How to conduct an interview: structure, questions, scoring — and the write-up for the hiring team (2026)

To conduct an interview, run the same structure for every candidate: a short opening, the candidate's story, four or five competency questions with follow-ups, ten minutes for their questions, and a close that names the next step — then score against a rubric within the hour and write it up before the debrief. The structure minute by minute, the questions and the follow-ups that get the real answer, a scorecard you can copy, and what the write-up for the hiring team must contain. Also where the interview record lives in Teams, Zoom and Google Workspace, and how it comes back to a hiring team whose members read in different languages.

The Mind.com Team

  • Run a structured interview: the same five parts and questions for every candidate, one interviewer leading, one scoring, the rubric filled in within the hour.
  • The write-up is where interviews fail: scores from memory a day later drift toward the loudest interviewer; in Teams, Zoom and Google Meet it is one file in one language.
  • InterMIND keeps the interview in each member's language, recap, transcript and chat, next to those tools rather than instead of them. Candidates join free by link, no account; you pay only for active members.
How to conduct an interview: structure, questions, scoring — and the write-up for the hiring team (2026)

How to conduct an interview: structure, questions, scoring — and the write-up for the hiring team (2026)

An interview is a meeting with one purpose: to collect evidence about whether a person can do a job, in a form that other people can read and weigh. A good one is structured — the same parts, the same core questions and the same scoring for every candidate — because the point is comparison, and you cannot compare answers to different questions. A conversation that goes where it goes is pleasant for both sides and produces one thing: an impression.

This guide is the craft first — the structure minute by minute, the questions and the follow-ups that get past the rehearsed answer, a scorecard you can copy — and then the part most guides stop short of: the write-up for the hiring team, where it lives in the tools you already use, and what happens when the panel's members, or the candidate, read in different languages.

What an interview is for — and what it isn't

Three documents surround an interview, and the middle one is the only one that gets any attention:

WrittenWhat it holdsWho reads it
Scorecard, blankBefore the first interviewThe competencies the job needs, a question or two per competency, the scale and what each score meansEvery interviewer, before the round
Interview notesDuringWhat the candidate said, as close to their words as the note-taker can manage, with the question it answeredThe interviewer who wrote them, within the hour
Write-upWithin the hourA score per competency with the evidence behind it, a recommendation, the risks, the questions left for the next roundThe hiring team, before the debrief

The scorecard is written once per role, not once per candidate; it is what makes the interviews comparable. The write-up is the scorecard filled in. If your hiring debriefs run on "I liked her" and "I'm not sure about him", the cause is upstream — there was no card to fill in, so the panel is comparing memories.

An interview is not a chat to see whether you get along, not a sales pitch for the company, and not a test of composure under hostility. "Tell me about yourself, 20 minutes" is a slot. "Describe a time you took over a project that was behind — what you did in the first week, and what the result was — 8 minutes, scored on ownership" is an interview question.

The structure: a 45-minute interview, minute by minute

A structured interview has five parts, and the proportions matter more than the exact minutes. The timings below are for 45 minutes; a 30-minute screen keeps the same shape with three questions instead of five, and a 60-minute technical round adds a work sample in the middle.

  1. Opening — 3 minutes. Say who you are, what the role is in one sentence, how the next 45 minutes will go, and that you will be taking notes. If the call is transcribed or recorded, say so now and say why: the hiring team reads the record, not just your memory. A candidate who knows the shape of the hour spends it answering rather than guessing.
  2. The candidate's story — 7 minutes. One question: "Walk me through the last two roles — what you were hired to do and what you actually ended up doing." You are listening for the gap between the two, and for where the candidate puts the credit. Stop at seven minutes; the rest of the hour is yours, not theirs.
  3. Competency questions — 25 minutes. Four or five questions from the scorecard, the same ones for every candidate, each followed by the follow-ups in the next section. Five minutes a question is enough when you cut off the preamble and ask for the specific case.
  4. The candidate's questions — 7 minutes. Not a courtesy: what a person asks after 35 minutes tells you what they will care about in month three. Write the questions down; they go into the write-up.
  5. Close — 3 minutes. Say what happens next and when: who decides, by what date, and who will tell them. A candidate who has not heard by that date has learned something about the company; make sure it is true.

