The most useful questions to ask someone you are interviewing are structured, behaviorally anchored prompts that ask candidates to describe specific past situations, the actions they personally took, and the measurable results. Questions about hypotheticals or self-assessments predict job performance poorly. Structured, past-behavior questions asked in the same order to every candidate are the most reliable interview method available.
How structured questions extract signal
An interview is a measurement instrument, and like any instrument it has a signal-to-noise ratio. The signal is evidence about how a candidate actually behaves under job-relevant conditions. The noise is everything else: first impressions, similarity bias, nerves, and the candidate's skill at telling you what you want to hear.
Past-behavior questions work because they exploit a simple mechanism. Adults are poor at predicting their own future behavior in unfamiliar situations, but their recent history is a reasonably stable sample of it. When you ask "Tell me about a time you missed a deadline," you force the candidate to retrieve an episodic memory rather than generate a socially desirable answer on the spot. Fabrication is possible but costly, because follow-up probes about sequence, names, and numbers expose inconsistencies.
Structured interviews beat unstructured ones by large margins in meta-analytic research, roughly doubling predictive validity. The mechanism is standardization: same questions, same scoring rubric, same order. That removes the confound of interviewers improvising different conversations and then comparing gut feelings. A structured interview is closer to a written test than to a chat.
The behavioral logic itself traces to the premise that past performance in similar situations predicts future performance in similar situations. That premise holds best when the past situation closely matches the target role. A question about resolving a conflict with a peer at a previous job predicts conflict handling in your workplace only if the stakes, authority structure, and culture are comparable. Probe for those details.
Plain language: ask what someone actually did, not what they would do.
Quantified context: validity, duration, and question counts
Interview structure is an old finding, but the numbers are worth knowing before you redesign your process. Structured interviews consistently show higher predictive validity than unstructured ones in meta-analytic work by Schmidt and Hunter and related industrial-organizational psychology research. The gap is not marginal; it is roughly the difference between a weak and a moderate predictor of job performance.
Duration matters less than people assume. A well-structured 45-minute interview with six to eight behaviorally anchored questions can outperform a loose two-hour conversation. Cognitive fatigue degrades interviewer scoring quality after roughly an hour of continuous interviewing, so spread panels across the day rather than stacking them.
Question banks for a single role typically work best at 6 to 10 core questions per competency set, with each competency covered by at least two independent questions scored separately. Fewer than 4 questions total makes scores statistically unstable. More than 12 in one sitting invites fatigue and declining inter-rater agreement.
Raters need calibration. A 30-minute calibration session where two interviewers independently score the same recorded mock interview and then reconcile differences measurably improves agreement. Without calibration, inter-rater reliability on unstructured scoring can be poor enough that two interviewers effectively evaluate different candidates.
The counterintuitive detail: adding more interviewers does not fix a bad question set. Averaging many noisy gut reactions produces a confident but still noisy number. Fix the questions and the rubric first.
Edge cases: roles, seniority, and remote contexts
Not every role rewards identical question types. Senior individual contributors and managers need questions that surface judgment under ambiguity and influence without authority. Early-career candidates may have thin work history, so shift from employment examples to project, academic, or volunteer situations with the same structure.
Remote interviews weaken the informal signal and strengthen the structured signal. Without hallway impressions, your rubric does more of the work. Ask explicitly for measurable outcomes, since remote candidates can more easily describe team results as personal results. Follow up with "What did you personally do, and who else was involved?"
Regulated and safety-critical roles often require scenario questions alongside behavioral ones. For example, a safety-sensitive position may need a "what would you do if" prompt to confirm the candidate knows the required protocol. That is a knowledge check, not a personality probe. Keep it separate in scoring.
Internal candidates are an edge case that trips up many processes. Internal applicants already have a performance record, so interviews should focus on the specific new demands of the role rather than replaying generic behavioral questions. Ask what they have done in the last 12 months that resembles the target role's hardest task.
Adverse-impact and accommodation considerations also apply. In the US, the Equal Employment Opportunity Commission enforces anti-discrimination law, and interview questions must be job-related and consistently applied. That is another practical argument for a fixed question set.
Misconceptions about interview questions
The biggest misconception is that clever questions reveal character. Brainteasers, "what animal are you," and hypothetical puzzles mostly measure composure and cultural familiarity, not job performance. They also invite interviewer bias because scoring is entirely subjective.
A second misconception is that candidates cannot prepare for structured interviews, so preparation invalidates them. Candidates can and do prepare. But preparation mainly improves the polish of an accurate story, not the underlying content. If a candidate must fabricate a detailed situation with names, timelines, and metrics under probing, the cost rises sharply.
