Structured Interview Kit

You design the interviews so that every candidate gets the same fair test of the same job-related competencies, and so that the decision rests on written evidence instead of who the panel liked. Meta-analyses of selection methods (Schmidt and Hunter, 1998; Sackett and colleagues, 2022) consistently rank structured interviews among the most valid predictors of job performance, and far ahead of unstructured conversation. The kit you produce is what makes an interview structured: fixed questions, anchored ratings, independent scoring, and a disciplined debrief.

When to use

• "Put together the interview process for the head of sales."
• "Write interview questions and a scorecard for the support role."
• "Our interviewers all ask different things, fix it."
• "Run the debrief for yesterday's finalists."

What you need

• The job description with outcomes and must-haves (see write-job-description).
• The interview panel: names, roles, and which of them have interviewed before.
• The time available: number of stages and their length.
• Access: the owner's ATS (a connected app) or calendar and documents through the agent's own browser signed in by the owner. Without either, work in files and hand them over.

Method

1. Choose four to six competencies. Derive them from the outcomes and must-haves: technical skills, ways of working, and values the company has written down.
   • Each must be observable and job-related.
   • "Culture fit" is not a competency; "gives and asks for direct feedback" can be.
2. Map competencies to stages. Build a matrix (competency by stage).
   • Each competency is assessed in one or two stages, not five.
   • Each interviewer owns two or three competencies, matched to what they are best at judging.
   • Include a work sample for the core skill wherever possible (a short, paid or time-boxed task that resembles the real work, under about 3 hours of candidate time).
3. Write the questions. For each competency write one primary question and three planned probes, using references/question-bank.md:
   • Past-behaviour questions ("Tell me about a time you...") for experienced candidates.
   • Situational questions ("What would you do if...") for early-career candidates or new situations.
   • Probes that dig for the person's own actions: "What did you do?", "What did you decide, and why?", "What happened next?", "What would you do differently?". Use the funnel: open question, narrow to specifics, find motivation and result, summarise back.
4. Anchor the ratings. For each question, write behavioural anchors for 1, 3 and 5 (behaviourally anchored rating scale).
   • A 3 is "meets the bar for this level".
   • Write what the candidate would actually say or do, not adjectives.
5. Build the scorecard. One per interviewer per candidate: competencies, rating 1 to 5, evidence notes (what the candidate said or did, quoted where possible), and an overall recommendation on a four-point scale (strong no, no, yes, strong yes; no "maybe"). Any score below the top must say why, so the next interviewer can probe it.
6. Write the interviewer brief. From references/interviewer-guide.md: the agenda (welcome and agenda, questions, candidate's questions), the legal and ethical limits, how to offer adjustments, how to sell the role without implying an offer, and the rule that feedback is submitted within 24 hours and before reading anyone else's.
7. Make it accessible. In the invitation draft, tell the candidate the format, length, who they will meet, whether there is a task, and how to ask for adjustments. Share the work sample brief in advance where reasonable.
8. Schedule (drafts only). Propose times to the owner. Invitations to candidates are drafts until the owner approves.
9. Run the debrief. Collect every scorecard before the meeting.
   • Open with the scores shown together, or have the most junior interviewer speak first, so seniority does not anchor the room.
   • Discuss competency by competency, against the bar, not candidate against candidate.
   • Note disagreements with evidence.
   • The hiring manager decides; record the decision and the reasons in job-related terms.
10. Close the loop. Draft the outcome for each candidate (next stage, offer handover, or rejection with the guidance in references/interviewer-guide.md) for the owner to approve.

Judgement calls

• Build panels with different roles and backgrounds where you can, and never let one person's view decide alone.
• An adjustment request (extra time, written questions in advance, a different format): agree it with the owner and apply it. The anchors and the bar stay the same.
• An interviewer asks a forbidden question or goes off script: note it for the hiring manager and HR, and keep that content out of the decision.
• Internal candidates take the same kit as external ones.
• A split debrief: gather more evidence on the specific competency in dispute (a short follow-up conversation or a reference with the candidate's consent) rather than averaging opinions.
• A candidate does brilliantly on one competency and poorly on a must-have: the must-have is not traded away for the highlight unless the hiring manager changes the role, in writing.

Output

• interview-kit-<role>.md: competency-by-stage matrix, questions with probes and anchors per stage, timings, the work sample brief if any.
• scorecard-<role>.md: the blank scorecard.
• interviewer-brief-<role>.md: one page for the panel.
• After interviews: debrief-<role>.md with scores per competency per interviewer, discussion notes, decision, owner of next step.

Scorecard layout:

Candidate ID · Stage · Interviewer · Date
| Competency | Rating 1 to 5 | Evidence (what they said or did) |
Overall: strong no / no / yes / strong yes
If not the top score, why:
Concerns for the next interviewer to probe:
Submitted before reading anyone else's feedback: yes / no

Checks before you finish

• Every competency traces back to an outcome or must-have.
• Every question has probes and 1/3/5 anchors; every candidate gets the same primary questions in the same order.
• No question touches a protected characteristic or salary history (check against the list in the guide).
• The candidate invitation explains format and how to ask for adjustments.
• Scorecards are collected before the debrief starts.
• Nothing has been sent to a candidate without the owner's approval.

Pitfalls

• The friendly chat. Unstructured interviews feel insightful and predict poorly. Keep to the kit.
• Brain teasers and trick questions. They measure nerves, not the job.
• Halo effect. One great answer colours every rating. Score each competency on its own evidence.
• Groupthink in the debrief. If anyone reads others' feedback first, independent judgement is lost. Lock scorecards before discussion.
• Notes you would not want read aloud. In many countries candidates can request their interview notes. Write evidence, not opinions about the person.
• "Where are you from originally?" and "Do you have kids?" Small talk can be unlawful. Brief interviewers to stick to the job.

See also: screen-cvs-fairly (the stage before) and draft-offer-letter (the stage after).