What Questions to Ask in an Interview: A Hiring-Science Deep Dive

The questions to ask in an interview should do one of three jobs: surface evidence about how a person actually works, test whether their stated claims hold under detail, or reveal how they think when the answer is not rehearsed. A good question changes what you know. A weak one only fills silence. The difference between the two is mostly structural, not conversational, and it can be engineered before the interview begins.

Most interview advice treats questions as rapport-building. Hiring research treats them as measurement. That reframing matters because unstructured conversation feels informative while predicting very little, whereas structured, behavior-anchored questioning predicts meaningfully more. The mechanisms below explain why, where they break down, and exactly what to ask.

How interview questions actually extract signal

A question produces signal only when it forces a candidate to retrieve a specific memory rather than construct a plausible generalization. “Tell me about your strengths” invites self-presentation; “Walk me through a decision you reversed after new data” invites episodic recall. Recalled events carry verifiable detail, sequence, and stakes. Constructed answers carry adjectives.

This is why past-behavior questions outperform hypothetical ones. A hypothetical (“What would you do if…”) measures verbal reasoning about a scenario the candidate can optimize in real time. A behavioral probe (“What did you do when…”) asks the candidate to reconstruct a real event, and real events have constraints that invented ones lack. When you follow up with “What happened next?” and “What did your manager say?”, you are checking whether a coherent event exists behind the claim.

The second mechanism is contrast. Ask the same competency from two angles and compare the answers. Someone who describes leading a project as “we decided” in one answer and “I drove the decision” in another is giving you a consistency check you did not have to construct. Divergence between the two stories is often more informative than either story alone.

A third mechanism is the follow-up, which does more work than the opening question. Skilled interviewers treat the first answer as an index and the follow-ups as the measurement. “Can you be more specific about your role?” and “What was the hardest part?” convert a polished summary into a testable account.

Quantified context: what structure buys you

Unstructured interviews predict job performance weakly, and the research literature has consistently found structured interviews outperform them. The direction is stable even where exact correlations vary by role and study design. The practical implication is not that you should script every exchange. It is that pre-decided, job-relevant questions scored against a rubric reduce the noise introduced by interviewer mood, similarity bias, and first-impression anchoring.

The scoring side matters as much as the asking side. A common failure is a strong question with no defined criteria, so two interviewers rating the same answer disagree. Anchor each question to what a strong, adequate, and weak answer looks like before you meet the candidate. Without anchors, comparisons across candidates drift toward whoever interviewed most recently.

Time allocation is another quantified lever. In a 45-minute interview, reserve roughly 10 to 15 minutes for candidate questions and 30 to 35 for your own. Candidates who spend most of the slot answering rather than asking is normal; your job is to keep the proportion deliberate rather than accidental. One well-followed question thread often outperforms eight shallow questions.

The counterintuitive detail: the questions candidates ask you are themselves a signal, but a weaker one than most interviewers assume. Well-prepared candidates ask good questions because they prepared, not necessarily because they will perform. Treat candidate questions as evidence of preparation and priorities, not as a proxy for competence.

Edge cases: when standard questions fail

Standard behavioral questions degrade in several predictable situations. Early-career candidates may lack the work history to supply rich episodes, so academic, volunteer, or project contexts must be explicitly invited. Senior candidates can answer almost any behavioral question from a deep backlog, which means you need questions about judgment trade-offs rather than event recall. Career changers may have relevant capability with no matching job title, so competency-anchored questions beat role-anchored ones.

Cross-cultural interviews introduce another edge case. Direct self-promotion is expected in some hiring cultures and penalized in others, so a candidate who seems vague may be following a different norm rather than lacking substance. Follow-up questions that ask for specifics without demanding bragging partially correct for this.

Remote and asynchronous formats compress the signal further. Video removes much of the incidental context that in-person meetings provide, which raises the value of structured questions and written follow-ups. If part of your process is a recorded or written response, ask the same core questions you would ask live so the comparison stays fair.

