How to structure an interview: an 8-step playbook

James Hitch

· 10 min read

How to structure an interview: an 8-step playbook

To structure an interview, decide what you are measuring before you meet anyone. Then ask every candidate the same job-related questions in the same order, take notes on what they say, and score each answer against a written scale before anyone compares opinions. That is the whole method in two sentences. The rest of this guide covers the detail: competencies, questions and probes, a usable scorecard, the clock, and a debrief where the loudest person in the room does not set the score.

An analysis of 5 search-ranking guides on how to structure an interview (Indeed, Workable, Criteria, VidCruiter, Michael Page) found that 0 of 5 give a minute-by-minute interview timeline or address staffing and high-volume hiring. Only 1 of 5 has interviewers score independently before discussing.

What interview structure actually means

Two different things get called structure, and it pays to keep them apart.

Structure of the conversation is the running order: welcome, background, questions, the candidate's questions, next steps. It keeps the meeting tidy and helps the candidate relax. On its own it does little for the quality of the decision, because two interviewers can follow the same agenda and still ask different things and reward different answers.

Structure of the measurement is what makes candidates comparable. The US Office of Personnel Management (OPM) sets out the standard in its structured interview training guide, which is older but remains one of the clearest free templates available. Four features define it:

  • Questions are built from job-related competencies.
  • Every candidate gets the same predetermined questions, in the same order.
  • Trained interviewers take detailed notes on the responses.
  • Every response is evaluated on the same rating scale, with the same standards for an acceptable answer.

If you only have the agenda, you have a well-run conversation. If you have all four, you have a measurement you can defend. Plenty of teams sit in between, using a fixed core of questions with some free time at the end. That semi-structured format is a reasonable place to start, as long as the free time is left out of the score.

Why structure is worth the effort

The strongest reason comes from a re-analysis of decades of selection research. Sackett, Zhang, Berry, and Lievens corrected what they describe as flaws in how earlier meta-analyses adjusted for range restriction, and their published table of revised estimates put structured interviews at the top of the list of widely used predictors.

Selection methodMean operational validity
Structured interviews.42
Job knowledge tests.40
Empirically keyed biodata.38
Work sample tests.33
General mental ability tests.31
Unstructured interviews.19
Years of job experience.07

Read those numbers with care, because the authors do. Validity for structured interviews varies widely between studies, with an 80% credibility interval of .18 to .66. In the paper's own words, it should really be viewed as ".42, plus or minus .24". The authors name one possible contributor to spread of this kind: a single label can cover measures that are really quite different.

The usable reading is simple. Calling your process structured does not earn you the .42. The design and the discipline decide where in that range your interviews land, and the eight steps below are where that gets decided.

An interviewer in a cream sweater writing with a pen in an open notebook on a wooden table, with a coffee mug nearby and another person blurred across the table

How to structure an interview in eight steps

These steps follow the sequence in OPM's guide, adapted for a team without an industrial psychologist on call. The first role takes the most work. Later roles in the same family can reuse most of it.

Step 1: Write down what the job requires on day one

Start with the job's real tasks and responsibilities, then list what someone needs to do them. Ask the hiring manager and one person who does the job today. OPM's test is useful here: focus on competencies that are required upon entry. Anything you plan to teach in the first few months does not belong in the interview, however nice it would be to have.

Step 2: Pick four to six competencies

OPM notes that structured interviews typically assess four to six competencies, unless the job is unique or very senior. It also lists where the format is well suited: teamwork, oral communication, interpersonal skills, conflict management, and influencing or negotiating.

Be strict. Eight competencies in 45 minutes means shallow evidence on all eight. If a skill can be checked by a test or a work sample, check it there and give the interview time to something a test cannot see. Then write a one-sentence definition of each competency, so that "communication" means the same thing to every interviewer.

Step 3: Write behavioral and situational questions

Two question types carry the load. Behavioral questions draw on what a candidate actually did in past experience. Situational questions present a realistic job scenario and ask what the candidate would do. Write them to be open-ended, clear, and free of jargon, and tie each one to a single competency.