Three rules hold the structure together. Two interviewers, two jobs: one leads and keeps the time, the other takes notes and scores; one person doing both produces a thin record and a thinner conversation. No decision in the room: the score is written after the call, alone, before anyone has said "so, what did you think?" Every candidate gets the same questions: the follow-ups can differ because the answers differ, but the core five are the role's, not the interviewer's.

What stays out: anything the scorecard does not ask for. A question that has no competency next to it is a question whose answer you cannot score — and a question about age, family plans, religion, origin or health is off the card in most jurisdictions and off the point everywhere; your HR partner or counsel has the exact list for yours.

Questions to ask — and the follow-ups that get the real answer

Good interview questions ask for a specific past case, not an opinion about how things should be done. "How do you handle conflict?" produces a philosophy; "Tell me about the last time you disagreed with your manager about a priority — what you said, what they said, how it ended" produces evidence. The STAR shape — situation, task, action, result — is a listening frame for the interviewer as much as an answer frame for the candidate: if you have heard the situation and the task but not what they did, you have not heard an answer yet.

Eight questions that cover the competencies most roles score, each with the follow-up that gets past the rehearsed version:

CompetencyThe questionThe follow-up
OwnershipTell me about a project that was yours to finish and was behind. What did you do in the first week?What would have happened if you had done nothing?
JudgmentDescribe a decision you made with incomplete information. What did you know, what did you not, and what did you choose?What did you learn afterwards that you wished you had known?
Working with othersTell me about the last time you disagreed with a colleague about how to do something. How did it end?What does that person think happened?
CommunicationWalk me through a time you had to explain something technical, or something bad, to people who did not want to hear it.What did you change in how you said it the second time?
LearningWhat is something you were plainly bad at two years ago and are competent at now? How did that happen?Who would confirm that?
DeliveryWhat is the piece of work you are proudest of in the last year? What was your part in it, specifically?Who else was on it, and what was theirs?
Handling failureTell me about something that went wrong on your watch. What was your part in it?What did you do differently the next time?
The role itselfHere is a problem we have right now — [one real, current problem, the same for every candidate]. How would you start on it?What would you need to know before you did anything?

Two habits of good interviewers. Ask for the case, then ask for the number: "How much was it behind?", "How many people?", "What was the result, in a figure?" — a story that has no numbers in it has usually been edited. Let the silence work: after the answer, wait three seconds; the sentence that follows the pause is often the one the candidate left out.

What to avoid: brain-teasers and puzzles, which measure comfort with puzzles; hypothetical questions where a real case exists; "what is your greatest weakness", which measures preparation for that question; and any question the candidate answered in the application, unless you want to hear it told differently.

Scoring: the rubric, filled in within the hour

Score each competency separately, on a scale with words attached to the numbers, and do it alone within the hour after the call. Each part of that sentence closes a known failure: a single overall score hides a weak competency behind a likeable candidate, a scale without anchors means "3" is a different grade from each interviewer, a score written the next day has been edited by memory, and a score written after talking to a colleague is partly their score.

A four-point scale with anchors — no middle point, so every score is a lean:

ScoreWhat it meansWhat you heard
4 — Clear evidenceClearly above the bar for this roleA specific case with the candidate's own actions, a result, and a lesson, without prompting
3 — EvidenceAt the barA specific case with their actions and a result, found with one follow-up
2 — Weak evidenceBelow the barGeneral statements, a shared or vague role, or a case with no result
1 — Contrary evidenceThe competency is a riskA case where the candidate's own account shows the opposite, or no case at all

The score is only as good as the sentence next to it. "3 — ownership" tells the hiring team nothing; "3 — took over the migration at week six, cut scope with the client the same week, shipped two weeks late instead of eight; result was his own account, not confirmed" is a score they can weigh against another interviewer's. Write the evidence first, then the number; the number written first tends to find its evidence.

