Structured interview

The design, and where it is decided

Structure is decided before sourcing starts, not before the interview. A structured interview assembled after the shortlist arrives has been fitted to the candidates you happen to have.

The build, stage by stage

1. 01

   Define the profile

   Run by Hiring manager, with the people who will manage the role

   Settles: What this person has to produce in the first year, and which three or four capabilities decide whether they do.

   Ends it: A profile listing more than five competencies has not been decided, it has been collected.

2. 02

   Write the questions and the anchors together

   Run by Hiring manager and recruiter

   Settles: What will be asked, by whom, and what each band of answer looks like.

   Ends it: If the anchors cannot be written, the competency is not measurable and should be cut rather than carried.

3. 03

   Run the same interview

   Run by The interviewer assigned to each competency

   Settles: Comparable evidence. Same primary question, same order, probes free.

   Ends it: An improvised new primary question. The candidate has now been assessed on something nobody else was asked, and that answer cannot be scored against anybody.

4. 04

   Score alone

   Run by Every interviewer

   Settles: An independent read, submitted before any discussion.

   Ends it: A score entered after the debrief conversation. It is no longer independent and should be recorded as an opinion rather than as a data point.

5. 05

   Decide against the profile

   Run by Hiring manager

   Settles: Whether the evidence clears the bar that was set before anybody was met.

   Ends it: Moving the bar to fit the strongest candidate you happened to see. That is the failure structure exists to prevent, and it is the one that survives every process change.

What breaks the structure

• Interviewers who can see each other’s scores before submitting their own.
• A competency with no anchors, scored anyway.
• Questions varied by candidate because "this one is clearly senior".
• A shortlist scored against each other rather than against the profile. Comparing candidates to candidates is how a mediocre field still produces a hire.
• A five-point scale, for the reason in the note below.

The scale

Sets the standard

Evidence of doing this at or above the level of the role, well enough to raise how your team does it. Rare, and it should be rare, or the anchor is written too low.

Does this

Clear, particular evidence of the thing the competency describes, at your scale, without a support structure you do not have. This is a hire.

Adjacent

Real evidence in the wrong shape. This is the band where most of the argument in a debrief happens, and where written anchors earn their keep.

No evidence, or evidence against

It never came up, or what came up predicts a problem. Record which of the two it is, because they mean opposite things for the next stage.

Four bands and no midpoint. On a five-point scale the middle absorbs every interviewer who was half paying attention, and a debrief where everybody scored three is a meeting with no information in it.

Writing anchors that survive a debrief

The scale is the part teams skip and the part that does the work. An anchor describes an observable answer, closely enough that two interviewers reading the same notes land on the same band without discussing it.

When an interviewer goes off script

• A probe is not off script. Chasing what the candidate said is the point, and probes are supposed to differ between candidates.
• A new primary question is off script. It produces an answer nobody else was asked for, and the honest thing is to record no evidence for the competency it replaced.
• The most common cause is an interviewer who was never briefed and is improvising to avoid dead air. Fifteen minutes of briefing fixes more of this than any amount of process documentation.
• The second most common cause is a senior person who believes their instinct beats the instrument. Give them the hardest competency and the same scale as everybody else, because the objection is usually to being deskilled rather than to structure.
• If somebody will not run the format, take them out of the loop and give them the debrief. A half-structured interview gets scored as though it were structured, which is worse than an honest unstructured one.

Why the unstructured version feels better

An unstructured interview is a more enjoyable hour. The conversation flows, rapport builds, and both people leave with a strong impression. The impression is the problem. It is made of things that are not in the job: shared vocabulary, similar background, comfort in the format, a good week.

• Rapport is a real signal for a small number of roles and a contaminant in most of them.
• The candidate who is best at interviews is not reliably the candidate who is best at the job, and the gap between those two is exactly where an inadequate hire comes from.
• An unstructured interview cannot be compared across candidates, so the decision defaults to whoever was seen most recently or most vividly.
• It also cannot be improved. There is nothing to change next time, because nothing was written down.

Structure without a script

Structure gets resisted because people picture an interrogation with a clipboard. What it actually requires is that the same questions get asked and the same scale gets applied. The conversation around that can be as human as the interviewer is.

Inside Continuity1 the equivalent step is deliberately called a structured conversation: the same bar for everyone, panel-ready, and still a conversation. The rigidity belongs in the sequence and the scale, not in the tone.

Structure is what makes the second hire cheaper

The return on this work does not arrive on the first hire. It arrives on the third, when the questions already exist, the anchors have been calibrated against people you actually hired, and each reject and select has sharpened what the bar means. A loop rebuilt from scratch for every role never gets past the first hire’s learning curve, which is why teams that hire steadily and still feel bad at it are usually rebuilding rather than improving.