Behavioural interview

The format, and what it settles

One interviewer, one hour, three or four competencies, and a written scale agreed before anybody meets the candidate. This is not a panel format. It works better one to one, because probing hard is easier when the candidate is not being watched by four people while they revise an answer.

The five moves, in order

1. 01

   The prepared story

   Run by Interviewer

   Settles: Nothing yet. Ask the question, let them run, do not interrupt. The rehearsed answer is the starting position and its job is to give you something to probe.

   Ends it: Cannot produce an example at all, after a rephrase and a pause. Either it did not happen or it cannot be retrieved under mild pressure, and both matter.

2. 02

   The particulars

   Run by Interviewer

   Settles: Whether this is a memory or a construction. Dates, names, sequence, who else was in it.

   Ends it: The story moves from "I" to "we" and never comes back, and they cannot say which part was theirs when asked directly.

3. 03

   The cost

   Run by Interviewer

   Settles: Judgement. Every real decision traded something away. Ask what it cost, who paid it, and what they would trade differently now.

   Ends it: A decision with no downside, told twice. Real work has costs and the people who did it remember them.

4. 04

   The aftermath

   Run by Interviewer

   Settles: Durability. What happened to the thing once their involvement ended.

   Ends it: No idea, and no curiosity about it. It is the clearest tell that somebody was present at something rather than responsible for it.

5. 05

   The counter-example

   Run by Interviewer

   Settles: Calibration. Ask for a time the same approach did not work.

   Ends it: Refuses the premise. A method that has never failed has either not been used much or will fail here without warning.

What ends it

• Every story is a success story, including the one you explicitly asked to be a failure.
• No name, no date and no number anywhere in an hour of examples.
• The scale of the story grows as your interest visibly grows.
• Blame sits entirely outside the candidate in every example, including the ones where you asked what they would do differently.
• The answer to "what did it cost" turns out to be a benefit.

How to score it

Owned it

First person throughout, the disagreement named, the cost named, and an accurate account of what they would do differently and why they did not do it then.

Was in it

Real involvement and real detail, with the decisions belonging to somebody else. Perfectly hireable at that level. A problem only if you are hiring for the level above it.

Was near it

Correct vocabulary, no particulars, the story told at team scale. Usually not dishonesty, usually somebody describing the environment they worked in.

Evidence against

A specific observation that predicts a problem: the failure that was somebody else’s fault twice, the cost that turned out to be a benefit, the peer who cannot be named.

Score the competency, not the story. One example can be strong evidence of ownership and no evidence at all of collaboration, and collapsing it into a single overall impression throws away most of what you just heard.

The probes that break a rehearsed answer

STAR is a preparation format, and by now it is a very well prepared one. Somebody who has practised will hand you situation, task, action and result in a clean four-minute block. Everything useful comes after that block.

• "When was this?" Prepared stories are undated. Real ones arrive with a quarter attached, and often with what else was going on at the time.
• "Who disagreed with you?" Every real project has one. Somebody who was actually there names the person and characterises the objection fairly, occasionally better than they characterise their own position.
• "What did you have to give up to do that?" It turns a result back into a decision.
• "What happened to it after you left?" It separates the person who built something from the person who was standing next to it at launch.
• "Whose idea was it originally?" Asked plainly and without suspicion. The answer is frequently generous and specific, and the generosity is itself evidence.
• "What would your manager at the time say was the hardest part of working with you?" A better version of the weakness question, because it is attributed and therefore harder to invent.
• Silence, for four seconds after an answer ends. The addendum people volunteer to fill it is regularly the most useful sentence in the hour.

Behavioural questions worth an hour, by competency

Behavioural and situational are different instruments

Where behavioural interviews go wrong

• Ten competencies in one hour, which produces ten prepared answers and no probing at all.
• A scale invented during the debrief, so every interviewer has been applying a private definition of a good answer.
• The same competency scored by four interviewers, each hearing a better-rehearsed version of the same story, with the improvement recorded as consistency.
• Treating a fluent narrator as a strong candidate. Storytelling is a real skill and it is a job-relevant one for only some jobs.
• Penalising a candidate whose examples are smaller than yours. Scale of company is not scale of ownership, and the reverse mistake is more common than people admit.

Behaviour you can watch instead of ask about

The strongest behavioural evidence is not always an answer. How somebody handles the parts of your process that are not the interview is behaviour too: whether they read the brief, how they respond to a reschedule, what they do when a task is deliberately underspecified, and whether they paste in an answer they were asked to write themselves.

Behavioural signals of that kind sit alongside the structured conversations in how Continuity1 screens, because the way somebody handles a test is evidence in a way that their account of handling a test is not.