The Uncomfortable Truth About Interview Questions
Most interview questions do not predict job performance. "Where do you see yourself in five years?" tells you nothing about how someone will handle a production outage. "What is your greatest weakness?" elicits rehearsed responses that reveal nothing genuine. Yet these questions persist because they feel like interviewing.
Industrial-organizational psychology has spent decades studying which questions actually work. The findings are clear, specific, and actionable.
The Two Question Types That Work
Behavioral Questions
Behavioral questions ask candidates to describe real past experiences. The premise is simple: past behavior is the best predictor of future behavior. The format is: "Tell me about a time when you [specific situation relevant to the job]."
Effective behavioral questions have three characteristics:
- Specificity: They target a precise competency, not a vague trait. "Tell me about a time you influenced a decision without authority" targets influence skills. "Tell me about your leadership style" targets nothing specific.
- Relevance: The situation described must actually occur in the job. Asking a junior developer about enterprise architecture decisions is testing hypothetical thinking, not real experience.
- Difficulty calibration: The question should be challenging enough to differentiate strong from weak candidates. If everyone can give a good answer, the question is not discriminating.
Situational Questions
Situational questions present a hypothetical scenario and ask how the candidate would handle it. They are particularly useful for roles where candidates may not have directly relevant past experience. The format is: "Imagine you are [specific scenario]. What would you do?"
The key is making the scenario realistic and detailed enough that the candidate must demonstrate actual problem-solving rather than generic platitudes. "How would you handle a difficult customer?" is too vague. "A customer's integration is down during their peak sales period, your engineering team says the fix requires 4 hours, and the customer is threatening to churn. Walk me through your first 30 minutes" forces genuine thinking.
Question Types That Do Not Work
Brainteasers
"How many golf balls fit in a school bus?" Google famously used these and famously abandoned them after their own research showed zero correlation with job performance. They measure one thing: how much a candidate has practiced brainteasers.
Self-Assessment Questions
"How would you rate your attention to detail on a scale of 1 to 10?" Nobody says 3. These questions invite socially desirable responses and provide no predictive signal.
Hypothetical Trait Questions
"Are you a team player?" "Do you work well under pressure?" Every candidate says yes. The question provides no mechanism for differentiating true from false claims.
Resume Walkthrough
"Walk me through your resume" wastes interview time gathering information you already have in writing. It rewards candidates who are good at self-narration, which is rarely a core job competency.
Designing Predictive Questions: A Framework
For each competency you are evaluating, follow this framework:
- Identify a critical incident. What specific situation in the job requires this competency? For a product manager, it might be "resolving conflicting priorities between engineering and sales."
- Formulate the question. "Tell me about a time when you had to resolve conflicting priorities between two stakeholder groups. What was the situation, what did you do, and what was the outcome?"
- Define the evaluation criteria. A strong answer includes: a clear description of the conflict, a specific action taken (not "I would..."), evidence of considering multiple perspectives, and a measurable or observable outcome.
- Write behavioral anchors. What does a 1, 3, and 5 response look like for this specific question?
The Compound Effect of Good Questions
Structured interviews are twice as predictive of job performance as unstructured ones (Schmidt & Hunter). But structure alone is not enough. The questions within the structure must also be predictive. Using the right question types within a structured framework compounds the validity gain.
Google's re:Work research confirmed this: structured behavioral interviews not only predicted performance better but also saved interviewers an average of 40 minutes per interview in preparation time. When you know exactly what to ask and how to evaluate the answers, the interview becomes more efficient and more effective simultaneously.
Putting It into Practice
Here is a concrete example for a customer success role:
Competency: Proactive risk identification
Behavioral question: "Tell me about a time you identified a client was at risk of churning before they raised concerns. What signals did you notice, what did you do, and what was the outcome?"
Anchor (score 5): Describes specific data signals (usage decline, support ticket patterns), proactive outreach with a tailored plan, and a measurable save (retained revenue, improved NPS).
Anchor (score 1): Cannot provide a specific example, describes reactive responses to explicit complaints, or gives a vague hypothetical.
Build Your Question Bank
Start by identifying the four to six competencies that matter most for your most common roles. Write two behavioral questions per competency. Test them with your current team to calibrate difficulty. Refine based on whether the questions actually differentiate strong from weak responses.
StormInterview includes research-backed question templates for dozens of roles, complete with behavioral anchors and scoring rubrics. Start your free trial and ask questions that actually predict success.