One score per competency, one competency per question, and a recommendation at the end that is one of three words: hire, no, or "need another round on X" — with X named.

An interview scorecard template

[Role] — [Candidate] — [Round, e.g. 2 of 3]
Date: [date] · Interviewers: [lead], [notes and scoring] · Link: [meeting link or recap]
Core questions: the five below, the same for every candidate in this round.

1. Ownership — [the question] — score [1–4]
Evidence: [the case in one or two lines, the candidate's own actions, the result, in their words where possible]

2. Judgment — [the question] — score [1–4]
Evidence: […]

3. Working with others — [the question] — score [1–4]
Evidence: […]

4. Delivery — [the question] — score [1–4]
Evidence: […]

5. The role itself — [the problem, the same for every candidate] — score [1–4]
Evidence: […]

The candidate's questions: [what they asked, in order]
Risks: [what you would want checked by a reference or the next round]
Recommendation: hire / no / another round on [competency]
Written at: [time] — within the hour, before the debrief.

The template is plain text, so it pastes into a document, a tracker's candidate record, or a message to the hiring channel without reformatting.

The write-up for the hiring team

The write-up is the scorecard with the blanks filled, plus three things the card does not hold: the recommendation in one line at the top, so a reader in a hurry gets it first; the candidate's questions, which are evidence about fit that no question of yours produces; and the open items — what the next round should test, what a reference should confirm, what you were not able to find out in 45 minutes.

Two rules about when. It is written within the hour, while the answers are still the candidate's words and not your summary of them. And it is written before the debrief, by each interviewer separately, and submitted before anyone has read anyone else's: a panel that reads each other's scores first converges, and a panel that converges has run one interview with four people in it. The debrief then has one job — the disagreements. The 30-minute hiring debrief agenda in our agenda examples is built on exactly that: scores submitted in writing beforehand, two minutes of evidence per interviewer with no reacting, then discussion of the disagreements only, then a decision and a name against "who tells the candidate by when".

Where the write-up lives matters as much as what is in it. One that sits in an interviewer's notebook is a write-up the team does not have; one in the role's channel, next to the other interviewers' write-ups and the recording of the call, is a record the next round can read and a reference check can be written from.

Where the interview record lives in the tools you already use

Most panels interview over the meeting tool the company already runs on, and each keeps a different part of the record for a different set of people. Documented behaviour only, from the vendors' own pages, checked November 2026:

PlatformWhat the meeting keepsWho can read itThe language of the record
Microsoft TeamsA recap with the recording, transcript, shared files and notes for events that were recorded or transcribed; AI-generated notes and follow-up tasks with Teams Premium or a Microsoft 365 Copilot license; recap content is "not currently available for ad-hoc events"People in the organization who were invited to the event; "anyone who wasn't invited or isn't in your org won't have access, even if they have the recording link"; AI summaries can be shared by email with participants outside the organization; guests and external attendees can view the recording only if it is explicitly shared with themAI summaries are produced for events transcribed in a supported language, 25 listed; the page does not describe a per-reader language for the recap
Zoom WorkplaceA meeting summary with AI Companion on paid Zoom Workplace plans, sent by email, posted to the meeting's group chat in Team Chat and listed on the Summaries page of the web portalParticipants signed in to a Zoom account during the meeting, when the host shares it; "unauthenticated users will not receive an email summary"; only the host can edit it"The generated summary will be provided in the language used by the majority of participants", or the host's language when there is no clear majority; a default summary language can be set per account
Google WorkspaceA recording saved to the organizer's Drive, with the link emailed to the organizer and the person who started it and added to the Calendar event; Take notes for me on an eligible Workspace edition or Google AI plan writes a Google Doc with a summary and next stepsThe notes are shared per the host's setting — all invited guests including those outside the organization, guests in the organization only, or hosts and co-hosts only; a Workspace admin must allow recording first"This feature supports one language at a time. Multiple languages spoken in the same meeting aren't currently supported"

