Defining a retail media audience from observable buyer roles and recent decisions

A retail media target audience is the population able to answer one stated buyer question. It is not everyone who shops, advertises or works in retail. Define the recent event and decision first, then recruit the people who actually performed or experienced that part of the exchange.

Write a population rule

Start with one England operation and one retail-media environment. IAB UK's retail media definition includes retailer or marketplace controlled advertising opportunities and data assets across on-site, off-site and in-store settings. Select one setting; each has different users and evidence.

Complete this blank statement:

Adults are eligible when they [performed or experienced the exact task] for [named retailer, advertiser or supplier role] during [date window], concerning [defined placement or process] connected to England.

An advertiser buyer who rejected a sponsored-search brief can inform the buying decision. A retailer operator can explain inventory approval. Agency staff can describe a delegated hand-off. A measurement specialist can test event definitions. Consumers can report whether they recognised a paid placement or completed an accessible task. One group cannot stand in for the others.

Build a recruitable frame

Name the source list and its owner: an approved client list, staff roster, campaign decision log or ethically recruited panel. Record its extraction date, fields, known omissions and contact permission. Public job titles and company filings do not establish that a person handled the event.

The GOV.UK guide to finding user research participants recommends actual or likely users, explicit recruitment criteria and attention to recruitment bias. Its public-service context does not determine a commercial sample size. Set the number from the decision, method and likely variation, with statistical or qualitative-method review.

Count the frame, eligible people, invitations, responses, completions and analysed records separately. Do not advertise the completion count as the audience size. Record why each person was excluded and whether non-response could hide a materially different view.

Before sending invitations, ask a second researcher to apply the eligibility rule to several synthetic records. Log disagreements about role, event date and England connection, then clarify the screener. This checks whether the population can be reproduced without exposing a real person's details. If reviewers cannot reach the same eligibility decision, recruitment stays on hold.

Recruit fairly and accessibly

The invitation should identify the researcher, purpose, activity, duration field, incentive, recording choice, data use, publication plan and withdrawal contact. Keep all values blank until approved. Do not combine recruitment with a sales message.

Ask what format or support enables participation rather than asking for a diagnosis. Offer an accessible screener and information sheet. Government guidance on research sessions with disabled people advises speaking directly to participants and not guessing what they can do or how they feel.

Children are excluded from this default frame. If the buyer's question cannot be answered without them or an adult at risk, stop recruitment until a safeguarding lead has approved a specific protocol and consent route.

Limit and test the inference

Predeclare what the evidence can change. A recurring advertiser objection may justify revising a brief; it does not prove England-wide demand. A consumer task failure may justify an accessibility repair; it does not establish media effectiveness.

The UK Statistics Authority's Code of Practice for Statistics 3.0 sets out principles for suitable sources, transparent methods and explained uncertainty. Use those as voluntary quality questions. Publish the exact population, frame, dates, denominators and limits, and retain UNKNOWN wherever the evidence cannot answer the buyer decision.

Set an expiry for the audience definition. A changed media environment, buyer authority, campaign workflow or research purpose can make the old frame unsuitable even when the names remain current. The decision owner must then revise the population rule rather than merging incompatible rounds.