Behavioral Interview Questions
A behavioral interview question asks a candidate to describe something they actually did: a problem they solved, a deadline they missed, a colleague they disagreed with. The premise is that past behaviour in a similar situation is the best available predictor of future behaviour, and the research on selection methods has borne that out for forty years. The catch is that the questions only work inside a structure: the same questions for every candidate, probes that get to what the person did, and a scoring scale agreed before the first interview. Without that, a behavioral interview is a pleasant conversation with the predictive value of one.
The builder below assembles a structured guide from twenty-five questions grouped by competency, with the probes and scoring anchors written in. The rest of the page covers why the format works, how to score answers, the questions that create legal exposure, and the mistakes interviewers make with it.
Interview guide builder
Five or six with follow-ups fills a 45-minute interview.
Nothing typed here leaves your browser. The guide includes a scoring anchor for each question and the same probes for every candidate, which is what makes the interview structured rather than a conversation.
Why structured behavioral interviews predict performance
The approach traces to Janz (1982), who compared "patterned behaviour description" interviews with unstructured ones and found the patterned version predicted performance far better. The largest recent re-analysis of selection research, Sackett and colleagues (2022) in the Journal of Applied Psychology, corrected a statistical over-adjustment in earlier meta-analyses and concluded that structured interviews are the single strongest predictor of job performance among common selection methods, ahead of cognitive ability tests, job knowledge tests and work samples. Unstructured interviews sit well down the same table.
What makes the difference is not the questions but the structure around them. Three elements carry the weight. The questions are the same for every candidate, so answers can be compared. The probes are planned, so every candidate is pushed to the same depth. And the scoring uses behaviourally anchored scales, agreed in advance, so a "4" means the same thing to every interviewer. Add a second interviewer scoring independently and the reliability rises again.
The STAR frame (situation, task, action, result) is the interviewer's checklist for a complete answer, not a script for the candidate. Most answers arrive as situation and a vague result. The probes exist to extract the action, in the first person, and the result with evidence. "We improved the process" is not an answer; "I rewrote the intake form, and the error rate fell from about one in ten to one in fifty over the next quarter" is.
The twenty-five questions by competency
The builder's list is grouped by the competency each question tests. Pick one or two from each competency the job actually needs, and do not ask leadership questions of people who will not lead. Five or six questions with probes fill a 45-minute interview; more than that and the answers get thinner as the clock runs.
| Competency | What a strong answer shows | What a weak answer sounds like |
|---|---|---|
| Problem solving | Diagnosed before acting; tested the fix; knows what would have happened otherwise | Jumped to a solution; the result is "it got better" |
| Ownership and results | Chose to act without being asked; quantifies the result; names what went wrong honestly | Speaks in "we"; blames the constraint; no numbers |
| Working with others | Describes the other person's view fairly; changed something in response to feedback | The other person was simply wrong; feedback was "taken on board" |
| Adaptability | Explains how the priority call was made and what was consciously dropped | "I just worked harder" |
| Judgment and integrity | Named the conflict, raised it through a proper channel, accepted a cost | Did what was asked and felt bad; or a hypothetical instead of an example |
| Leadership | Specific conversations, specific dates, a documented process, a clear outcome either way | "I believe in giving people a chance" |
| Customer and quality | Knows the standard, knows when to break it, can say why | Every story ends with a delighted customer |
Two questions are worth asking of everyone regardless of role. "Tell me about a deadline you missed" and "describe a mistake that affected other people" separate candidates who have examined their own work from those who have not. A candidate who cannot produce a single missed deadline in a career has either never been stretched or is not telling you about it, and either is information.
Scoring answers
Score each answer on its own scale immediately after the interview, before talking to anyone else, and write the evidence next to the score. The anchors in the builder are generic; the better practice is to write role-specific anchors before interviewing. For a support lead the "5" on problem solving might read: identified a recurring ticket cause from the data, proposed a fix to engineering with reproduction steps, and can quote the change in volume afterwards.
- Score the behaviour, not the story. A dramatic story with a vague action scores below a modest story with a precise one.
- Discount hypotheticals. "What I would do is" answers a different question. Redirect once; if the candidate has no example, score the absence.
- Weight recent and relevant examples. A university group project counts for a graduate; it counts for little from a ten-year professional.
- Compare across candidates on the same question. The panel discussion is question by question, evidence first, score second. Overall impressions come last, if at all.
