How to Create Interview Questions That Show Who Can Actually Do the Job
Don't start with the question.
Start with what you need to prove.
Interview questions have never been easier to create.
Give AI a job description and you can have 20 of them in seconds.
But that creates a different problem.
Are they the right questions?
A question can sound thoughtful.
It can create a great conversation.
It can even produce an impressive answer.
And still tell you very little about whether this person can succeed in the role.
So before asking:
What questions should we ask?
Ask:
What do we need this candidate to prove?
That's where a better interview begins.
Questions are easy. Evidence is hard.
Consider this question:
"Tell me about a time you demonstrated leadership."
Sounds reasonable.
But what are you trying to learn?
How they make difficult decisions?
How they influence people?
How they develop others?
How they lead through uncertainty?
How they take accountability when something goes wrong?
"Leadership" can mean almost anything.
And when what you're assessing is vague, the evidence can be vague too.
Harvard Business School recommends choosing interview questions purposefully and considering what candidates' answers will actually reveal. Harvard Business School
So don't start by writing a leadership question.
Start here:
What does leadership need to look like in this job?
Then build the question.
Start with the job.
Suppose you're hiring a manager whose first year will require them to:
Turn around an underperforming team.
Gain the trust of skeptical stakeholders.
Make difficult decisions with incomplete information.
Now you have something useful.
You don't just need "leadership."
You need evidence of:
Developing performance.
Stakeholder influence.
Judgment under uncertainty.
McKinsey recommends defining the skills required for a role and using objective interviewing and assessment methods to evaluate those skills. McKinsey
The sequence matters.
Role
What success requires
What must be proven
Evidence
Question
The question comes last.
Decide what a strong answer would prove.
Suppose judgment under uncertainty matters.
You could ask:
"Tell me about a time you had to make an important decision without having all the information you wanted."
Better.
But you're not finished.
Before asking it, decide what evidence you're listening for.
Perhaps a strong answer would reveal:
What the candidate knew.
What they didn't know.
How they evaluated risk.
What information they prioritized.
What tradeoffs they considered.
What decision they made.
What happened afterward.
Now the question has a job to do.
A question without an evidence standard is just a question.
Research on structured employment interviews has examined not only question consistency, but also job-related question development, rating scales, probing, and how candidate answers are evaluated. Personnel Psychology Structured interviewing research
The goal isn't consistency for its own sake.
It's collecting evidence you can actually use.
Match the question to the evidence.
Not everything should be tested the same way.
Sometimes you want to know:
What has this person actually done?
A behavioral question can help.
"Tell me about a time you inherited an underperforming team. What did you do?"
Sometimes you want to know:
How would this person think through a situation they may not have encountered before?
A situational question can help.
"You inherit a team where two strong performers are undermining a major change. What would you do first?"
These aren't interchangeable.
Research comparing matched situational and past-behavior interview questions suggests the two formats should not simply be assumed to provide identical information. Human Resource Management Review
So the better question isn't:
Should we use behavioral or situational questions?
It's:
What kind of evidence do we need?
Then choose how to collect it.
Sometimes, don't ask a question at all.
Imagine you're hiring someone whose job requires them to:
Analyze a complicated problem.
Make a recommendation.
Defend that recommendation.
Respond when someone challenges their thinking.
You could ask:
"Tell me about a time you demonstrated analytical thinking."
Or you could give them a relevant problem.
Harvard Business School discusses work samples as a way to assess capabilities by evaluating candidates based on their work rather than relying only on how they describe their abilities. Harvard Business School
McKinsey also identifies skill-specific work samples as an objective assessment method that can complement interviews. McKinsey
Sometimes the strongest interview question is:
Show us.
Follow the evidence.
A candidate says:
"I led a major transformation across the organization."
Sounds impressive.
Don't move on yet.
What did you personally own?
Who resisted?
What did you do?
What changed because of your actions?
What didn't work?
What would you do differently now?
The purpose of a follow-up isn't to make the interview conversational.
It's to get underneath the claim.
Research on structured interviewing treats probing and follow-up practices as meaningful parts of interview design. Personnel Psychology
So when an answer sounds impressive but the evidence is thin:
Stay there.
Find out what actually happened.
Don't ask five questions that prove the same thing.
You have ten interview questions.
It feels comprehensive.
But look closer.
Questions 1, 4 and 7 all test communication.
Questions 3 and 8 both test collaboration.
Nothing meaningfully tests execution.
You have a long interview.
Not necessarily a complete one.