A third misconception is that gut feel is a valid tiebreaker. It is a valid data point only if you record it separately and weight it consciously. Most hiring failures are not caused by bad instincts about people; they are caused by unstructured processes that let those instincts silently override evidence.
Red flags to watch for during the interview itself:
Vague "we" answers that never resolve into what the candidate personally did. Repeated blame of prior managers or teammates with no reflection on their own role. Numbers that shift when you ask the same question from a different angle. Refusal to name a real failure, or a "failure" that is really a humblebrag. Answers that describe only intentions and plans, never outcomes. Discomfort with follow-up probes about sequence, dates, or specific names.
Comparison of common question formats
| Format | What it measures | Predictive strength | Best use | Main risk |
|---|---|---|---|---|
| Behavioral (past situation) | Demonstrated actions and judgment | High when structured | Core competency evidence | Fabrication or rehearsal |
| Situational ("what would you do") | Reasoning and role knowledge | Moderate | Safety and protocol checks | Rewards articulate guessers |
| Hypothetical puzzle/brainteaser | Composure, cultural fit | Low | Rarely justified | Bias, no job relevance |
| Self-assessment ("rate yourself") | Self-image and confidence | Low | Almost none | Measures self-promotion |
| Work sample or exercise | Actual task performance | High | Skills-heavy roles | Setup cost and time |
| Reference-check style probes | Corroboration of claims | Moderate | Final-stage verification | Requires candidate consent |
Work samples and structured behavioral interviews are complementary, not substitutes. Use the work sample to see what someone can do and the behavioral questions to understand how they got there and how they work with others.
Timeline: how a well-run interview conversation unfolds
| Time (minutes) | Phase | Purpose |
|---|---|---|
| 0–3 | Opening and role framing | Set expectations and reduce nerves |
| 3–8 | Warm-up background question | Establish baseline and rapport |
| 8–30 | Core behavioral questions | Collect scored evidence per competency |
| 30–38 | Follow-up probes | Verify specifics: sequence, numbers, role |
| 38–43 | Candidate questions | Assess motivation and role understanding |
| 43–45 | Close and next steps | Confirm timeline and process |
| Within 24 hours | Independent scoring | Score before discussing with other panelists |
Scoring independently before discussion is the step most teams skip, and it is the one that protects against anchoring. If one interviewer shares an opinion before others score, the panel's independence collapses.
Practical actions: building your question set
Start by listing the four to six competencies the role genuinely requires, drawn from a job analysis rather than from the last person who held the job. For each competency, write two behavioral questions and a three-to-five point anchored rubric describing what a weak, adequate, strong, and exceptional answer looks like.
Draft questions that begin with "Tell me about a time when" and specify a situation type. Then write two follow-up probes per question, such as "What was your specific contribution?" and "What was the measurable result?" Rehearse the full set on a colleague playing a candidate and time it.
Use concrete thresholds to keep the process honest. Cap the core question set at 8, cap each interview at 60 minutes, and require scores before any group discussion. If two calibrated interviewers differ by more than one point on a rubric scale, flag the case for a structured debrief rather than averaging silently.
Blind the first review where feasible. Hiding names and schools during initial scoring reduces affinity bias, which is well documented in hiring research. Even partial blinding improves consistency.
Plain language: write the questions and the scoring guide before you meet anyone.
Limitations and what interviews cannot tell you
Interviews measure candidates' ability to describe past behavior and reason aloud. They do not directly measure on-the-job performance, and any single interview is a small sample. Predictive validity is real but far from perfect; structured interviews are among the better predictors, not a guarantee.
Motivation, adaptability to your specific team, and long-term retention are weakly captured by any question set. Work samples, structured reference checks, and trial projects add signal that interviews miss. No combination eliminates the cost of a bad hire.
Scores also drift over time as interviewers recalibrate informally. Refresh your rubric and run periodic calibration to keep agreement from decaying. Store scored notes for auditability and consistency.
Finally, decline to rely on questions that touch protected characteristics. In the US, keep every question job-related and apply it uniformly to every candidate. If a question would feel inappropriate asked of you, it likely carries legal risk for the employer.
This article is general educational information about interview design, not legal, HR compliance, or employment-law advice. It does not promise hiring outcomes or predict individual performance. Consult a qualified employment attorney or human-resources professional in your jurisdiction before changing hiring practices or relying on these methods for a specific decision.