Audience research
Part of Audience research for an online store that starts with the journey, not the recruit
Running store journey interviews that recruit from a recent event, not an opinion
Run safe, reproducible store journey interviews using recent-event eligibility, neutral questions, verified notes and explicit evidence limits.
Method: controls were selected where a current official record supports recruitment, participant treatment, safe journey design or interpretation. Research date: 6 September 2026. Scope: adult research for a defined England store journey; cited UK or Great Britain guidance retains its original geography. Inclusions: practices tied to observable evidence and merchant decisions. Exclusions: invented respondents, quotations, findings, prevalence, sample-size promises and product tests. Ranking: non-ranked. Conflicts: no participant, supplier or retailer funded the selection.
What to take away
- Recruit participants based on a recent journey state or decision, not general interest in optimisation.
- Keep the exercise safe by avoiding real payments, unsafe products and broken cancellation routes.
- Obtain specific participation agreement covering purpose, data, recording, withdrawal and contacts.
- Ask about the event before the opinion to capture what happened, not just preferences.
- Interviews can reveal mechanisms and language but cannot establish prevalence or causal uplift.
1. Recruit from a recent journey or decision
Write the target population, sampling frame, field period and observable eligibility event. Invite a customer who recently reached the relevant journey state, or a merchant role-holder who owns the decision. General interest in optimisation is not enough.
The Government Service Manual's participant guidance says research should involve actual or likely users and describes how recruitment choices create bias. Publish counts for screened, eligible, invited, delivered, responding, enrolled and completed cases.
2. Keep the exercise commercially and physically safe
Do not require a purchase, real payment, misleading price, unsafe product interaction or broken cancellation route. Use synthetic products, test accounts and an authorised payment simulation. The CMA's online choice architecture collection explains that interface design affects shopping decisions, while official online selling guidance records seller information duties.
Stop if a participant reaches a live charge, cannot find a safe exit, sees an undisclosed mandatory amount or exposes another person's information. Consumer, product-safety and payments owners decide whether work can resume.
3. Obtain specific, usable participation agreement
Tell people the purpose, activities, data, observers, recording, recipients, retention, withdrawal route, controller and complaint contact. The Service Manual's consent guidance recommends retaining evidence of what each participant agreed to.
Participation is not permission for marketing or optional tracking. A privacy and PECR specialist must approve those separate purposes. The interview should end, and collected research data be handled under the approved protocol, if a person withdraws.
4. Ask for the event before the opinion
Freeze a versioned guide. Begin with what the person intended, saw, did and did next in the recent journey. Ask what record could corroborate the sequence. Only then ask what they thought or preferred. Do not propose a desired answer, forecast spending or ask the participant to interpret the law.
The Service Manual's round-planning guidance recommends clear objectives, a practice session and planned recording and analysis. Log deviations from the approved question order.
5. Make the session accessible to the individual
Ask what format, technology, pace or assistance the participant wants. Do not infer access needs from identity or treat one person's experience as representative of a disability group. Official guidance on running research with disabled people emphasises accessible information, consent and direct respectful communication.
An inaccessible consent form or prototype is a stop condition, not missing data to ignore.
6. Preserve statement, observation and inference
Mark notes as system observation, participant statement, researcher interpretation or unknown. Verify any quotation with the speaker before external use, including wording and context. The Service Manual's analysis page separates observations, findings and actions.
Report item denominators, non-response, missing evidence, sample limits and conflicts. Interviews can expose mechanisms and language. They do not establish England prevalence, a conversion benchmark, attributed revenue or causal uplift. A named researcher and statistician must approve those boundaries before publication.
Before you act
- Define the target population and observable eligibility event.
- Use synthetic products and authorised payment simulation only.
- Tell participants the purpose, data use and withdrawal route.
- Freeze a versioned guide and ask about the event first.
- Ask what format, pace or assistance the participant wants.
- Mark notes as observation, statement, interpretation or unknown.
Common questions
How should recruitment for a store journey interview be designed?
Set out who you want, how you will sample them, the dates, and a clear eligibility trigger. Pick someone who has just been through that journey stage, or the merchant who makes the call.
What safety conditions must be met before running the session?
Nobody should have to buy anything, hand over money, meet a fake price, touch a dangerous item, or hit a dead end when cancelling. Use dummy goods, test logins and an approved payment simulation instead.
What can interviews establish and what can they not establish?
They can show you how something works and the words people use. They cannot tell you how common a problem is in England, give a conversion figure, or prove that a change caused a gain.
