Interview Protocol Builder Prompt

Build a review-ready, semi-structured interview protocol from the inputs below.

Inputs
- Research objective: [Research objective]
- Participant population and sampling context: [Participant population and sampling context]
- Study context and decisions: [Study context and decisions]
- Interview constraints: [Interview constraints]
- Source materials: [Source materials]
- Definition of done: [Definition of done]

Claude’s operating boundaries
- Analyze only information available in this conversation or in accessible uploaded materials. Do not claim to have opened inaccessible links, contacted participants, consulted systems, or reviewed documents that were not provided.
- Produce a proposed protocol, not evidence that interviews, pilots, approvals, recordings, or analyses have occurred.
- Do not publish, schedule, send, approve, or administer the protocol. Human authorization is required before participant contact, recording, recruitment, or field use.
- Treat supplied requirements and source documents as evidence. Distinguish them from assumptions, hypotheses, conflicts, and unknowns. Do not invent participant characteristics, legal requirements, organizational policy, or prior findings.

Input assessment
1. Determine whether the minimum prerequisites are available: a research objective, intended participant population, decisions the study should inform, interview duration or timing constraints, and relevant safety or consent constraints.
2. Ask no more than five focused clarification questions if missing or conflicting information would materially affect participant safety, consent, question relevance, or interpretation. Do not draft a field-ready protocol while a safety-critical prerequisite is unresolved.
3. If gaps are non-blocking, continue with a bounded draft. Record each assumption, its consequence, and what a human reviewer must confirm. Preserve conflicting source statements rather than silently choosing one.
4. Create a compact evidence ledger classifying each important input as supplied fact, source-supported requirement, assumption, hypothesis, unknown, or conflict. Cite the supplied document name or input section where possible; do not fabricate quotations or citations.

Protocol design workflow
1. Convert the research objective into a small set of answerable research questions. For each one, identify the decision it informs, evidence needed, relevant participant group, and topics that are outside scope.
2. Build an objective-to-question traceability map. Every core question must support at least one research objective, and every objective must have adequate question coverage without unnecessary duplication.
3. Select an interview structure appropriate to the objective and population. Explain consequential trade-offs such as standardization versus exploration, breadth versus depth, recall versus recency, and sensitivity versus evidentiary value.
4. Design a realistic timed flow: interviewer preparation, introduction, identity or eligibility confirmation only when necessary, consent, warm-up, core sections, reflection, participant questions, and close. Include transition language and a time-recovery plan identifying questions that may be shortened or skipped.
5. Write neutral, plain-language questions that address one construct at a time. Prefer concrete experience, behavior, sequence, critical-incident, comparison, and example requests over speculation. Avoid leading wording, double-barreled questions, unsupported presuppositions, forced agreement, and unnecessary personal-data collection.
6. Add optional probes for clarification, chronology, examples, exceptions, consequences, frequency, and disconfirming evidence. Clearly separate probes from required questions so interviewers do not administer every probe mechanically.
7. Add analysis tags linked to the research questions. Define each tag, inclusion and exclusion rules, likely evidence indicators, and overlaps with related tags. Do not present the tags as validated findings or claim that themes have been observed.
8. Add interviewer guidance for active listening, neutral follow-up, handling silence, documenting exact language versus interpretation, and avoiding promises. For multi-interviewer studies, define standardization points, permitted adaptation, handoff notes, and calibration checks.
9. Provide an opening and consent script that states purpose, expected duration, voluntary participation, right to decline or stop, recording status, intended data use, confidentiality limits, and a contact or escalation placeholder described in prose for human completion. Flag any consent statement requiring privacy, legal, ethics, or institutional approval; do not represent template language as legally sufficient.
10. Define proportionate safeguards. Minimize collection of names, protected characteristics, health details, credentials, confidential business information, and other unnecessary sensitive data. Recommend redaction or access restrictions where relevant. Include pause, skip, withdrawal, distress, disclosure, and recording-failure procedures.
11. If the protocol concerns employment candidates, use only approved, job-related competencies and consistent core questions. Exclude questions designed to elicit protected characteristics, family status, medical information, political or religious views, or other non-job-related personal data. Map evidence to the approved scorecard rather than making a hiring decision.
12. Stop and identify the required human review if the request involves covert recording, deceptive consent, discriminatory screening, avoidable collection of sensitive data, vulnerable populations without an approved safeguard path, or claims that the protocol has already been approved or validated without evidence.

Required deliverable
A. Protocol status and scope
- Status must be one of: Draft, Blocked pending clarification, or Ready for human review.
- State the objective, participant population, intended decisions, interview mode, target duration, exclusions, dependencies, and unresolved constraints.

B. Evidence and uncertainty ledger
- Use columns: item, classification, supporting source, effect on protocol, confidence, and required confirmation.

C. Research framework
- List the research questions and decision links.
- Provide a traceability table with: objective ID, research question, evidence sought, participant relevance, protocol question IDs, and coverage status.

D. Interview run sheet
- Provide a timed sequence with: phase, minutes, purpose, interviewer script or action, consent checkpoint, and adaptation rule.
- Reconcile section timings to the stated total duration and show the arithmetic.

E. Question and probe bank
- Use columns: question ID, phase, exact wording, objective ID, evidence sought, required or optional status, neutral probes, analysis tags, sensitivity level, interviewer caution, and skip condition.
- Clearly distinguish core questions from optional probes and contingency questions.

F. Consent, privacy, and safety notes
- Supply editable opening and closing scripts.
- List data-minimization measures, recording and note-taking requirements, confidentiality limits, withdrawal handling, distress or disclosure escalation, and approvals still required.

G. Analysis tag codebook
- Use columns: tag ID, tag name, definition, include when, exclude when, evidence indicators, related tags, and linked objective IDs.

H. Administration and calibration plan
- Include interviewer preparation, standardized elements, allowed adaptations, note-taking conventions, multi-interviewer calibration, escalation routes, and materials needed before field use.

I. Validation and acceptance register
For each check, report the expected condition, the actual observation from the drafted text, status as Pass, Revise, Unverified, or Not applicable, and corrective action. Check at minimum:
- Every research objective maps to one or more core questions.
- Every core question has a stated evidentiary purpose.
- Questions are neutral, singular, understandable, and non-duplicative.
- Probes do not introduce assumptions or pressure participants.
- Consent precedes recording or substantive questioning.
- Sensitive-data collection is necessary, minimized, and safeguarded.
- The timed sections reconcile to the target duration.
- Analysis tags have definitions plus inclusion and exclusion rules.
- Skip, withdrawal, distress, and recording-failure paths are usable.
- Candidate protocols, when applicable, are job-related and consistently scoreable.
- Claims of approval, pilot validation, field completion, or observed findings are supported by actual evidence.

J. Pilot and handoff plan
- Propose a cognitive review and pilot procedure covering question comprehension, neutrality, timing, probe usability, participant burden, missing response options, note quality, and tag applicability.
- Define what evidence a human reviewer should record, how revisions should be reconciled, and which unresolved issues prevent field use.
- Keep proposed pilot checks marked Unverified until results are supplied. Never state that the protocol is approved, validated, administered, or complete without corresponding evidence.

End with the smallest safe next action for the human owner, naming the specific approval, clarification, or pilot activity required before field use.