Audience research
Part of Audience research for an online store that starts with the journey, not the recruit
Store personas as evidence cards rather than invented characters
Build evidence-linked store journey personas for England without invented people, stereotypes, false prevalence or unsupported conversion assumptions.
A useful store persona is a dated summary of evidence about a particular journey and decision. This method is written for England and UK stores. It is not a fictional customer with a stock photograph, a catchy name and invented buying motives. Treat the card as a hypothesis that can be corrected or retired.
What to take away
- A store persona is a dated summary of evidence about a journey and decision, not a fictional customer.
- Keep observation, preference and interpretation in separate fields on the card.
- Leave out fictional names, portraits, ages, incomes, occupations and family stories.
- Add a confidence note covering frame gap, non-response and missing items.
- A card cannot approve a price, claim, cancellation route, tracker or accessibility treatment.
Name the evidence boundary
Begin with the legal seller, store, channel, journey stage, England geography rule and research period. State the target population, sampling frame and eligibility event. Then list the number screened, eligible, invited, responding and represented in the evidence behind the card.
Recruitment numbers to report
- Screenedtotal contacts reviewed
- Eligiblemet criteria
- Invitedasked to take part
- Respondingcompleted research
The event should be observable and recent: for example, a purchaser who reached a defined delivery step during the field window, or an authorised merchant owner reviewing that step. Interest in online shopping, a postcode or a device type does not establish eligibility.
The Government Service Manual's participant recruitment guidance advises defining criteria and recognises recruitment bias. It is a method source rather than evidence about England buyers. For England stores, practical sources are the store's own checkout analytics and a UK user-research panel.
Use an evidence card, not a character sketch
A concise card can contain:
Evidence card fields
- Task and decision context
- Observed journey actions and version
- Participant statements, labelled
- Barriers and workarounds seen
- Contradictions and missing evidence
- Sample and item denominators
- Source dates, owner, review date
- Permitted decision and retirement trigger
Use an evidence card
- task and decision context;
- observed journey actions, with system and definition version;
- participant statements, labelled as statements;
- barriers or workarounds seen in the research;
- contradictions and missing evidence;
- contributing sample and item denominators;
- source dates, owner and next review date;
- the decision for which the summary is permitted; and
- a correction or retirement trigger.
Keep observation, preference and interpretation in separate fields. "The participant returned to delivery information" can be an observation. "I expected collection" is a statement. "The label may be unclear" is an editorial interpretation. None is a conversion rate, an attributed sale or a causal effect.
Worked example: a completed evidence card for a UK homeware retailer's checkout delivery step, dated 12 March 2025, with illustrative figures.
Use an evidence card
- Task and decision context
- Choosing between home delivery and store collection at checkout step 3.
- Observed journey actions
- Participant selected collection, then returned to home delivery twice; system version 4.2.
- Participant statements
- "I expected free collection." Labelled as a statement.
- Barriers or workarounds
- Delivery label was ambiguous; participant used the browser back button.
- Contradictions and missing evidence
- Contradicts an earlier finding that collection was clear; postcode missing.
- Contributing sample and item denominators
- 6 participants from 18 invited; 42 session recordings (illustrative).
- Source dates, owner and next review
- Checkout analytics 1 to 31 January 2025; owner: research lead; review 12 June 2025.
- Permitted decision
- Improve collection label for the England store checkout.
- Correction or retirement trigger
- Retire if the checkout flow or delivery options change.
Do not manufacture personal detail
Leave out fictional names, portraits, quotations, ages, incomes, occupations and family stories. Do not infer disability, ethnicity, sex, vulnerability, residence, purchasing power or technical skill from a profile, device or transaction. If a relevant access need is volunteered and lawfully handled, describe the task and arrangement rather than making it a personality.
The government's Equality Act guide for service providers explains the Great Britain duty to anticipate barriers and consider reasonable adjustments. That duty supports inclusive design; it does not support assuming that people sharing a protected characteristic behave alike.
Research files also require minimisation. The Service Manual's participant privacy guidance recommends collecting only what is needed, limiting access and deleting recruitment details when no longer required. A persona should point to controlled evidence, not copy identifiable notes into a broad-access slide.
Show uncertainty and revision
Add a confidence note that explains the frame gap, non-response, missing items and whether participants used the live journey or an authorised prototype. Do not generalise qualitative themes to prevalence across England. If assisted buyers or people blocking optional tracking are absent, say so prominently.
Example confidence note: The frame covered purchasers who reached checkout step 3 on the England store between 1 and 30 June 2025; it missed assisted buyers and people who blocked optional tracking. Of 24 invited, 7 responded, so non-response was 17. Missing items: delivery postcode and device type. These figures are illustrative.
Use a version history. A product owner may promote the card for one bounded decision only after the researcher confirms traceability and the statistician checks the limitation language. Retire it when the interface, eligibility event, evidence definition or relevant policy changes. Reopen it when contradictory research arrives.
Keep assurance outside the persona
A card cannot approve a price, product claim, cancellation route, tracker or accessibility treatment. Consumer, privacy and PECR, product-safety, security and accessibility specialists retain separate gates. Publication also requires named user-research, statistical, equality and privacy review. These cards organise evidence; they do not certify a customer type or predict commercial behaviour.
Before you act
Before you act, check the card against the field list under 'Use an evidence card' and confirm the confidence note is present.
Common questions
What should a store persona actually contain?
See the field list under 'Use an evidence card'. The card must stay dated and traceable to the evidence boundary.
Why should personal detail be left out of a persona?
Fictional names, portraits, ages, incomes, occupations and family stories are not evidence. Do not infer disability, ethnicity, sex, vulnerability, residence, purchasing power or technical skill from a profile, device or transaction. If an access need is volunteered and lawfully handled, describe the task and arrangement instead.
When should a persona card be retired or reopened?
Retire it when the interface, eligibility event, evidence definition or relevant policy changes. Reopen it when contradictory research arrives. A product owner may promote the card for one bounded decision only after the researcher confirms traceability and the statistician checks the limitation language.
