What a structured interview is
A structured interview is one where the questions, the competencies being assessed and the scale used to score them are fixed before the first candidate is seen, and applied identically to every candidate after. The interviewer decides what to ask before knowing who will answer, and decides how to score before hearing the answer.
The opposite is the unstructured interview: a conversation that goes where the interviewer’s curiosity and the candidate’s answers take it, ending in a general judgement of fit. That flexibility can be useful, but without shared criteria it becomes harder to compare what each interviewer learnt.
Why a shared structure helps
A shared structure gives the panel a consistent way to explore the requirements of the role. Asking agreed questions makes it easier to compare relevant experience across candidates.
Written scoring criteria also give interviewers a common reference point. Each person records their own observations before the debrief, where the team can examine both agreement and differences.
A good conversation still matters. The scorecard helps the panel distinguish the quality of that conversation from the evidence of how a candidate would approach the work.
The parts of a structured interview loop
A structured interview loop has nine parts: an interview plan, rounds and modes, competencies, panel roles, a rubric with behavioural anchors, a scorecard, blind submission, a debrief and a committee decision. Each is defined once per role and reused for every candidate.
- Interview plan. The written definition of the loop for a role: how many rounds, in what order, who sits on each, what each is for, and how long the panel has to submit feedback. It is attached to the role, not to the candidate, and it does not change between candidates.
- Rounds and modes. Each round has a purpose, a duration and a mode: video, in person or telephonic. A common loop for a professional role runs a recruiter screen, one or two skill rounds, a hiring manager round and a cross-functional or leadership round.
- Competencies. The handful of things the role actually requires, named specifically enough to be observed. “Communication” is too broad; “explains a technical decision to a non-technical stakeholder” can be asked about and scored.
- Panel roles. Who interviews, and what each is responsible for. Two interviewers assigned the same competency in different rounds give the panel a second reading; two interviewers assigned everything give it a blur.
- Rubric with behavioural anchors. For each competency, a scale, usually four levels, with a written description of what each level looks like. The next section shows how to write one.
- Scorecard. The form each interviewer completes: a score per competency they were assigned, the evidence for it in the candidate’s own words or actions, and an overall recommendation. It is short by design.
- Blind submission. No interviewer sees any other scorecard until their own is submitted. This is the rule that preserves the independence of each observation.
- Debrief. The panel meets with all scorecards visible, side by side, and examines where scores diverge.
- Committee decision. A hire or no-hire decision recorded with reasons, by a named group, before any offer is raised. No scorecards, no decision; no decision, no offer.
How to write behavioural anchors
A behavioural anchor describes what a specific score on a specific competency looks like, in terms of what a candidate at that level does or says, so that two interviewers reading the same anchor will give the same behaviour the same score. Write one anchor per level per competency, and make each anchor describe behaviour rather than qualities.
Take the competency Ownership, for a mid-level role, on a four-level scale:
| Level | Anchor |
|---|---|
| 1. Below bar | Describes outcomes as things that happened to them. Attributes delays and failures to other teams or circumstances. Cannot name a decision they took without being told to. |
| 2. Approaching bar | Completes what is assigned and escalates blockers promptly. Describes their own contribution clearly but waits for direction when scope is unclear. Fixes problems inside their remit; flags those outside it. |
| 3. At bar | Gives a specific example of noticing a problem outside their stated remit and acting on it without being asked. Names what they decided, what they gave up and what happened. Follows through to the outcome, not the handoff. |
| 4. Above bar | Gives an example of changing how a team or process works so the problem stops recurring, and of bringing others along to do it. Is candid about a decision that went wrong and what they changed as a result. |
Three rules keep anchors useful. Describe observable behaviour, not adjectives: “gives a specific example” can be checked, “is proactive” cannot. Make adjacent levels clearly different, so that the interviewer is choosing between descriptions rather than points on a line. And write the anchors with the people who will use them, then test them on a few past interviews to see whether two readers score the same evidence the same way.
Running the debrief
The purpose of a debrief is to surface where the panel disagrees and find out why, not to average the scores into a verdict. An average hides exactly the information the meeting exists to examine: one interviewer saw something the others did not, and the panel needs to decide whether it was signal or noise.
Run it in this order. Put every scorecard side by side, by competency, so that the pattern is visible at a glance: agreement is dull and can be passed over; divergence is the agenda. For each point of divergence, ask the interviewers who scored differently to read out the evidence they wrote, in their own words, unedited. Very often the disagreement dissolves once the evidence is heard, because one interviewer asked a question the other did not. Sometimes it does not dissolve, and the panel has learnt that the candidate is inconsistent on that competency, which is itself evidence.
Two disciplines protect the meeting. Nobody edits their scorecard after reading the others; the scores are what they were, and any change of mind is spoken in the room and recorded in the decision log. And the most senior person in the room speaks last, because a hiring manager’s opinion delivered first turns the rest of the panel into an audience.
The output is a decision with reasons, written down, by the committee, before any offer is raised. The reasons should be traceable to the scorecards. “Strong on ownership and systems design, below bar on stakeholder communication, panel agreed the role is senior enough that the gap is disqualifying” is a decision that can be revisited a year later. “Good fit” is not.
Feedback SLAs and why they matter
A feedback SLA is the number of hours an interviewer has to submit a scorecard after the interview, and it matters because scorecards written days later are written from memory of an impression rather than from notes of the evidence. The longer the gap, the more the score records how the interviewer feels about the candidate and the less it records what the candidate said.
There is a second cost, to the candidate. The candidate is usually employed already and usually in more than one process. Every day the loop waits for a scorecard is a day the candidate hears nothing, and the process that moves fastest tends to be the one the candidate chooses. A loop with a two-day feedback SLA and an escalation when it is missed protects the decision and the candidate at the same time.
Make the SLA part of the interview plan, per loop, and measure it: feedback turnaround per interviewer and per round is one of the few metrics that a hiring team can act on the same week. Escalate overdue feedback within the agreed response time, not after.
Scaling structure: interview days and panel capacity
Structure scales when the interview plan, rubric and scorecard are defined once per role and the scheduling is driven by the panel’s real capacity, so that adding candidates adds interviews without adding new decisions about how to interview. The constraint is almost never the number of candidates; it is the number of trained interviewers and the hours they can give.
Two practices make the difference. The first is treating interviewer capacity as a resource to be planned: an interviewer roster with availability, a load limit per person per week, and interviewer groups by competency, so that a round can be staffed from anyone qualified to run it rather than from one over-booked expert. The second is the interview day: concentrating a role’s loops into a scheduled block of slots and rooms, physical or virtual, so that a panel completes many loops in a day with the same plan and rubric, and the debriefs run the same evening while the evidence is fresh.
The rubric does not change because the day has two hundred slots in it. Structure is what makes the day possible, because it is the reason interviews conducted by forty different interviewers can be compared at all.
In Elevator
Elevator by Superset implements the loop as described here: interview plans per role with rounds, modes, panel roles, competencies and a feedback SLA; scorecards anchored to four-level behavioural rubrics, blind until submitted; a debrief view that puts every scorecard side by side on a competency heat-map with divergence surfaced rather than averaged and the panel’s words unedited; a committee decision logged before an offer can be raised; escalation when feedback is overdue; and a day board with candidate self-scheduling on real panel capacity. See interviews and scheduling and Elevator for hiring managers, or request a demo. Terms are defined in the recruiting glossary.