Red flag: answers with no named people, teams, or timeframes after two follow-ups. Red flag: every failure is attributed to a colleague, manager, or circumstance with no owned contribution. Red flag: enthusiasm that collapses into vagueness when you ask for a concrete example. Red flag: inconsistent timelines across two questions covering the same period. Red flag: refusal to engage with a reasonable counterexample rather than reasoning through it.

Misconceptions worth retiring

The biggest misconception is that the best questions are the cleverest ones. Brainteasers and “gotcha” questions measure composure under surprise more than job capability, and they invite interviewers to over-weight their own reaction to the answer. A plain question asked well and followed up rigorously beats a clever question asked once.

A second misconception is that culture-fit questions reveal culture fit. Open-ended “Would you fit here?” prompts mostly measure similarity to the interviewer, which is exactly the bias structured questions are meant to reduce. Ask about observable working preferences instead: how the candidate prefers feedback, how they handle competing priorities, what environment makes them do their best work. Those are answerable and comparable.

A third is that more questions mean more signal. Interview fatigue is real on both sides. After roughly the sixth substantive question, answer quality often flattens as candidates pattern-match to what they think you want. Fewer questions with deeper follow-ups generally outperform long lists.

Finally, interviewers often believe they can detect deception through body language. The evidence for reliable lie detection from nonverbal cues is weak, and confidence in one's own detection ability tends to exceed accuracy. Behavioral specificity and consistency checks are more defensible than reading posture.

Comparing question types

The table makes the trade-off explicit: the question types with the strongest validity are the ones that cost the most interviewer effort. Work samples and well-scored behavioral questions require preparation and a rubric. Brainteasers are cheap to ask and expensive in noise.

Practical actions: what to ask and how to score it

Build a small, fixed core of questions tied to the competencies the role actually requires. Three to five competencies, each with one behavioral question and one follow-up, is enough for a single interview. Write the strong, adequate, and weak answer anchors before the session, and score immediately afterward rather than at the end of the day when impressions have blended.

Use a consistent follow-up ladder. First ask for a specific example. Then ask what the candidate personally did. Then ask what happened as a result. That three-step sequence converts most vague answers into either concrete evidence or a visible gap, and it takes under two minutes.

Separate evidence collection from evaluation. Take notes on what was said and done, not on how likable the candidate seemed, then score against the anchors. This ordering reduces halo effects, where one strong answer colors the whole assessment.

For the candidate side of the same coin, the questions to ask in an interview often work best when they mirror the structure above: ask about a specific past decision, ask what happened next, and ask what the interviewer would change. Questions that invite a concrete story from the interviewer produce more information than questions that invite a policy statement.

Reconcile thresholds explicitly. If one part of your process rewards specificity and another rewards enthusiasm, decide in advance which carries more weight for the role rather than letting the loudest signal win. For most roles, specificity should dominate, because it is harder to produce without underlying experience.

Timeline: running a structured interview

The timeline is a guardrail, not a script. The point is that core questions get the largest block and candidate questions are protected rather than squeezed out by overruns. Interviews that drift usually lose time in the warm-up and core sections, not the close.

Limitations and what questions cannot tell you

Interview questions sample behavior under artificial conditions. A candidate who performs well in a structured interview has demonstrated that they can recall, structure, and defend real examples, which correlates with capability but does not guarantee on-the-job performance. The interview is one instrument among several; work samples and reference checks measure different things.

Questions also cannot fully correct for interviewer bias, and no question design removes the effect of a shared background or a strong first impression. Structured questions and pre-written anchors reduce the influence of those factors; they do not eliminate them. Calibration across interviewers, where two people score the same recorded or mock answer and compare, is the practical remedy.

Finally, a good question can only test what a candidate can articulate. Some capable people describe their work poorly, and some less capable people describe theirs fluently. Pairing questions with work samples is the most direct correction, which is why the strongest hiring processes rarely rely on conversation alone.

This article is general educational information about interview practice, not a hiring, legal, or employment-compliance opinion. Hiring and employment decisions can carry legal obligations that vary by jurisdiction, so consult a qualified local professional for advice specific to your situation.