision role, not to a made-up individual. Retailer commercial buyers, advertiser decision-makers, agency operators, and platform or data users can face different tasks, so one blended profile would hide which role supplied each finding.

Keep summaries for consumers and retail staff apart from buyer cards: they encounter the advert or the working process, but they are not the media buyers described here. The person approving key metrics also needs a distinct record, because judging reports is a different responsibility from purchasing placements.

Which decision should the card describe?

Describe one defined decision or work event, such as accepting, rejecting, revising, or operating an exchange. State whether the setting is on-site sponsored products, off-site activation, or in-store screens; evidence about one setting must not silently stand in for another.

Do not carry a finding from retailer staff across to advertiser demand without separately collected advertiser evidence. Define the eligible group and sampling frame before turning an observation into a role summary, and record them in the planning template; the audience research work is relevant when those populations need to be kept apart.

What belongs in the record?

Leave a field empty when no source supports it. For each entry, distinguish the participant's report, what a researcher directly observed, and what an analyst inferred; record the source identifier, fieldwork date, analyst, and question version so a later reviewer can retrace the entry.

Record the sample, dates, method, and source IDs rather than giving a confidence score without explanation. Where relevant, state the eligible population identified in market size work, and keep a retailer-specific claim attached to the retailer sample that supports it.

What a completed card looks like

The card below is invented to show how the fields work. It contains no research findings, and its sources are labelled as examples. Any real card replaces each line with its own evidence.

How should uncertainty appear?

Label every item as observed, reported by a participant, or inferred by the analyst. Mark inferences as weaker than direct observations, and show contrary cases rather than forcing the evidence into a neat, single story. A card is a working hypothesis, not a portrait of everyone who has the role.

Founders' and sales teams' views can suggest questions, but they remain assumptions until user research tests them. Opinions that do not come from users are assumptions to prove through research, as the service manual says about learning user needs.

Show the eligible population, sampling frame, invited, responding, and analysed counts, field dates, missing answers, and conflicts on the card. This makes the boundary of the evidence visible and prevents a pattern in one group from being presented as a representative account of buyers more widely.

Which personal details should be excluded?

Do not infer buying authority from seniority, technical ability from device choice, company size from postcode, or access needs from appearance. These fields can make unsupported guesses look factual; leave them out unless an evidence-based research purpose requires a particular detail.

Collect personal information only when it changes recruitment, support during participation, or a defined analysis. Keep opinion visibly distinct from fact, and ask functional access questions, such as which format or support a participant needs. Apply data minimisation when deciding whether a detail earns a place in the record.

A privacy reviewer should challenge demographic entries that do not affect a research decision. Avoid turning a visible characteristic into a proxy for access needs; ask about the support needed instead, and record only what serves recruitment, participation support, or the stated analysis.

When should you compare or retire a card?

Review cards against each other, not only one at a time, and retain contradictions for investigation. For example, agency operators may describe a clear hand-off while retailer users cannot identify the approval owner; the difference is evidence of a gap, not a reason to choose one account.

Separate evidence from claims when checking a comparison, as in the competitor research approach. Retire a card when the workflow changes, contrary cases overturn its working summary, or the sample no longer fits the decision; do not preserve a label just because it was previously useful.

Where no approved evidence exists, publish a blank template and leave the persona marked UNKNOWN. Use it as one part of the wider target audience evidence, and place it in the context of the England market guide, rather than treating a single card as a complete market picture.

What should you do first?

Start from the research plan and name the decision, role, and environment involved. Then gather evidence before filling fields: if a source is absent, leave the entry blank. This sequence keeps a proposed persona tied to its evidence rather than to a stereotype.