Here is what that looks like for a recruiter role:

CompetencyBehavioral questionSituational question
Conflict managementTell me about the last time a hiring manager rejected a candidate you strongly believed in. What did you do, and what happened?A hiring manager has turned down six candidates and blames your sourcing. You think the brief is the problem. How do you handle the next conversation?
PrioritizationDescribe the worst week you have had juggling open roles. What did you drop, and how did you decide?It is 4 p.m. You have ten callbacks, two offers to chase, and a new urgent role. What do you do first, and why?

OPM also recommends superlatives such as "last", "worst", and "most", which point a candidate toward one specific event. Good questions should draw out three things: the situation or task, the action taken, and the result. Have a colleague read each question without its label and name the competency it tests. If they guess wrong, rewrite it.

Step 4: Write your probes before you need them

A probe is a follow-up that gets missing information out of an answer. Unplanned follow-ups are an easy place for structure to leak, because an interviewer can end up digging harder with the candidates they already like. So decide beforehand whether probes are allowed, and make a list of acceptable ones for each question.

  • Situation: Who was involved? What led up to this?
  • Action: What was your role? How did you respond?
  • Result: What was the outcome? Would you do anything differently?

OPM adds two rules worth keeping. Ask for clarification when a candidate says "we did" or "our group", because you are hiring the individual. And never use a leading probe that signals the right answer.

Step 5: Build a scorecard with written anchors

A scale fails when its numbers mean nothing. A 1 to 5 scale running from "very unsatisfied" to "very satisfied" records how an interviewer felt, which is the thing you were trying to get away from. OPM's guidance is to choose one range for all competencies, typically between three and seven levels, label at least three of them, and write example answers for each level. For more on the mechanics, see this guide to building an interview score card.

Here is an illustration for the conflict question above. Write your own against your own job:

ScoreWhat the answer sounds like
1: UnsatisfactoryGives no specific event, blames the manager, or dropped the candidate without a conversation.
3: SatisfactoryDescribes a real event, raised the disagreement directly, and backed it with some evidence. The outcome is stated.
5: SuperiorDescribes a real event, found out what sat behind the rejection, changed the brief or the search as a result, and can say what changed afterward.

Get the anchors from people who know the job. OPM's method is to have each subject matter expert write example responses alone, then discuss until they agree on the most representative answer for each level. On weighting, keep it simple: OPM calls equal weights across competencies generally the most effective and defensible approach, and asks for a documented reason if you use anything else. A related guide covers using rubrics for interviews in more depth.

Step 6: Set the running order and the clock

Every candidate should get the same amount of time. Use this as a starting template for a 45-minute interview with five questions. The timings are a suggestion to adjust, and the fixed part is that they stay the same across candidates once set.

MinutesSegmentWhat happens
0 to 5WelcomeIntroductions, how the interview works, note-taking explained
5 to 35Scored questionsFive questions at about six minutes each, including probes
35 to 42Candidate's questionsUnscored, with room for an honest talk about the role
42 to 45Wrap-upThanks, next steps, and when they will hear back
AfterRatingBlock 15 minutes straight after, before the next meeting starts

That last row is the one teams skip. Book it with the interview itself.

Step 7: Assign roles and write the interviewer's guide

Name a chair. On OPM's model, the chair handles logistics, opens and closes the interview, keeps it on topic, and runs the consensus step. Assign each remaining interviewer their questions, so nobody improvises on the day.

Then put everything into one short interviewer's guide: how the process runs, the competency definitions, questions and rating scales with example responses, the allowed probes, and the tips and practices to avoid. Before it goes live, pilot it on a colleague. You will find the question that takes twelve minutes and the one nobody understands. The pilot is also the time to settle how final ratings will be reached: consensus, majority, or average.

Step 8: Keep a record of how you built it

Save who helped, the materials used, and how the questions and scale were developed. The next hiring manager can then reuse the kit, and you can explain your process if a decision is ever questioned.

How to run the interview once it is built

A structured interview should still feel like a conversation with a person. OPM's guide treats rapport as part of the job: greet people properly, use their name correctly, make eye contact, and thank them for coming. Candidates are evaluating you at the same time.

