Structured Interviews: Why They Predict Performance and How to Run Them at Scale

A structured interview asks every candidate for a role the same job-related questions, in the same order, and scores each answer against the same written rating scale. It's one of the most dependable ways to predict job performance: a 2022 re-analysis of decades of hiring research ranked structured interviews first among the selection methods it reviewed.

This guide covers what makes an interview structured, what the research says, a complete six-question example with scoring anchors, and how to keep the structure when you're hiring at volume.

What is a structured interview?

A structured interview fixes three things before the first candidate arrives: the questions, the order they're asked in, and how the answers will be scored. The U.S. Office of Personnel Management's guidance on structured interviews describes structure as rules for eliciting, observing and evaluating responses, and notes that those rules raise agreement between interviewers by limiting how much each one is left to improvise.

In practice, the difference looks like this:

Most real interviews sit somewhere in between. A semi-structured interview keeps the core questions and the scoring fixed and lets the interviewer follow up more freely. Partial structure still helps, as the research below shows.

Why do structured interviews predict job performance?

Two large research summaries are worth knowing.

In 1998, Frank Schmidt and John Hunter summarized 85 years of research on hiring methods in Psychological Bulletin. Their abstract reports that a test of general mental ability combined with a structured interview was one of the three most predictive combinations they examined, with a mean validity of .63. Validity here is the correlation between a method's results and later job performance, and that figure belongs to the combination, not to the interview alone.

In 2022, Sackett, Zhang, Berry and Lievens revisited those estimates in the Journal of Applied Psychology. They concluded that earlier meta-analyses often overcorrected for range restriction (job performance can only be measured for people who were hired), so the validity of many selection methods had been substantially overestimated. The methods that ranked high before mostly still did, with mean validity estimates .10 to .20 lower, and structured interviews emerged as the top-ranked selection procedure. Their bottom line: selection methods remain useful, but their link to job performance is considerably weaker than previously thought. Our evidence-based hiring playbook turns these findings into seven steps.

Why does structure help? The mechanics aren't mysterious:

• The questions come from the job, so the answers are evidence about the job.
• Everyone gets the same questions, so their answers can be compared.
• Written anchors make one interviewer's 4 mean the same as another's.
• Notes on what was said give the decision something to rest on besides memory.
• Independent ratings stop the loudest voice in the debrief from setting everyone's score.

Michael Campion, David Palmer and James Campion's 1997 review in Personnel Psychology identified fifteen components of structure, some about the content of the interview and some about how answers are evaluated. They concluded that interviews can be improved by using even some of them, which matters for teams that can't change everything at once.

None of this turns an interview into a precise instrument. A structured interview is better evidence than a conversation, not a forecast.

What are the parts of a structured interview?

A job analysis

Before writing a question, work out what the job requires: the tasks, and the competencies someone needs on day one to do them. OPM's structured interview training starts here: review the job's tasks, identify the competencies required on entry, and confirm them with people who know the job. It adds that structured interviews typically assess four to six competencies. That list becomes both the question plan and the scorecard. Here's how to turn a job description into interview questions.

The same questions, in the same order

Every candidate for the role hears the same questions in the same sequence. Fixing the order means everyone gets the same warm-up and the same amount of time before the hard questions.

Behavioral and situational questions

Behavioral questions ask what the candidate did in a past situation: "Tell me about a time…". Situational questions describe a realistic scenario from the job and ask what they would do. OPM's guidance treats both as effective and builds both from real incidents on the job. Behavioral questions show what someone has actually done; situational questions test judgment on your problems, including ones a candidate hasn't faced yet. A good structured interview uses both.

Planned follow-ups

Follow-ups are allowed if they're planned. OPM's training says to decide in advance whether probes are allowed, to list the acceptable ones for each question, and to avoid leading probes that hint at the answer. Neutral ones such as "What was your role?" and "What was the outcome?" work for almost any question. Our list of probing questions groups 40 of them by what an answer leaves out.

Anchored rating scales

Each answer is rated on a fixed scale, usually 1–5, against written descriptions of what a weak, a solid and a strong answer sounds like. The interview scorecard template shows how to write these anchors, with examples for three roles.

Notes on what the candidate said

The interviewer writes down what the candidate said, not what they thought of it. "Moved two pickers to packing and shipped by 1:45" is a note. "Strong leader" isn't.

Independent ratings before discussion

Each interviewer scores alone before the panel talks. OPM describes the usual panel method the same way: members rate individually, then discuss significant differences as a group. A score said out loud first pulls everyone who hasn't written theirs yet.

A structured interview example: warehouse shift supervisor

Here's a complete structured interview for a warehouse shift supervisor who runs a day-shift team of about 15. The job analysis produced three competencies, each tested with one behavioral and one situational question.

