Audience research
Part of Audience research for a website project that keeps buyers and end users apart
Customer interviews for web design agencies: six practices that hold up
Six non-ranked interview practices for England website research, covering recruitment, access, consent, neutral questions and evidence limits.
An interview can reveal a person's recent experience and language. It cannot establish market prevalence, prove what people generally do or replace observation of a working journey.
What to take away
- Interviews reveal recent experience and language, not market prevalence or general behaviour.
- Each session must link to a decision and an assumption it could challenge.
- Recruit eligible participants and record recruitment details to manage bias.
- Ask about access needs before the session and avoid unnecessary health details.
- Protect evidence with codes, restricted access and deletion triggers.
How the practices were selected
The source review for this interview list closed on 6 September 2026.
England and UK scope: these practices are for sessions concerning an England website project. The sources are UK government or regulator guidance, much of it written for public services, and are adapted rather than treated as a private-sector legal code.
Inclusions: a practice needed a direct source, a defined research action and a clear safeguard against biased or unusable evidence.
Exclusions: invented quotations, anonymous agency tips, claimed sample-size rules, sales scripts and product promotions.
Ranking and conflicts: the six entries are non-ranked. No participant, agency or software provider contributed to this desk review.
1. Tie the session to one decision
Write what the owner must decide and which assumption the interview could challenge. GDS guidance on planning a research round links objectives to problems, assumptions and information needed next. Stop if the team cannot explain how an answer will change the work.
2. Recruit people eligible for the task
Separate organisational buyers from end users. The Service Manual page on finding participants recommends actual or likely users and warns that recruitment route, time and location introduce bias. Record the population, England rule, screen, invitation route, refusals, non-response, attendance and withdrawal when fieldwork occurs.
3. Ask about access before the session
Offer communication formats, equipment, breaks, travel support and a suitable remote or physical setting. The GDS guide to sessions with disabled people says to plan for individual needs, speak directly to the participant and ask before helping. Do not request unnecessary health details.
4. Make participation and recording explicit
Provide an information sheet naming the researcher, purpose, observers, data, sharing, retention, complaints and withdrawal route. The Service Manual's informed-consent guidance says participants should be told whether and how a session is recorded. A privacy reviewer must still assess lawful basis and any special-category information.
5. Ask for recent episodes, neutrally
Start with open prompts about what happened, then ask for concrete detail. GDS guidance on in-depth interviews favours current or likely users, real examples and open questions. Avoid presenting a design and asking whether it is good. Mark what the participant said, what the researcher observed and what remains an inference.
6. Protect evidence and dissent
Use participant codes, restrict raw material and set deletion triggers. The GDS page on participant data recommends collecting the minimum required information and limiting access. Include contradictory cases and explain sampling limits in the decision note.
If employees are interviewed or routinely recorded, check whether the process becomes worker monitoring. Current ICO monitoring guidance is under review and requires a facts-based assessment of purpose, necessity and transparency.
The final output should show the question, eligible population, recruitment path, session record, evidence type, limitation and decision. There are no customer findings to report until those sessions genuinely occur.
Before you act
- Write the decision and assumption the interview will test.
- Recruit actual or likely users and record the recruitment path.
- Ask about access needs before the session.
- Provide an information sheet and get informed consent.
- Ask for recent episodes with open, neutral prompts.
- Use participant codes and set deletion triggers.
Common questions
What can an interview reveal and what can it not establish?
An interview can reveal a person's recent experience and language. It cannot establish market prevalence, prove what people generally do or replace observation of a working journey. These limits mean findings must be treated as evidence of individual accounts, not as proof of wider patterns.
How should recruitment be handled to reduce bias?
Separate organisational buyers from end users. The Service Manual recommends actual or likely users and warns that recruitment route, time and location introduce bias. Record the population, England rule, screen, invitation route, refusals, non-response, attendance and withdrawal when fieldwork occurs.
What should be done about consent and recording?
Provide an information sheet naming the researcher, purpose, observers, data, sharing, retention, complaints and withdrawal route. The Service Manual's informed-consent guidance says participants should be told whether and how a session is recorded. A privacy reviewer must still assess lawful basis and any special-category information.
