Audience research
Part of Audience research for property creators with populations and permissions kept apart
Building defensible research personas with evidence cards and no fictional profiles
A method for building defensible research personas in England, with evidence cards, coded notes, recruitment counts, denominators and no fictional identities.
A defensible persona is a summary of observed needs within a defined population. It is not a fictional character made plausible with a name, portrait, salary, house type or invented quote. No England-wide persona dataset exists for property creator services.
What to take away
- Set an indicative recruitment threshold for each role and region before treating a pattern as defensible.
- Keep buyer and user roles separate, because a commissioning organisation, approver, viewer and resident are different evidence subjects.
- Use an evidence card with population, dates, recruitment counts and coded notes, and report the denominator behind every claim.
- Leave out identity details such as stock photos and addresses, since combinations of facts can identify a person indirectly.
- Retire a persona when the decision changes, sources expire or new evidence overturns the pattern.
Do not mix the buyer and the user
Start with separate evidence files. A commissioning organisation may buy a production service. Its content approver may use the workflow. An adult home seeker may watch the result. A property operator may supply facts. A resident captured in footage is a rights and privacy subject, not a buyer persona.
Decide which role the persona represents before recruiting. For organisation buyers, a useful sampling frame may come from the ONS business activity, size and location tables, filtered by selected industry and English region. The unit is a VAT and/or PAYE registered enterprise. It neither identifies media purchasers nor includes every operator.
For adult readers, use a stated housing task rather than assumed demographics. The English Housing Survey can show the range of household and tenure contexts in England under its own design. It does not license a publisher to attach survey findings to an invented individual or claim that a sampled reader represents a national segment.
Use an evidence card
Give the card a functional label, such as "regional residential media approver for an agency covering Greater Manchester". Record:
Evidence card contents
- Population and England geography
- Research dates and recruitment sources
- Invitations, eligible participants, completed cases
- Recent decisions or tasks discussed
- Supporting observations and contrary cases
- Access needs, missing roles, known bias
- Editorial decision informed; review and expiry dates
- the population and England geography;
- research dates and recruitment sources;
- invitations, eligible participants and completed cases;
- recent decisions or tasks actually discussed;
- supporting observations and contrary cases;
- access needs, missing roles and known bias;
- the editorial or service decision the evidence informs;
- review and expiry dates.
Set an indicative threshold before you treat a pattern as defensible. For one role in one English region, a first round might aim for at least 8 completed cases. This is a working threshold, not a national benchmark. If you have fewer than 8 completed cases, report the gap.
Worked example of a completed card. Counts are anonymised and illustrative, not a benchmark.
Field / Completed card example
- Label
- Regional residential media approver, Greater Manchester
- Population
- Content approvers in agencies with 10 to 49 staff
- Research dates
- 3 February to 14 March 2025
- Recruitment source
- Two agency networks
- Invitations
- 18
- Eligible
- 12
- Completed cases
- 9
- Coded notes
- P01 sought legal sign-off before publication; P04 wanted portrait and wide crops; P07 rejected drone shots over named developments; P09 asked for captions checked by compliance
- Contrary cases
- Two approvers did not mention legal review
- Decision
- Revise the approval checklist
- Review date
- 3 February 2026
- Expiry date
- 3 February 2027
Link each statement to coded notes. If five participants mention a problem, report five and the relevant denominator. Do not replace the count with words such as "most" unless the calculation and exclusions support it. A vivid comment does not outweigh contradictory cases.
The GOV.UK guidance on managing participant data recommends minimising recruitment information, separating access and removing identifiers from shared outputs. This is government research-method guidance, not automatic compliance for a commercial creator. A UK privacy reviewer should set lawful basis, retention, access and deletion for the actual study.
Code notes and report denominators
- Give each participant a code that carries no identity, such as P01.
- Give each theme a short label, such as APPROVAL-DELAY.
- In the note, record the date, the role, the task and the exact count.
- Report the denominator as the number of completed cases in that role, not the number of invitations.
- If five of nine completed cases mention a delay, write "5 of 9 completed cases" and list the four contrary cases.
- Keep the codebook with the card so another researcher can repeat the count.
Leave identity out unless it is necessary
Do not give the card a stock photograph, exact address, employer, financial position or distinctive property story. A combination of neighbourhood, job and moving date may identify someone even if their name is absent. The ICO says personal data can identify a person directly or indirectly.
Keep research permission separate from publicity. If a real participant's words or image will appear in an advert, CAP's testimonials guidance requires evidence that the testimonial is genuine and permission for its use. Show the participant the proposed context and obtain a distinct publication decision.
Give each card a version number and an evidence owner. When two roles need different information, create two cards rather than adding incompatible needs to a composite person. The owner should be able to show which coded records support a statement, which participants disagreed and whether later evidence changed the decision.
A persona is ready only when another researcher can trace every statement, see disagreements and understand the sampling limits. Retire it when the decision changes, its sources expire or new evidence overturns the pattern. If the sample is too narrow, publish the gap in the research record instead of decorating the card with assumptions.
Before you act
- Decide which role the persona represents before recruiting.
- Start with separate evidence files for each distinct role.
- Record invitations, eligible participants and completed cases on the card.
- Link each statement to coded notes and report the denominator.
- Keep research permission separate from publicity permission.
- Give each card a version number and an evidence owner.
Common questions
Why should a persona not be given a name, portrait or invented quote?
Because those details are assumptions, not evidence. They can also identify a person indirectly. Link each claim to coded notes instead.
What should go on an evidence card?
The fields listed in the evidence card section, plus counts, coded notes and the denominator. Omit identity details unless the decision requires them.
When is a persona ready to use, and when should it be retired?
It is ready when a second researcher can trace each statement and see the sampling limits. Retire it when the decision changes, sources expire or new evidence overturns the pattern.
