Human ResourcesInterviewing & Selection

AI Prompts for Structured Interview Design

Structured interviews — the same job-related questions, asked of every candidate, scored against the same anchors — predict performance far better than unstructured conversations and are far fairer. The structure is in three places: which competencies the loop tests and who tests each; how questions and work samples are designed so answers produce evidence; and how interviewers are trained to record that evidence and score it.

These prompts design the loop, the questions and the interviewer training. The question bank page holds the ready-made questions; the scorecard page holds the scoring; candidate comparison holds the debrief. This is the architecture that makes those work together.

Before you use these

Have these ready to replace the highlighted [variables]:

Success profile and the competencies that matter most for the role
Interview stages, interviewers available and time per stage
Whether work samples or exercises are appropriate for the role
Your policy on questions to avoid and reasonable adjustments

These prompts structure evidence against stated, job-related criteria and draft documents. They do not make hiring decisions. Do not ask the model to rank, reject or recommend candidates on protected characteristics or their proxies (name, address, school, gaps, age indicators), and do not treat its output as a decision: a person accountable for the hire reviews every score and makes the call. Check local law before using any automated or AI-assisted screening.

The prompts

1. Design the interview loop around competencies
2. Write behavioral and work-sample questions with anchors
3. Train interviewers to gather evidence
4. Design a work-sample exercise

1. Design the interview loop around competencies

Best forDeciding what the loop tests, who tests it, and in what order, so nothing is assessed three times and nothing important is missed.

Inputs needed

Competencies with priority
Stages and interviewers
Time available

How to use itGive the competencies from intake. The model assigns each to a stage and interviewer, adds a work sample where it adds signal, and produces the loop plan.

Expected outputCompetency × stage matrix, interviewer assignments, work-sample recommendation, timing, candidate-facing loop description, and the coverage check.

Act as a talent assessment designer building a structured interview loop for [role].

Competencies (prioritized): [from the success profile, with the evidence each requires]
Stages available: [e.g. screen, hiring manager, panel, work sample, executive] with interviewers and durations
Constraints: [candidate time, interviewer availability, accessibility]

1. Competency × stage matrix: each competency assessed in exactly one primary stage (and at most one secondary), by an interviewer with the standing to judge it. No competency assessed nowhere; none assessed everywhere.
2. Work sample or exercise: where it adds predictive signal, what it simulates (a real task, time-boxed, with the same brief for everyone), how it is scored, and the candidate time it costs.
3. Loop order and timing: the sequence that lets early stages screen efficiently and later stages go deep; total candidate hours.
4. Interviewer assignments and the reason for each; the calibration each interviewer needs.
5. Candidate-facing loop description: stages, what each assesses, how to prepare, adjustments available.
6. Coverage and fairness check: any stage that assesses 'culture fit' or general impression — replace with a defined competency or remove.

Keep the loop proportionate to the role. More interviews do not add signal; better-defined ones do.

2. Write behavioral and work-sample questions with anchors

Best forQuestions that produce evidence for a specific competency, with probes and anchored scoring.

Inputs needed

Competency and its behavioral indicators
Stage and interviewer
Level

How to use itOne competency at a time. The model writes past-behavior questions, a work-sample brief where relevant, probes, and the 1–4 anchors — with what each level of evidence sounds like.

Expected outputPer competency: 2–3 behavioral questions, probes, an optional work-sample brief, anchored 1–4 scoring with example evidence, and the pitfalls for interviewers.

You are writing interview questions for the competency [competency] for [role] at [level].