- Keep the notes. They are the record that the decision was made on job-related evidence, which is the defence if it is ever questioned.
The notes and scores go in the hiring file with the job description the questions were built from, and they inform the reference check: a referee asked about the specific example the candidate gave is far more useful than one asked for a general opinion.
Questions that create exposure
A structured interview is also the safest interview. The EEOC guidance on selection procedures treats an interview as a selection procedure like any test: it must be job-related, applied consistently, and not screen out protected groups without business necessity. Behavioral questions tied to the competencies in the job description satisfy that by design. Improvised questions are where the trouble starts.
- Anything touching a protected characteristic: age ("when did you graduate?"), family ("do you have children?"), religion ("what do you do on Sundays?"), national origin ("where is your accent from?"), disability ("have you had any health problems?"), pregnancy, marital status, military discharge type, arrests.
- Salary history, banned in around twenty states and several cities; our pay transparency page lists them.
- "Culture fit" questions with no competency behind them ("what do you do for fun?") that invite similarity bias and produce nothing scorable.
- Brain teasers and puzzles, which several large employers abandoned after finding they predicted nothing.
If a candidate volunteers protected information, do not pursue it, do not write it down, and return to the question. Where a candidate needs an accommodation for the interview itself (extra time, a different format), provide it; our accommodation guide covers the process.
Where behavioral interviews go wrong
- Different questions for different candidates. The comparison collapses. Use the guide.
- Accepting the first answer. The first answer is the rehearsed one. The second probe ("what did you personally do?") is where the interview starts.
- Leading probes. "So you escalated it, presumably?" hands the candidate the answer. Probes are open: "What happened next?"
- Talking too much. The interviewer should speak for a fifth of the time or less. Silence after an answer produces more than a follow-up question does.
- Scoring in the room. Scores written during the interview drift toward the impression of the moment. Write evidence in the room, score afterwards.
- Deciding in the first five minutes. The best-documented interviewer bias. The structure is the defence: a scored guide forces evidence to be collected after the impression has formed.
- Skipping the debrief. Interviewers who discuss candidates before scoring converge on the most confident voice. Score alone, then meet.
Once the candidate is chosen, the offer letter and the new hire forms follow, and the examples gathered in the interview make a good starting point for the 30-60-90 day plan.
Key takeaways
- Behavioral questions ask for what the candidate did, not what they would do. Their value comes from structure: same questions, planned probes, anchored scores.
- The 2022 Sackett re-analysis puts structured interviews first among selection methods for predicting performance. Unstructured interviews sit far below.
- Use STAR as the interviewer's checklist. Probe until you have the candidate's own action and a result with evidence.
- Five or six questions across the competencies the job needs fill a 45-minute interview. Ask everyone about a missed deadline and a mistake.
- Score each answer alone, immediately, with the evidence written next to it, then compare question by question in the debrief.
- Stay off protected characteristics and salary history. A guide built from the job description is the safest interview you can run.
Frequently asked questions
What are behavioral interview questions?
Questions that ask a candidate to describe a specific past situation and what they did in it, on the premise that past behaviour predicts future behaviour. They usually begin 'tell me about a time' or 'give me an example of', and are followed by probes to establish the situation, the candidate's own actions and the result.
What is the STAR method?
A frame for a complete answer: Situation (the context), Task (what the candidate was responsible for), Action (what they personally did) and Result (what happened, with evidence). Interviewers use it as a checklist and probe for whichever part is missing, most often the action in the first person and a measurable result.
How many behavioral questions should an interview have?
Five or six, with probes, for a 45-minute interview. Each takes six to eight minutes done properly. More questions means shallower answers. Choose them across the competencies the job description actually requires.
Do behavioral interviews really predict job performance?
Structured interviews, which behavioral interviews are when run with consistent questions and anchored scoring, are the strongest common predictor of job performance according to the 2022 re-analysis of selection research by Sackett and colleagues. The structure does the work; the same questions asked conversationally without scoring predict much less.
How do you score a behavioral interview?
Each answer gets a rating, typically 1 to 5, against anchors written before the interview that describe what a weak, adequate and strong answer contains for that competency. Score immediately after the interview, alone, with the evidence noted, then compare candidates question by question in a debrief.
What questions should you not ask in an interview?
Anything that touches a protected characteristic: age, family status, religion, national origin, disability, pregnancy, marital status, arrests. Salary history is banned in about twenty states. Questions with no competency behind them, such as culture-fit small talk, produce nothing scorable and invite bias.