A structured interview is one where every candidate for a role answers the same job-related questions, in the same order, and every answer is scored against the same rating scale. That structure makes candidates comparable, and it is why structured interviews predict job performance better than a free-flowing conversation.
The US Office of Personnel Management puts it in one line: a structured interview "uses a standardized questioning and scoring process for all candidates." Everything else in this post is detail on how to do that well.
How is it different from a normal interview?
In an unstructured interview, each interviewer picks their own questions, follows the conversation where it goes, and forms an overall impression at the end. Two candidates for the same job can get completely different interviews, and the final call often comes down to who the interviewer liked.
A structured interview fixes four things in advance:
The questions. Each one is tied to a skill or behaviour the job needs, and every candidate gets the same set. Follow-up probes are planned too, so interviewers dig into the same areas for everyone.
The order. Asking questions in a set sequence keeps the experience the same from one candidate to the next.
The scoring. Each answer is rated on a scale with written examples of what a weak, acceptable and strong answer looks like. OPM's guidance on building a rating scale is to decide the number of proficiency levels, then write example behaviours for each level with input from people who know the job.
The record. Answers are noted or recorded so another reviewer can check the score later. Google's re:Work guide lists detailed recording of answers alongside vetted questions, standard rubrics and interviewer calibration as the parts of its approach.
Does a structured interview predict job performance?
The research says yes, more than most people expect. In 2022, Paul Sackett and colleagues re-ran widely cited meta-analyses of hiring methods in the Journal of Applied Psychology. They argued that earlier studies had over-corrected for range restriction, which inflated some well-known validity figures. With that fixed, structured interviews came out with the highest mean operational validity of the methods reviewed, at .42, ahead of job knowledge tests (.40), work samples (.33) and general cognitive ability (.31), as summarised by SIOP and HumRRO.
A validity of .42 is far from perfect, and the range across studies is wide. But it means the interview many teams already run can carry real predictive weight, if it is built with structure rather than run as a chat.
Google reports the same pattern internally. Its re:Work guide says structured interview scores predicted job performance better than unstructured interviews across functions and levels, that prepared questions and rubrics saved interviewers an average of 40 minutes per interview, and that rejected candidates who had a structured interview were 35% happier with the process than those who did not. Those are Google's own figures from its own hiring, so treat them as one employer's experience rather than a benchmark.
Why does structure matter more in 2026?
Two things have changed. First, candidates now prepare with AI. Polished, generic answers to "tell me about yourself" are easy to produce, and some candidates use answer copilots during live calls. Structured interviews hold up better because the planned follow-up probes ask for specifics that a script does not contain. Our post on stopping candidates cheating with AI in interviews covers this in more detail.
Second, AI interviewers are going mainstream, and they are structured by design. In May 2026 Greenhouse completed its acquisition of Ezra AI Labs and now describes its Voice AI as a way to "ask the same role-specific questions and evaluate candidates against the same criteria." That is a description of a structured interview. If an AI runs your first round, the quality of your question bank and rubric decides the quality of the output.
Structure also helps with compliance. When a regulator, an auditor or a rejected candidate asks why a decision was made, a scored answer to a job-related question is far easier to explain than an impression. Our 2026 guide to AI interviewing law explains why documented, consistent criteria matter.
How do you set up a structured interview?
Start from the job, not from a list of favourite questions. Pick four to six skills or behaviours that separate strong performers in the role from average ones.
Write one or two questions per skill. Mix behavioural questions ("tell me about a time you had to recover a slipping project") with situational ones ("what would you do if a client changed the scope a week before launch"). Plan two or three follow-ups for each.
Write the rating scale before the first interview. For each question, describe what a weak, acceptable and strong answer contains. Keep the scale short enough that interviewers will use it.
Score each answer straight after it is given, then discuss. Panel members should rate independently before comparing, so the most senior voice does not set everyone's score.
Calibrate. After the first few candidates, compare scores across interviewers and tighten any question where ratings drift apart.
Keep the human decision. The scores inform the hiring decision. A person who has read the answers makes it.
How GetHirePlus helps
GetHirePlus is built around structured rounds. AI phone interviews ask every candidate for a role the employer's questions and return a transcript with per-question scores against the same rubric. Live video interviews include panels, scorecards and recording, so human rounds follow the same structure. Candidate assessments add a work-based test where the job needs one, and the applicant tracking system keeps every score on one candidate record. The AI scores and summarises; recruiters make the decision.
To estimate what your current process costs, try the free cost per hire calculator.
Sources
Google re:Work, A guide to structured interviewing for better hiring practices
Greenhouse, Greenhouse acquires Ezra AI Labs (Voice AI), May 2026