05 Hiring Leadership
Questions That Predict
Performance
Good interviews ask better questions.
The question system
Question
Designed to surface specific signal
Signal
Observable evidence of capability
Follow-up
Probes depth and consistency
Evidence
Concrete basis for decision
The quality of a hiring decision is upstream of the quality of the questions that produced the evidence.
Why this matters
Most interview questions generate conversation, not evidence.
The default interview question is abstract: 'Tell me about yourself.' 'What's your greatest weakness?' 'Where do you see yourself in five years?' These questions produce answers. They rarely produce evidence of how a candidate will perform in the specific role being filled.
Performance-predicting questions are different in structure. They ask for specific past behaviour — because past behaviour in comparable situations is the most reliable predictor of future behaviour available. They probe for context, decision, action, and outcome. They follow up on vagueness with specificity.
The difference between a good interview and a weak one is often not the interviewer's instinct — it's the quality of the questions they prepared. Strong questions produce strong evidence. Strong evidence produces better decisions.
Founder reality
Evaluate the questions in your current interview process:
Could a strong candidate answer our interview questions without revealing anything specific about their capability?
Are our questions role-specific — or could they be asked in any interview for any role?
When candidates give vague answers, do we have follow-up questions ready that push for specificity?
Are we asking for past behaviour — or for hypothetical responses to hypothetical situations?
After the interview, can the panel point to specific evidence for each capability signal they're evaluating?
If interviewers can't recall specific evidence after the conversation, the questions didn't produce what was needed.
The framework
Five question types that predict performance
Use each type for specific capability signals. Combine them across a structured interview.
01
Behavioural — specific past situations that reveal how the candidate actually operates
'Tell me about a time you had to make a technical decision with incomplete information. What was the situation, what did you decide, and what happened?' Behavioural questions are the gold standard because they require real examples — not hypothetical responses. Follow up with 'what specifically did you do' to prevent abstract storytelling.
02
Problem-solving — structured problems that reveal how they think under observation
A system design question, a data analysis prompt, or a product decision scenario — evaluated not for the correct answer but for how the candidate approaches ambiguity, asks clarifying questions, and communicates reasoning. The process matters more than the conclusion.
03
Ownership — questions that surface accountability versus deflection patterns
'Tell me about a project that didn't go as planned. What happened and what was your role in the outcome?' Strong candidates own outcomes — including difficult ones. Candidates who deflect to external factors or team failure without personal accountability signal something specific about how they operate.
04
Learning — questions that reveal how the candidate develops over time
'What's the most significant thing you've changed your mind on professionally in the last two years, and why?' This question separates candidates with genuine intellectual growth from those with fixed mental models. The quality of the answer reveals both self-awareness and learning velocity.
05
Role-specific signal — questions built for the specific challenges this role will face
The most predictive questions are role-specific. For a distributed systems engineer: 'Walk me through a consistency vs availability trade-off you navigated in production.' Generic questions produce generic signal. Role-specific questions produce actionable evidence.
Common mistakes
01
Asking hypothetical questions instead of behavioural ones
'What would you do if...' produces the answer the candidate thinks you want. 'Tell me about a time when...' produces actual behaviour. The past is more reliable than the hypothetical.
02
Accepting vague answers without follow-up
When a candidate says 'we built a system that scaled well', the signal is in the follow-up: 'What specifically was your contribution to the architecture decision?' Vague answers accepted without follow-up produce vague evidence.
03
Using the same questions for every role
A senior infrastructure engineer and a product designer require entirely different signal sets. Generic interview questions produce generic signal — which is not useful for specialist hiring decisions.
04
No question-to-capability mapping
Every interview question should map to a specific capability being evaluated. Questions without a mapped capability are conversation, not evaluation.
Example scenario
An ML infrastructure company, hiring a principal engineer. Previous four interviews ended in no-hire decisions — but the panel couldn't articulate why. Interviews felt 'off' but produced no specific evidence.
The question audit
Interview questions reviewed: 8 of 12 were hypothetical or generic. 'Tell me about your experience with distributed systems' — produces a narrative, not evidence. 'What would you do if two engineers disagreed on architecture?' — produces a hypothetical, not behaviour.
The redesigned question set
Behavioural: 'Tell me about a specific architecture decision you made that you later reversed. What was the decision, why did you make it, and what caused you to change it?'
Ownership: 'Tell me about a production incident you were responsible for. Walk me through what happened, what your role was, and what changed as a result.'
Role-specific: 'Describe a time you had to design for both consistency and low latency in the same system. What trade-offs did you make?'
Learning: 'What's a strongly-held technical belief you've revised in the last 18 months?'
The outcome
Next interview panel could articulate specific evidence for every signal after the conversation. Two strong candidates identified within 3 weeks. One hire made with unanimous panel confidence.
Takeaway
Interview quality is question quality.
Design questions that require specific evidence. Follow up on vagueness. Map every question to a capability. The panel's ability to make a confident decision at the end is a direct function of the evidence the questions produced.
Related thinking
Hiring Leadership
Interview Signals vs Gut Feeling
Confidence is not capability.
Hiring Leadership
Structured Interviews Without Feeling Robotic
Structure improves conversations — not replaces them.
Hiring Leadership
Interviewing Without Guessing
Better hiring starts with better signals.
Continue reading
Interviewing Without Guessing