Two things are the same everywhere. The record is a file beside the meeting — a recording, a transcript, a Doc — in the language the room spoke, and the candidate, who joined as an outsider, is outside the record by default. A panel that bolts a notetaker onto one of these tools gets a second record with its own access rules; the Otter.ai alternatives comparison lists what each keeps and who can read it. And when the candidate's working language is not the panel's, what the tool itself translates in the call is documented in the guides to Microsoft Teams live translation and Google Meet live translation — a translated caption in the call, in each case, not a translated record afterwards.

How the interview comes back to the hiring team

InterMIND is a communication space where a meeting lives in a channel, so an open role has an obvious home: one channel per role, where the scorecard is posted before the round, the interview links are started from, and the write-ups are posted after. The channel's chat is translated per reader — each interviewer types in their own language and reads the others in theirs — so a panel in Milan and a panel in Lyon read the same write-ups without anyone translating them. A case brief or a job description shared in the channel as a file opens for each reader in their language, PDF, Word and PowerPoint included; a note shared during the call with Share to conference goes on everyone's screen next to the video tiles, read by each participant in their own language.

When the call ends, the recap is written from the meeting itself: a summary, the full transcript speaker by speaker with each line keeping its timestamp, and the recording when the session was recorded. It arrives by email, as a message in the channel and as a link in the calendar event, and every signed-in member reads it in their own language — the transcript is captured in each speaker's language and translated when a member opens it in theirs. Any member can edit the summary, every edit keeps its author and revision history, and the channel keeps the recap of every past session, so the second-round interviewer reads the first round before the call rather than asking the candidate to repeat it. The host can add a recap copy to any address that accepts email — a tracker's inbound mailbox, for instance — so the record reaches the system where the candidate's file lives without anyone forwarding anything. The recap is members-only: a link forwarded outside the team opens nothing.

The candidate joins by link, without an account and at no cost — from a phone, in the browser, nothing to install — gives a name, waits for the host to admit them, and hears the panel live in their own language while the panel hears them in theirs. There is no record for the candidate afterwards: the recap, transcript and recording stay with the host's account and its members, and the anonymous guest identity is deleted within 24 hours. The same channel is where the hire's first weeks happen: the host adds the new colleague by email onto a free account, and they read the role's history — the brief, the decisions, the recaps — in their own language. The host provides the space and, on a paid plan, pays only for the members active in the previous 28 days; members who go quiet keep their access at no cost.

Interviewing as a craft survives every tool: somebody still has to write the scorecard, ask the same five questions, and score alone within the hour. What does not have to survive is the second job — rewriting what was said, from memory, in one language, for a hiring team that reads in several.

FAQ

How do you conduct an interview step by step?

Write the scorecard first: the competencies the role needs, a question per competency, and a scale with words attached to the numbers. Then run the same five parts for every candidate — a three-minute opening that names the shape of the hour, seven minutes on the candidate's story, 25 minutes of competency questions with follow-ups, seven minutes for their questions, a close that says who decides and by when. Score alone within the hour, write the evidence before the number, and submit the write-up before the debrief.

How do you start an interview as the interviewer?

In three minutes: who you are, what the role is in one sentence, how the time will be used, and that you will take notes — and, if the call is transcribed or recorded, that it is and why. Then one question that gets the candidate talking about work rather than about themselves: "Walk me through the last two roles — what you were hired to do and what you actually ended up doing."

What questions should an interviewer ask?

Questions that ask for a specific past case tied to a competency on the scorecard — "tell me about the last time…", "describe a decision you made when…" — followed by a follow-up that tests the account: "what would have happened if you had done nothing?", "what does that person think happened?", "who would confirm that?" The same core questions go to every candidate; the follow-ups differ because the answers do. Puzzles, hypotheticals where a real case exists, and questions about protected characteristics stay out.