Open by explaining the format. Something like: "I'm going to ask you five questions, the same ones we ask everyone. I'll take notes while you talk, so if I look down, I'm still listening. You'll have time for your own questions at the end."

Ask the questions as written, in order. Interviewers ask only their assigned questions. Probe when an answer lacks a situation, an action, or a result, and watch that your tone and body language do not hint at the answer you want.

Control the time politely. Some candidates will talk for nine minutes on question one. OPM is direct about it: tactful interruption is the chair's job when someone runs long or goes off topic. Try: "That's helpful, and I want to be fair to your time. What was the result?" Doing this for every candidate is fairness, because the person who rambles should not get more airtime than the person who answered clearly.

Take notes on the answer, not the person. OPM's rule is that notes should summarize the actual response and hold enough detail to justify a rating. They should not contain judgments about the candidate's personality or mention demographic characteristics. "Called the manager the same day, asked for three rejected resumes, and rewrote the brief" is a note. "Seemed confident, great energy" is a verdict with nothing behind it.

Close cleanly. Thank them, ask them to keep the details confidential, and explain the next steps. Then keep your word on the timeline.

Two colleagues sitting at a table by a window with open notebooks, one pointing at a page while the other listens

How to score and debrief without groupthink

The order of events matters, because the first opinion spoken can color everyone else's. OPM's sequence is built to prevent that:

  1. Share the evidence. Right after the interview, panelists read their notes to one another. This is what the candidate said, with no scores yet.
  2. Rate alone. Each interviewer rates every question and competency individually, checking the definition, the example responses, and their notes before choosing a score.
  3. Compare. Ratings go onto one form and the panel checks them against the agreement rule set in advance.
  4. Discuss only the gaps. Where ratings disagree, each interviewer explains their rating using their notes until the panel reaches the required level of agreement. Any change gets initialed.

Then watch for the errors OPM lists. They are worth printing on the rating form:

Rating errorWhat it looks like
Similar to meHigher ratings for someone with your school, your hometown, or your sense of humor
HaloA strong answer on one competency lifts the scores on the others
Leniency or strictnessOne interviewer rates everyone high, or everyone low
Central tendencyEverything lands on 3 on a 5-point scale
First impressions and candidate orderThe first five minutes, or whoever came before, sets the score

Questions to keep out of the room

A fixed question set protects you here, because the risky questions tend to appear in the improvised small talk. OPM tells interviewers never to ask or comment on a candidate's age, sex, race, national origin, religion, marital or family status, disability or health status, or any other job-irrelevant factor.

On disability specifically, the US Equal Employment Opportunity Commission (EEOC) says an employer may not ask an applicant whether they have a disability or require medical questions before making a job offer, though it may ask whether they can perform the job and how they would do it. The EEOC's broader guidance on tests and selection procedures is also worth reading before you lock in a process. Local rules vary, so have employment counsel review your kit. This section is general information and is not legal advice.

Also make the setting accessible, and brief your interviewers on what to do if a candidate volunteers personal information: move on, and leave it out of the notes.

Adapting the structure for staffing desks and high volume

The OPM model assumes a panel and plenty of preparation time. A recruiter running ten screens a day has neither. The principles hold, and the format shrinks.

A recruiter wearing a headset, talking on a call at a tidy desk beside a window and writing in a notebook
  • Build one kit per role family. Many desks fill the same few job types again and again. A kit for warehouse supervisors or registered nurses gets reused for every job order of that type, with one or two questions swapped for the specifics.
  • Shrink the screen, keep the rules. A 20-minute call can cover two or three competencies with fixed questions and a short scale. The phone screening questions you already use are a good starting point, once each one has anchors.
  • Be your own panel. A solo screener has no second opinion, so score right after the call and before the next dial. Recording, with proper consent, lets you check your notes later.
  • Use one-way video for the first round. Recorded answers suit this format well, since every candidate gets the same question, in the same order, with the same time limit. Reviewers can then score against anchors in their own time.
  • Write the submittal from the evidence. A scorecard gives a recruiter specific examples to put forward: what the candidate did, in their words, against what the role needs.

