Structured Interview

What a structured interview is

A structured interview is a hiring interview in which every candidate for the same role is asked the same questions in the same order, and their answers are scored against a rating scale defined in advance.

The word “structured” describes how the interview is run, not what it is about. It is a statement about the procedure, and the procedure is the only thing that makes the resulting scores addable at all.

What gets fixed in advance

• The questions. Written before any candidate has been seen, and derived from the requirements of the role rather than from the interviewer’s interests on the day.
• The order. One order, for everyone.
• The rating scale. For each question, a written description of what counts as a weak, an adequate and a strong answer.
• The raters. Each scores independently before any discussion takes place.

Those four are not a checklist of good practice that can be partially adopted. Each one closes a specific route by which the comparison stops being a comparison, and dropping any of them reopens that route.

Why fixing them matters

An unstructured interview produces an impossible comparison. If one candidate is asked how they handle pressure and another is asked about working in a team, then any comparison between the two scores is comparing two questions rather than two people. The scores are numbers, they can be averaged, and the average means nothing. Fixing the questions is what makes the scores addable in the first place.

Writing the questions before seeing any candidate does a second thing that is easy to overlook. It forces the requirements of the role to be stated before anyone is in a position to write them to fit a person already in mind. A question set drafted after the shortlist is a description of the shortlist.

Structure also narrows the space available to competency judgements that are really impressions. A common question set prevents the interview from being tailored around one candidate, and independent scoring prevents opinions from converging on whichever view is voiced first. The second of these is the one most often dropped, because scoring independently feels like a formality when the panel is going to discuss the candidate anyway. It is not a formality: once a senior voice has stated a view, the other scores are no longer independent evidence of anything, and the panel has quietly turned three opinions into one.

Where a structured interview keeps the label and loses the design

All three below are run as structured interviews and described as such, and each has dropped exactly one of the four elements above. In each case the dropped element is the one that was doing the work, which is why the label goes on surviving after the thing it names has stopped.

• A fixed list with no scale. Everyone is asked the same questions and each interviewer forms an overall impression at the end. The question set was the visible half of the design; the scale was the half that turned answers into something addable, so what remains is a well organised conversation that still ends in a preference.
• A fixed list scored after the panel has conferred. The numbers exist and they were written down. They were written down after the view in the room had settled, so they record the discussion rather than the answers, and they will agree with each other almost perfectly, which is usually read as a sign the process is working rather than as the symptom it is.
• A template reused from another role. This one keeps the form of the first element and loses its substance: the questions are fixed, written in advance and asked of everyone, but they were derived from the requirements of a different role, so the requirement the hire is actually being made against was never written down. Comparability is fully preserved, which is what makes it the hardest of the three to notice: the scores are addable, consistent and about something else.

The failure that survives all four elements

There is one more way the comparison breaks with nothing missing from the design, and it is worth naming because none of the four elements catches it: raters who apply the same scale differently. Two interviewers can agree entirely about what happened in an interview and still record a three and a four, because one of them treats the middle of the scale as the default position and the other treats it as a concession. Averaging those two numbers produces a figure that describes neither rater.

The guard is calibration, and it is cheap. Score one candidate jointly before the real interviews begin, compare the numbers, and argue about the gap then, when the argument costs nothing and no decision depends on it. A scale that has never been calibrated is a shared form rather than a shared standard, and the difference between the two only becomes visible when somebody asks why a candidate was rejected.

The semi structured interview

Between the unstructured and the fully structured form there is a middle version: a fixed core of questions asked of everyone, with room for follow up questions that vary according to the answer given. It keeps comparability in the core and allows depth where depth is warranted.

Its condition is precise, and it is the point at which the middle version usually fails. The follow up has to remain a clarification of the answer already given, not a new question that enters the score. The moment a follow up is scored, the candidates are no longer being scored on the same instrument, and the core comparability the design was built to preserve has been spent.

A practical way to hold the line is to keep the score sheet limited to the fixed questions, and to record follow up material as notes that inform the discussion rather than the number. That separates the two functions without forbidding the flexibility that made the middle version attractive.

How it differs from a behavioural interview

Structure describes how the interview is run: fixed questions and uniform scoring. A behavioural interview describes what kind of questions are asked: questions about situations that actually happened.

The two are independent dimensions, and treating them as the same thing is the most common confusion in this area. An interview can be structured and consist entirely of knowledge questions. It can be behavioural and run with no structure at all, which is in fact the common case, since asking everyone to “tell me about a time when” is not the same as asking everyone the same question and scoring it the same way.

What it costs, and where that cost sits

Structure moves effort earlier. Writing the questions, agreeing the scale and calibrating the raters all happen before the first interview, and none of that work is visible in the interview itself. What is visible is that each interview then takes about as long as an unstructured one and produces a result somebody else can check.

Because the preparation is reusable across every candidate for the role, and often across future openings for the same role, the cost per interview falls as the number of candidates rises. A single hire for a one off role is the case where the preparation is hardest to justify, and a repeated or high volume role is where it pays for itself most clearly. That preparation time is internal recruiting time, and it belongs in cost per hire like any other internal hour, which is a reason to record it rather than a reason to avoid it.

What the candidate experiences

Structure is usually discussed from the employer’s side, as a measurement question. It is also visible from the other side of the table, and it is visible as fairness: every candidate is asked the same things, so the process can be explained to somebody who was not selected. An unstructured process cannot be explained that way, because there is no consistent account of what was compared. That is one of the places where the hiring process and candidate experience meet, and it costs nothing extra once the structure is in place.

Related terms

• Competencies
• Candidate Experience
• Cost per Hire
• Employee Onboarding
• Succession Planning