Think of the interview as an evidence map.
What matters?
Where will we test it?
What evidence will we look for?
What's still untested?
More questions don't make a better interview.
Better coverage does.
Ask every question for a reason.
Before a question makes it into the interview, make it pass five tests.
1. What does this question assess?
If you can't answer clearly, reconsider it.
2. Why does that matter for this job?
Tie it to actual success in the role.
3. What would strong evidence look like?
Know what you're listening for before the candidate answers.
4. What would weak or incomplete evidence look like?
Make the difference visible.
5. Is a question the best way to test it?
Sometimes a situation or work sample will tell you more.
If a question can't survive those five tests:
Why is it in the interview?
Don't write questions to create conversation.
Write them to create evidence.
A great conversation can still be a poor assessment.
The candidate can be engaging.
The interviewer can enjoy the discussion.
Everyone can leave feeling like they learned a lot.
Then the hiring meeting starts.
Someone asks:
What did we actually learn about whether this person can do the job?
And suddenly the notes feel thin.
McKinsey cautions against allowing assessment to collapse into generalized questions or whether an interviewer personally likes the candidate. Instead, it recommends questions designed to test the skills that matter for the job. McKinsey
That's the difference.
Conversation creates impressions.
Good interview design creates evidence.
You can have both.
Don't confuse them.
The test of a good interview question.
Before asking it, finish this sentence:
"We're asking this because we need evidence of..."
Can you finish it clearly?
Good.
Now finish another:
"And we'll know strong evidence when we hear..."
If you can't answer that one, you're not finished.
Now you have more than an interview question.
You know:
Why you're asking it.
What you're listening for.
How the answer will help you decide.
That's what makes the question useful.
A better way to build the interview.
The sequence is simple.
1. Define success.
What will this person actually need to accomplish?
2. Decide what must be proven.
What capabilities would give you confidence they can accomplish it?
3. Define the evidence.
What would you need to hear or see?
4. Choose how to test it.
Behavioral question?
Situational question?
Work sample?
Realistic challenge?
5. Build the question.
Now write it.
6. Know how you'll evaluate the answer.
Don't wait until the candidate speaks to decide what "good" sounds like.
And before the interview begins, ask one final question:
Does everything that matters get tested?
That's the interview.
The questions collect the evidence. The evidence helps you decide, which is covered in How Do You Know Which Candidate to Hire?
That's Evidence-Based Hiring™.
A better interview doesn't start with a list of better questions.
It starts with better thinking.
What matters?
What must be proven?
What evidence would prove it?
How should we test it?
Then write the questions.
Don't start with the question.
Start with what you need to prove.
Proof over persuasion.
Build an interview around what actually matters.
InterviewKits helps you define what needs to be proven first, then build interview questions designed to uncover the evidence you need to make the decision. Start with what belongs in an interview kit.
Build your first Interview Kit in about 90 seconds.
Start freeNo credit card required.
Research behind this guide
- Harvard Business School
Interview Strategies to Connect with a Wider Range of Candidates
Harvard Business School recommends standardized questions, purposefully selecting questions based on what answers will reveal, focusing on capabilities, and using work samples where appropriate.
View source (opens in a new tab) - McKinsey & Company
Taking a Skills-Based Approach to Building the Future Workforce
McKinsey recommends defining the skills required for roles and using objective assessment methods, including behavioral and situational questions, standardized scoring rubrics, and skill-specific work samples.
View source (opens in a new tab) - McKinsey & Company
Skills-Based Hiring: Right Person, Right Role
McKinsey discusses assessing job-relevant skills through specific questions rather than allowing interviews to fall back on generalized measures of whether the interviewer personally likes the candidate.
View source (opens in a new tab) - Personnel Psychology
The Structured Employment Interview: Narrative and Quantitative Review of the Research Literature
This major academic review examines decades of structured-interview research, including question design, situational and past-behavior questions, rating scales, probing, impression management, and interview structure.
View source (opens in a new tab) - Personnel Psychology
Structured Interviewing: Raising the Psychometric Properties of the Employment Interview
This foundational research describes a structured interviewing approach beginning with job analysis, followed by job-related questions, consistent administration, and anchored criteria for evaluating answers.
View source (opens in a new tab) - Human Resource Management Review
A Tale of Two Formats: Direct Comparison of Matching Situational and Behavior Description Interview Questions
This meta-analysis compares situational and past-behavior interview questions designed around matching job attributes and examines whether the two formats provide equivalent information.
View source (opens in a new tab)