Behavioral indicators: [what this competency looks like in the role]
Stage and interviewer: [who asks, how much time]

1. Behavioral questions (2–3): past-behavior framing ('Tell me about a time when…'), specific to the indicators, asking for a real situation, actions and results. Avoid hypotheticals except as a follow-up.
2. Probes for each: the follow-ups that get to the candidate's own actions (not the team's), the decision points, what they would do differently.
3. Work-sample brief (if suitable): the task, materials, time limit, and what it reveals about the competency.
4. Anchored scoring (1–4): for each level, what the evidence sounds like — concrete, observable, not adjectives. Include the 'insufficient evidence' option distinct from a low score.
5. Pitfalls: leading follow-ups, rewarding fluency over substance, penalizing unfamiliar contexts, similarity bias; the counter-practice for each.
6. Equivalent-experience note: how the questions apply to candidates from non-traditional backgrounds (e.g. 'a project' not 'a job').

Questions must be job-related and asked of every candidate for the role.

3. Train interviewers to gather evidence

Best forA short interviewer guide and calibration exercise so scores mean the same thing across the panel.

Inputs needed

Loop design
Questions and anchors
Interviewer experience

How to use itThe model writes the one-page guide, a calibration exercise using a sample answer, and the note-taking template.

Expected outputInterviewer guide (structure, note-taking, scoring, what not to do), a calibration exercise with a sample answer to score, and the note template.

Act as an interviewer trainer preparing the panel for [role].

Loop: [stages, competencies per stage]
Questions and anchors: [summary]
Panel: [experience level, known habits]

1. One-page interviewer guide: the purpose of structure; ask the same questions in the same order; take verbatim notes of what the candidate did; score after the interview against the anchors, not against other candidates; separate 'evidence' from 'impression'; what to do when a candidate goes off track.
2. Note-taking template: question, situation, actions, result, evidence for the competency, score, confidence.
3. Calibration exercise: a sample answer (write one, realistic) to a question from the loop; have interviewers score it independently; the model's anchored score with the reasoning; the discussion points where panels usually disagree.
4. Bias checkpoints: the five most common in this kind of loop (halo, similarity, recency, contrast, fluency) with the concrete counter-practice for each.
5. Questions not to ask and how to redirect if the candidate volunteers personal information.
6. Candidate experience rules: timekeeping, explaining the format, leaving time for questions, the same warmth for everyone.

Short enough to read before an interview. Practical, not a lecture.

4. Design a work-sample exercise

Best forA realistic, time-boxed exercise that shows what a candidate can do, scored the same way for everyone.

Inputs needed

The competency and the real task it corresponds to
Materials you can share
Time the candidate can reasonably give

How to use itGive a real task from the role. The model designs the exercise around it, with a brief, materials, time limit, scoring anchors and the accessibility and fairness rules.

Expected outputExercise brief, materials list, time limit, what it assesses, anchored scoring, reviewer guide, and fairness rules.

Act as an assessment designer creating a work-sample exercise for [role] to assess [competency].

Real task it mirrors: [describe the actual work]
Materials we can share: [data, documents, a scenario — anonymized]
Candidate time: [maximum minutes/hours] Format: [take-home / live / paired]

1. Brief to the candidate: the scenario, the task, the deliverable, the time limit, what will be assessed and what will not (polish, tooling familiarity), and the statement that the work will not be used commercially.
2. Materials: what is provided so every candidate starts equal; what is deliberately omitted and why.
3. Scoring: anchored 1–4 for each assessed dimension (e.g. approach, judgment, quality, communication of reasoning) with what evidence looks like at each level; reviewers score independently before discussing.
4. Reviewer guide: what to look for, common good answers that look unconventional, what not to reward (length, speed beyond the limit, familiarity with our tools).
5. Fairness rules: the same brief and time for everyone; adjustments available on request; no exercise that requires unpaid work of commercial value; the time limit proportionate to the role.
6. Debrief with the candidate: a short conversation about their choices, so the exercise measures reasoning and not only output.

Keep it short and realistic. A four-hour take-home tests availability, not capability.

Related prompts

Logical next step

After this, most HR teams move on to Interview Question Banks.

Get the free HR & Recruiting AI Starter Kit → Nine of these prompts as an intake → assess → offer hiring workflow with an intake worksheet, delivered by email. See what's inside

All Human Resources prompts · Search the full library