By Oliver Bennett, MSc in Work and Organizational Psychology
In short
Structuring an interview means fixing the design before any candidate is seen: the competencies each interviewer owns, a common question set, independent scoring recorded before any discussion, and a defined order of stages. Structure increases the consistency of what is observed and makes candidates comparable to each other rather than to the interviewer's memory of the last one. It is a property of the process, not of how formal the conversation feels.
A panel of four meets a candidate for an hour. Everyone asks what interests them. Afterwards they talk, agree quickly, and write "strong communicator, some concerns about pace". The next candidate gets a different hour from the same four people, and the two are compared as though the same thing had been measured twice.
Nothing in that room was badly done. The design was missing before anyone walked in.
Structure is a set of decisions made in advance
The word "structured" is usually taken to mean formal — same questions, no small talk, a scoring sheet. That is part of it and the smaller part. Structure is the architecture of the whole stage: who assesses what, when judgements are recorded, when they are allowed to meet, and where the interview sits relative to everything else.
Those decisions are made once, before the first candidate, and they are what makes the hour comparable across people. Decisions made inside the room are, by definition, made differently each time.
Divide the competencies
The default panel has everyone assess everything. It feels thorough and it is the main source of wasted interview time: four people produce four readings of the same two or three impressions, and the agreement between them is read as confirmation when it is mostly shared exposure.
Assign instead. Each interviewer owns two or three competencies from the blueprint and asks the questions that reach them. Nobody covers everything, and between them the panel covers the role.
| Everyone assesses everything | Competencies divided | |
|---|---|---|
| What you get | several readings of the same impressions | evidence on distinct competencies |
| Agreement means | shared exposure | genuine convergence, where it happens |
| Gaps | invisible — everyone assumed someone covered it | visible on the coverage map |
| Interview length | grows with panel size | stays bounded |
The division also makes the gap honest. If no interviewer owns "handles conflict with a peer", the coverage map shows a blank rather than four people each assuming another had it.
Record before you reveal
This is the single highest-value rule in panel design, and the easiest to drop under time pressure.
Each interviewer scores independently against the rubric and records it — written down, submitted — before any discussion. Then the scores are revealed, and then the panel talks.
The reason is anchoring. Once a confident impression is in the room, the others move toward it. The panel experiences this as reaching consensus; what actually happened is that one judgement was repeated by three people. A debrief that begins "I thought she was excellent, what did everyone else think?" has already collected its answer.
Revealed-after-recording also makes disagreement visible, which is the useful part. Two interviewers two points apart on the same competency is a conversation worth having — about the rubric, or about what each of them saw — and it is a conversation that never starts if the first number spoken is the one everyone settles on.
The order of candidates is part of the measurement
Interviewers calibrate against what they have recently seen. The third candidate of the morning is scored partly against the first two. A strong opener raises the bar for everyone after; a weak one lowers it.
Two cheap mitigations. Randomise the order, so position in the queue is noise rather than a systematic penalty for whoever scheduled late. And if the panel is new to the rubric, re-score the first candidate at the end — the first interview is also the panel's calibration session, and nobody should pay for that.
Where a panel runs across several days, a short re-grounding in the rubric at the start of each day costs five minutes and prevents the standard from quietly resetting overnight.
Sequence the stage, not just the hour
An interview is one stage in a process, and its design depends on what comes before it.
Run the work sample first. Then the interview is not spent establishing whether someone can do the work — that question has an answer already — but on what the work sample cannot show: the reasoning behind the choices, how they would handle the case that escalates, what they would do differently with more time.
This ordering shortens the interview and sharpens it. It also removes the most common failure of the unstructured hour, where a persuasive account of competence is weighed against another candidate's demonstration of it, and the account wins because it was more fluent.
What structure does not buy
Structure makes observations consistent. It does not make them correct.
A structured interview still reports what a person says about their work, filtered through how comfortable they are being interviewed. That is genuinely useful evidence about reasoning, self-awareness and how someone handles a live exchange — and it is not a substitute for seeing the work. Keeping both stages is not belt and braces; they observe different things.
And structure does not remove judgement. It relocates it: from the room, where it is made under time pressure by whoever speaks first, to the design, where it is made deliberately, written down, and applied the same way to everyone. That is the whole of the improvement, and it is enough.
- structured interview
- interview panel
- independent scoring
- hiring process design
- evaluation
Key takeaways
- Divide competencies between interviewers rather than having everyone assess everything — overlap produces agreement, not information.
- Score independently and record it before the panel talks. A judgement spoken first becomes the panel's judgement.
- Candidate order is part of the measurement. Randomise it, and rescore the first candidate at the end if the panel is new to the rubric.
- A structured interview is still an interview: it reports what a person says about their work, which is why the work sample stays in the process.
Sources & further reading
- Levashina, Hartwell, Morgeson & Campion (2014), The Structured Employment Interview: Narrative and Quantitative Review — Personnel Psychology
- Macan (2009), The Employment Interview: A Review of Current Studies and Directions for Future Research — Human Resource Management Review
- AERA, APA & NCME — Standards for Educational and Psychological Testing