Mistakes that quietly undo the structure

  • Changing questions mid-search. The candidates who came before and after the change can no longer be compared. If a question is truly broken, change it, say so in the record, and treat the two groups separately.
  • Treating a culture chat as data. Unscored conversation is fine. Letting it decide the outcome brings the unstructured interview back in through a side door.
  • Skipping interviewer training. OPM's model specifies trained interviewers. A session on the kit, a practice round, and a look at the rating errors above is a sensible minimum.
  • Never checking the results. Look back at who you hired and how they scored. If top scorers struggle in the job, the anchors need work, and the .18 to .66 spread in the research is a reminder that the design has to earn its place.

How Hivemind handles structured interviews

A structure only lasts if the system people work in keeps it in place. Hivemind.HR is the hiring OS: it can run hiring from requisition to offer, or connect to the tools a team already has. For the process in this guide, these are the concrete steps it runs:

  • Pipeline builder. A recruiter drags steps onto a canvas: score the resume, run an assessment, request a video answer, book an interview, wait a set time, or branch on a score. Every candidate on that pipeline follows the steps the team set.
  • AI-graded async video. Candidates record answers on their own time. AI transcribes, summarizes, and scores them, so the team reviews a shortlist. Reviewers can read the transcript against their own anchors.
  • Assessments. Skills checks with scoring rubrics sit beside the application, which frees interview time for the competencies a test cannot see.
  • Live interview support. Scheduling a manual interview creates the Zoom or Google Meet link and sends it to both sides. Video call intelligence takes notes and writes a summary of the live call, and the recording, notes, and scores sit together on the candidate's profile, so the panel reviews against the same anchors in one pass. For how booking automation works in general, read the Hivemind guide to automated interview scheduling.
  • Automatic candidate updates. Emails and texts fire from pipeline events, including interview reminders and status changes, so the candidate hears what comes next.

The team still chooses the competencies and writes the anchors, and a person makes the hiring decision. A team can try this on a real role with a 28-day free trial.

Frequently asked questions

What are the main parts of a structured interview?

Job-related competencies, the same predetermined questions in the same order, planned probes, detailed notes on each answer, and one rating scale with written examples for each level. A set running order and a fixed time per candidate hold it together.

How many questions should a structured interview have?

Work back from your competencies and your clock. With four to six competencies, one or two questions each is a sensible range. At about six minutes per question, five questions fill 30 minutes. Fewer questions with real probing beat a long list answered on the surface.

What is the difference between structured, semi-structured, and unstructured interviews?

A structured interview fixes the questions, the order, and the scoring. A semi-structured interview fixes a core set and leaves room for free discussion. An unstructured interview leaves the questions and the scoring to each interviewer. In the revised estimates from Sackett and colleagues, unstructured interviews had a mean validity of .19, against .42 for structured.

Can I ask follow-up questions in a structured interview?

Yes, if they come from a list you agreed in advance and you offer them on the same basis to every candidate. Use them to get the situation, action, or result that an answer left out, and keep them free of hints about the right answer.

Do structured interviews feel robotic to candidates?

They can if the interviewer reads like a script and skips the human parts. Explain the format at the start, use the candidate's name, make eye contact, and leave real time for their questions. Fair treatment and clear next steps also tell a candidate how your company operates.

Start with one open role

Do not rebuild every process at once. Take one role you are hiring for now, agree four competencies with the hiring manager, write a question and a set of anchors for each, and use the kit on the next five candidates. Then look at where your interviewers disagreed, because those gaps show you which anchors to rewrite. Once that kit holds up, copy it to the next role family.

If you want to see the process running as a pipeline on one of your own roles, book a call.

Switch on Hivemind.

Where Hivemind fits

Hivemind runs these steps for you, in one platform.

Written by

James Hitch

COO

James Hitch is COO of HiveMind, the hiring OS for modern staffing firms. He leads the company’s day-to-day operations, aligning teams across product, growth, customer success, recruiting, and internal systems to turn strategic priorities into consistent execution. His writing translates those operational conversations into clear, practical guidance for staffing teams: how to unlock value from an existing candidate database, reduce time to first submittal, build meaningful desk-level metrics, and apply AI where it materially improves speed, judgment, and recruiter capacity.

More from James Hitch

All 36 articles →