Define the research decision and evidence boundary
Record the content or product decision, target situation, intended participants, research questions, excluded claims, round owner, budget, and stopping rule. Every control is pass, fail, or unknown with evidence and date. Unknown remains a blocker when it concerns consent, participant fit, or whether a key finding was observed before the concept was introduced. The audit approves a bounded qualitative inference; it does not turn a convenience sample into a population estimate or promise commercial demand.
Evidence: Government Digital Service; Government Digital Service
Recruitment and consent checks
Verify written criteria for actual or likely users, recent relevant experience, inclusion of unsuccessful or non-adopting cases, recruitment source, incentive timing, accessibility, and reasons for exclusions. Confirm participants received understandable information about purpose, recording, use, retention, and withdrawal. Separate contact data from analyzed notes and collect only what the round needs. Fail when employees or loyal customers are silently treated as the whole audience, or when a session recording lacks the agreed consent.
Evidence: Government Digital Service
Discussion-guide neutrality checks
Mark where the guide asks for a recent episode, follows chronology, probes workarounds and consequences, and permits an answer that contradicts the idea. Flag embedded solutions, assumed problems, praise-seeking language, forced choices, double questions, and concept exposure before open recall. Pew's method guidance and the Oxford review both show that wording and order can influence responses. Pilot the guide with a colleague for clarity, then inspect whether the pilot changed the intended evidence rather than merely making the script sound smoother.
Evidence: Pew Research Center; Oxford Academic
Session-conduct and note checks
Confirm the interviewer allows silence, uses neutral follow-ups, asks what broad terms mean in the specific event, and does not reward agreement. The note-taker should distinguish participant words, described action, observed action, and researcher interpretation. Record which concepts were introduced by the moderator. Protect sensitive details and use anonymous participant codes in the working table. Fail when only polished summaries survive, because another reviewer cannot test whether a finding came from the participant or the team's prompt.
Evidence: Government Digital Service; Government Digital Service
Analysis and contradiction checks
Enter one row per episode with situation, trigger, action, alternative, workaround, consequence, frequency claim, evidence type, uncertainty, and contradiction. GOV.UK recommends extracting observations before findings and involving multiple people to reduce individual bias. Require a second review of a sample, actively search for deviant cases, and retain negative results. A repeated phrase passes only as a language insight until separate evidence establishes that the underlying obstacle recurs across relevant situations.
Evidence: Government Digital Service
Actionability and triangulation checks
Every proposed problem must name the decision it changes, supporting episode IDs, conflicting evidence, and the next source that could disconfirm it. Compare interviews with legitimate behavioral, support, search, community, or transaction evidence where appropriate. Do not average incompatible measures into one confidence percentage. Fail when the team cannot say what it will do differently, when a quote is treated as market size, or when all corroboration comes from sources shaped by the same recruitment or prompt.
Evidence: Government Digital Service; Oxford Academic
Hard stops, sign-off, and recheck
Stop the round for unresolved consent, unnecessary sensitive collection, undisclosed sales activity, systematic leading prompts, or findings that cannot be traced to episodes. Exceptions need scope, owner, expiry, and correction evidence. A second reviewer should reconstruct one finding from the anonymized table before sign-off. Reopen when the audience, problem context, guide, interviewer, or proposed solution changes. A pass means the current method supports a testable qualitative problem statement—not prevalence, willingness to pay, or guaranteed content performance.
Evidence: Government Digital Service; Pew Research Center
Maintain an anonymized decision trail
Archive the question version, recruitment criteria, participant codes, evidence table, excluded rows with reasons, synthesis date, decision, and later outcome. Limit access and retention according to the consent and actual need. When a future article cites an interview-derived problem, link to the anonymous finding identifier rather than copying personal detail. This trail lets editors revisit an interpretation after the context changes without pretending the old sample represents the new audience.
Evidence: Government Digital Service; Government Digital Service
Sources and further reading
These references informed this article. A source supports a claim; it does not imply endorsement of TenMultigure or any future product reference.
- Using in-depth interviewsGovernment Digital Service · Accessed August 10, 2026
Provides the planning, recruitment, consent, neutral-moderation, and recent-example requirements audited before sign-off.
- Analyse a research sessionGovernment Digital Service · Accessed August 10, 2026
Supports traceability from raw observations to findings and the mandatory second-review and deviant-case controls.
- Writing Survey QuestionsPew Research Center · Accessed August 10, 2026
Grounds wording, answer-option, and question-order checks that prevent the guide from manufacturing agreement.
- The Effect of the Question on Survey Responses: A ReviewOxford Academic · Accessed August 10, 2026
Adds independent evidence on response effects used to justify hard stops for systematically leading interview designs.
Reviewed by TenMultigure Editorial Review. See an error or a source that has changed? Tell the editorial team.
Review method: AI-assisted desk research with editorial checks. Reviewed ; next scheduled review . Converted audience interviews into a production audit spanning recruitment through actionability, with pass-fail evidence, contradiction controls, privacy stops, and reproducible sign-off.