How to run a group interview: agenda, exercises and an observer scoring grid

To run a group interview, bring several candidates together for a set agenda of short, job-related exercises, have trained observers each watch a fixed set of candidates against the same scoring grid, add a brief one-to-one conversation for each person, and score independently before observers compare notes. The format works when you are hiring several people into the same role and the job depends on working with others: retail, hospitality, customer service, seasonal and graduate programs.

A group interview is one where several candidates are assessed at once. It is different from a panel interview, where several interviewers meet one candidate. The planning problems overlap, but a group interview adds one that panels do not have: every candidate is also watching, and reacting to, the others.

When a group interview fits, and when it does not

Before booking, write down in one sentence what the group format will show that a one-to-one screen would not. If the answer is only "it saves time", run faster individual screens instead; the screening more candidates per day page has the method.

Decide what you are scoring before you design anything

Pick three or four behaviors the job genuinely needs and that people can show in 90 minutes. Define each in terms an observer could see or hear. "Teamwork" is not scoreable; "builds on another person's idea by name" is.

Leave out anything a group setting measures badly, such as confidence with strangers, speed of speech or how well someone performs being watched. These rarely predict how the person will do the job, and they penalize quieter candidates who may be excellent. The wider method for behaviors like these is in how to assess soft skills in interviews.

What to tell candidates before they arrive

Group interviews go badly when candidates do not know what they are walking into. Send this with the booking confirmation.

Subject: Your group interview for [Role], [Day] [Date] at [Time]

Hi [First name],

You're booked for a group session for [Role] on [Day, Date],
[start time] to [end time], at [address / link].

What to expect: you'll be with [number] other candidates. After a
short introduction to the job, there are two short team activities,
a written scenario question, and a five-minute one-to-one chat.
There is nothing to prepare and no trick questions.

If you need any adjustment, such as extra time for reading or a
particular seat, reply to this email or call [name] on [number].
It stays between us.

We'll let everyone know the outcome by [date].

[Your name]

A 90-minute agenda

0:00  Welcome (10 min)
      Who the observers are, the agenda, breaks, that notes are taken,
      that there are no trick questions. Check adjustments quietly.

0:10  Role preview (10 min)
      The real job: shifts, pay range if shared, what a hard day looks
      like. Invite anyone who realizes it is not for them to leave at
      the break with no awkwardness.

0:20  Exercise 1: paired task (15 min)
0:35  Exercise 2: group discussion (20 min)
0:55  Break (5 min)
1:00  Exercise 3: short individual scenario (10 min)

1:10  One-to-one conversations (15-20 min, in parallel)
      Five minutes each, same three questions for everyone.

1:30  Close (5 min)
      What happens next and by when. Thank everyone.

The role preview is worth its time. Candidates who hear the realities of the shift pattern and choose to leave save everyone a later no-show or early resignation.

Three exercises that produce evidence

1. Paired task

Two candidates solve a small, job-like problem together, such as planning a shift rota from a list of constraints or ordering a set of customer requests by urgency. Pairs mean nobody can hide and nobody can dominate a crowd. Observers watch two pairs each.

2. Group discussion with a decision

Give the group a short written brief and a decision to make together in 15 minutes. It needs a real trade-off, so there is something to discuss, and no hidden right answer.

Example brief (invented):
The store has budget for one change before the holiday season:
a second till at peak hours, a staffed returns desk, or longer
opening hours on Saturdays. Customer comments and last year's
queue times are below. Agree one recommendation and two reasons,
and choose who will present it in one minute.

3. Individual scenario

Each candidate writes or says, in private, how they would handle one situation from the job: a customer complaint, a co-worker not pulling their weight, a safety issue. This gives every candidate, including the quiet ones, a chance to show judgment without an audience.

Observers: numbers, roles and briefing

• Ratio. One observer per three or four candidates at most. Assign candidates to observers by name before the session, and rotate them between exercises if you have enough people.
• Roles. One facilitator who runs the clock and does not score. Observers who do not join the discussion or answer questions during an exercise.
• Briefing. Thirty minutes before candidates arrive: the behaviors, the grid, what each score looks like, and the rule that notes record what was said and done, not impressions.
• Mix. Include at least one person who does or supervises the job. They spot the behaviors that matter on a real shift.

Observers should write short, factual notes as they go: "Suggested splitting the brief into three parts; asked Sam for her view when Sam had not spoken." Those notes are what the score is built from, and what you rely on if anyone later asks why a candidate was not selected.

The observer scoring grid

Use one grid per observer. Score each candidate alone after each exercise, before talking to anyone.

Add a notes column with one piece of evidence for each score. A score without evidence should not count in the wash-up. The interview rating scale page explains why anchored 1 to 4 scales are easier to use consistently than 1 to 10.

Fairness in a room full of candidates

• Same brief, same timings. Every group gets identical materials and the same time limits, even if one group finishes early.
• Adjustments. Ask in the invitation and again privately on arrival. A candidate who is hard of hearing may need a seat facing the group, or a written copy of spoken instructions. See ADA accommodations in interviews.
• Language load. Keep written briefs short and plain, unless reading complex text is part of the job.
• Group composition. Avoid a single candidate who is visibly different from everyone else in their group where you can, and never let group dynamics excuse a low score. Score what each person did.
• Privacy. Do not discuss any candidate's background in front of others, and handle questions about pay or availability in the one-to-one part.
• Questions you would not ask one-to-one are still off limits in a group. The list is in illegal interview questions.

The wash-up and what happens next

Hold the decision meeting immediately after the session, while the notes are fresh.

1. Each observer reads out their scores for their assigned candidates, with evidence, before any discussion.
2. Discuss only where scores disagree by two points or more, or where the evidence is thin.
3. Combine the grid with the one-to-one answers. A strong group performer who cannot work the required shifts is still a no.
4. Decide yes, no or second conversation for each person, and record the reason in a sentence.

Tell candidates the outcome by the date you promised at the close, and offer brief, factual feedback where you can. Candidates who met a room of other applicants talk about the experience to each other and to friends, and a slow or silent follow-up damages the next event. If you run group sessions regularly, the interview calibration session is a useful way to keep observers scoring the same way over months.