How long should an interview be?

Long enough for four or five scored questions with follow-ups and ten minutes that belong to the candidate: 45 minutes for a competency round, 30 for a screen with three questions, 60 for a technical round with a work sample. An interview that runs over is not extended; the question that did not fit goes to the next round or the reference check.

What is the STAR method in an interview?

A frame for a behavioural answer — situation, task, action, result — and, for the interviewer, a checklist for whether an answer has been given: if you have heard the situation and the task but not what the candidate personally did and what came of it, you have a story, not evidence. The follow-ups "what was your part, specifically?" and "what was the result, in a figure?" fill the missing letters.

How do you score an interview?

One score per competency on an anchored scale — this guide uses four points with no middle, from contrary evidence to clear evidence — written alone within the hour after the call, with the evidence next to the number: the case, the candidate's own actions, the result. Then one recommendation in one of three forms: hire, no, or another round on a named competency. Scores are submitted in writing before the debrief, which discusses only the disagreements.

Should you record a job interview?

The record is what the hiring team reads instead of each interviewer's memory, so a transcript is worth having; whether the candidate must consent, and how, depends on where you and they are — the meeting tools above notify participants when a recording starts, and your HR partner or counsel has the rule for your jurisdiction. Either way, say it in the opening and say why, so the candidate hears it from you rather than from a notification.


Sources: Microsoft — Recap in Microsoft Teams; Microsoft — Start, stop, and find meeting recordings in Microsoft Teams; Zoom — Using Meeting Summary with AI Companion; Google Meet — "Take notes for me" in Google Meet; Google — Calendar & Meet settings for "Take notes for me"; Google Meet — Record a video meeting; all pages checked November 2026. Vendors change plans, features and language lists over time; check their pages for the current state.

More in Meetings & recaps

All posts in Meetings & recaps
AI note taker for Google Meet: the ways to get notes from a Meet call, and what each leaves the team with (2026)
Meetings & recaps

AI note taker for Google Meet: the ways to get notes from a Meet call, and what each leaves the team with (2026)

An AI note taker for Google Meet is one of four things: Gemini's own Take notes for me, a notetaker bot that knocks and sits in the participant list, an app that records on your device with no bot, or a meeting room with the notetaker built in. Each one leaves something different behind the call — where the notes land, who receives them, which Workspace edition it needs, in which language they read, and what the guest who knocked walks away with. The four ways by the vendors' own documentation, a verdict table, and how to choose in 30 seconds.

The Mind.com Team

AI note taker for Microsoft Teams: the ways to get notes from a Teams meeting, and what each leaves the team with (2026)
Meetings & recaps

AI note taker for Microsoft Teams: the ways to get notes from a Teams meeting, and what each leaves the team with (2026)

An AI note taker for Microsoft Teams is one of four things: Teams' own Recap with Copilot or Facilitator, a notetaker bot that waits in the lobby and sits in the participant list, an app that records on your device with no bot, or a meeting room with the notetaker built in. Each one leaves something different behind the call — where the notes land, who receives them, in which language, and what the external attendee on the link gets. The four ways by the vendors' own documentation, a verdict table, and how to choose in 30 seconds.

The Mind.com Team

AI note taker for Zoom: the ways to get notes from a Zoom meeting, and what each leaves the team with (2026)
Meetings & recaps

AI note taker for Zoom: the ways to get notes from a Zoom meeting, and what each leaves the team with (2026)

An AI note taker for Zoom is one of four things: Zoom's own Meeting summary with AI Companion, a notetaker bot that sits in the participant list, an app that records on your device with no bot, or a meeting room with the notetaker built in. Each one leaves something different behind the call — where the notes land, who receives them, in which language, and what the guest on the link gets. The four ways by the vendors' own documentation, a verdict table, and how to choose in 30 seconds.

The Mind.com Team

Get new posts and product updates by email

One email a month with new posts and product updates. Unsubscribe anytime.