Planned follow-ups, the same for every candidate: "What did you say to them?", "What did you do first?", "What happened next?" and "What would you do differently?"

The anchors for each competency apply to both of its questions.

Safety (questions 1 and 2)

• 1 – Weak: no real example, or treats the shortcut as acceptable when the team is behind; blames the person without asking why it happened.
• 3 – Solid: stops the unsafe act, explains why, has the job done the right way, and reports it as the site requires.
• 5 – Strong: stops it at once, even under time pressure; talks to the person privately; asks why the shortcut was tempting, such as a missing lift truck or an unrealistic target, and fixes that; follows up with the whole team.

Running the shift (questions 3 and 4)

• 1 – Weak: jumps in to do the work themselves or waits for a manager to decide; no plan, and nobody is told.
• 3 – Solid: re-plans the work around the people available, puts the urgent order first, and tells their manager early.
• 5 – Strong: re-plans in minutes using each person's skills, borrows help from another area, protects the 2 p.m. order with checkpoints through the morning, warns the people who need to know if anything will slip, and reviews afterward what to change.

Leading people (questions 5 and 6)

• 1 – Weak: avoids the conversation, or handles it publicly and angrily; no follow-up.
• 3 – Solid: talks to the people involved privately, is specific about the standard or the behavior, agrees on next steps and checks back.
• 5 – Strong: acts early, listens before judging, separates the issue from the person, sets a clear expectation with a date, offers support, follows up, and can describe how it ended either way.

Each interviewer rates every answer alone and writes down the evidence before the debrief. With two questions per competency, decide before the first interview whether you'll average the two scores or give one score drawing on both answers.

How do you run structured interviews at scale?

Build a question bank

Write questions once, by competency, and reuse them. Within a role, every candidate gets the same questions; across roles, the bank saves you from starting over. The 60 competency-based interview questions make a starting bank, and the free interview question generator builds a set from the competencies you pick.

Lock the scorecard before the first interview

Competencies, anchors and weights are fixed before anyone meets a candidate. Changing them halfway through a search makes early and late candidates impossible to compare.

Train interviewers on the scale, not just the questions

An hour on the anchors does more than an hour on the questions. Have every interviewer score the same sample answer, compare, and talk through the gaps. OPM's training names the rating errors to watch for: putting everyone in the middle of the scale, rating everyone high or everyone low, and the halo effect, where a rating on one competency colors the others.

Let an AI interviewer take the first round

Volume is where structure tends to slip: a dozen interviewers will ask the same question a dozen ways, and nobody has time to write notes on the 40th screen of the week. An AI interviewer asks every candidate the role's questions and records what they said.

DIO, the AI interviewer Dio makes, holds a structured voice interview with each candidate in the browser. The recruiter reviews and edits the interview plan before publishing, and every candidate for the role is interviewed from that plan, whenever suits them, with nothing to schedule. Most interviews take about 15 minutes. DIO asks a follow-up when an answer is thin or vague, rates each competency 1–5 against anchors written for the role, and has to quote the candidate's exact words for every score. Two different AI models grade each competency independently, and when the evidence isn't there, the score is left blank rather than guessed. A person on your team reviews the evidence and makes every hiring decision.

Check the numbers

Look at scores by interviewer and by question. A question that everyone scores 3 isn't telling you anything, and an interviewer whose scores never vary needs another look at the anchors.

What should structured interview software do?

Whatever you use, check that it:

• Keeps questions organized by competency and fixed for each role.
• Shows the anchors next to each question while the interviewer scores.
• Collects each interviewer's scores before showing anyone else's.
• Ties notes, or a transcript, to every score.
• Reports scores by interviewer and by question, so drift shows up.
• For an AI interviewer, shows the evidence behind each score, leaves a score blank when the evidence is missing, and leaves the decision to a person.

For a fuller vendor checklist, see how to choose AI interview software.

What are the common objections to structured interviews?

"It feels robotic"

OPM's guidance notes that interviewers and applicants tend to prefer less structured formats. The fix is in the delivery, not the structure: explain the format at the start, keep a warm tone, and leave time at the end for the candidate's own questions. Structure governs what you ask and how you score, not how friendly you are.

"I can tell in the first five minutes"

That first impression is exactly what the structure checks. OPM's training lists relying on first impressions among the things to avoid. If your read was right, the scores will agree with it.

"Senior candidates expect a conversation"

Then make the questions harder, not looser. Situational questions about real problems the role will face make for a good conversation, and they're still scored the same way for everyone.

"We don't have time to build one"

Three competencies, two questions each, and anchors at 1, 3 and 5 is an afternoon with the hiring manager, and you reuse it for every hire into that role.

"Candidates will rehearse their answers"

Many will, and that's fine: a prepared candidate gives you more evidence, not less. Planned follow-ups such as "What did you personally do?" and "What happened next?" get past the script.