Behavioural interview questions ask a candidate to describe something they actually did, rather than what they would do. The premise is simple and well supported: how someone handled a real situation is better evidence than how well they can describe good practice in the abstract.
The format is widely used and frequently used badly. Asking "tell me about a time you showed leadership" and accepting the first polished answer is not a behavioural interview. It is a prompt for a prepared story. The value is entirely in the follow-up.
Design the question from the competency, not the other way round
Start with the four or five things the role genuinely requires, taken from the scorecard. For each, ask what situation would have forced someone to demonstrate it, then write a question that asks for that situation.
If the competency is "works effectively without supervision", the situation is a period with an absent manager or an ambiguous brief. The question becomes: "Tell me about a time you had to make progress on something without clear direction. What was the situation and what did you do."
Do this and your questions are automatically job-related, which matters for defensibility as well as accuracy. Generic question banks produce generic evidence.
Write three or four core questions. Any more and you are trading depth for coverage, which is the wrong trade, because depth is where the signal is.
The follow-ups are the interview
A candidate's first answer is usually rehearsed, and often genuine but summarised to the point of being uninformative. Everything useful comes from what you ask next.
Four probes do most of the work:
"What was your part specifically?" The most valuable question in the set. Listen for the shift between "we" and "I". People who did the work can describe their own contribution precisely. People describing a team's achievement usually cannot, and will keep returning to "we".
"What else did you consider?" Reveals whether there was actual judgement or a single obvious path. Strong candidates describe the option they rejected and why.
"What went wrong?" Every real project has a complication. An answer with no friction in it is a summary, not a memory.
"What would you do differently?" Tests reflection. The answer "nothing" is nearly always either a lack of self-awareness or a story they did not really own.
Keep going until you can picture the situation. If you cannot, you do not have evidence yet, however confident the delivery was.
Questions by competency
Adapt the wording to the role. The structure is the part that carries over.
Ownership and delivery. "Tell me about something you owned end to end that did not go to plan. What happened and what did you do about it."
Working without direction. "Describe a time the brief was unclear and you had to decide how to proceed."
Influence without authority. "Tell me about a time you needed someone outside your team to change how they worked. How did you go about it."
Handling conflict. "Describe a disagreement with a colleague about how to do something. How was it resolved."
Learning quickly. "Tell me about a time you had to become useful in something unfamiliar fast. How did you approach it."
Prioritising under pressure. "Describe a period when you could not do everything. How did you decide what to drop."
Handling a difficult customer or stakeholder. "Tell me about the most difficult client situation you have handled. What did you do."
Managing performance, for people-leadership roles. "Tell me about a time you had to address someone's performance. What did you do and what happened."
Score against a rubric, not an impression
Before the interviews, write what a weak, adequate and strong answer looks like for each question. Three or four lines each. This is the step that gets skipped and it is the one that makes scoring reliable.
For "works without supervision", strong might be: identified the ambiguity early, made a defensible decision, communicated it, adjusted when new information arrived, can explain the reasoning. Weak might be: waited for direction, or proceeded without telling anyone and describes the resulting problem as someone else's fault.
Score each competency separately and immediately after the interview, before speaking to the other interviewers. Independent scores compared afterwards are evidence. Scores formed after a group discussion are a record of who spoke first.
A useful discipline: write the quote, not the rating. "Escalated to the vendor at day four after the second missed checkpoint" is evidence anyone can re-score. "Good judgement" is not.
Where behavioural interviewing struggles
It is not a complete assessment method, and it helps to be clear about the edges.
Early-career candidates have fewer workplace situations to draw on. Widen the frame to study, part-time work, volunteering or personal projects. The competency is the same, the setting is not.
It rewards articulacy. Someone who tells a story well scores better than someone who did the same work and describes it flatly. This is a real bias, it correlates with background and first language, and it is the strongest argument for pairing behavioural interviews with a work sample that does not depend on narration.
It can be prepared for. The follow-ups are the defence, since preparation rarely survives three layers of specific probing.
It looks backwards. For a genuine step up in scope, past behaviour at a smaller scale is informative but not conclusive. Pair it with something that tests reasoning at the new level.
Ask everyone the same questions
Same core questions, same order, same rubric, for every candidate on the role. Follow-ups vary because answers vary. The core does not.
Without this you cannot compare, and the interview reverts to a conversation you enjoyed with some people more than others. With it, you can put two candidates' answers to the same question side by side, which is the entire point.
Common questions
What is a behavioural interview question?
A question asking a candidate to describe a specific situation they actually experienced and what they did, rather than what they would hypothetically do.
How many behavioural questions should an interview include?
Three or four, with real follow-ups, beats eight rushed ones. Depth produces the evidence.
What is the STAR method?
Situation, Task, Action, Result. A structure candidates use to organise answers. As an interviewer, use it as a checklist for completeness rather than as a script, and probe the Action, which is where individual contribution shows.
How do you tell a genuine answer from a rehearsed one?
Follow up on specifics: their individual role, the options they rejected, what went wrong, what they would change. Rehearsed answers thin out quickly under that, genuine ones get richer.
Do behavioural interviews work for graduates?
Yes, with the frame widened to study, part-time work, volunteering and projects. The competency being assessed is unchanged.
Build questions from the competencies the role actually needs, ask three or four with real follow-ups rather than eight without, probe for what the person personally did, write the rubric before the interviews, and score independently before anyone discusses.