How to Run a Panel Interview That Actually Works

Three of you sit down across from one candidate. The ops manager asks about a system they used two jobs ago. The finance lead asks a curveball question nobody agreed on beforehand. You ask something you already asked in the phone screen. Twenty minutes in, the candidate is answering to whoever spoke last, and afterward the three of you stand in the corridor going "so... what did we think?" and nobody has an actual answer.

That's not a panel interview. That's three interviews happening at once, badly.

A panel can be genuinely useful — it saves everyone's time, it gives you more than one read on the same person, and it looks less intimidating to a candidate than three separate rounds with three separate calendars to chase. But it only works if you treat it as one structured conversation with assigned parts, not three people improvising in the same room.

When a panel is the right call

Use a panel when:

• The role touches multiple teams and each panelist genuinely needs to judge a different thing — a hire for a small ops team might need a read from finance, from the person they'll manage, and from you.
• You're short on time and stringing together three solo interviews would take two more weeks the candidate doesn't have.
• The role is senior enough that one person's judgement alone feels risky.

Don't use a panel when:

• It's really just you wanting moral support for a decision you're not confident making. A second opinion is fine — get it from a colleague afterward, not by putting them in the room.
• Nobody has actually agreed what each person is there to assess. If you can't answer "why is this person on the panel" in one sentence, they shouldn't be.
• The role is junior and low-stakes. Three people interviewing someone for their first job out of school is intimidating for no return.

Most small companies over-panel. Two or three people is plenty. Once you're past four, candidates start performing for the room instead of answering the question, and you lose more signal than you gain.

Before the interview: assign the parts

The single biggest fix is deciding, in advance, who owns what. Send a short brief to your panel the day before — an email like this works fine:

> Subject: Panel for [Candidate name] tomorrow, 2pm
>
> We're 45 minutes. Split like this:
> - Amira (10 min): walk-through of their current role, what they actually did day to day
> - Jun (15 min): the [specific work scenario] — we want to hear how they'd handle it, not whether they've done it before
> - Me (10 min): questions about working with our team, availability, expectations
> - Last 10 min: their questions for us
>
> Please don't repeat each other's questions. Take your own notes — we'll each score independently before we compare notes.

That email takes two minutes to write and it's the difference between a structured conversation and a pile-on. It also stops the awkward moment where the candidate answers the same "tell me about yourself" question three times to three different faces.

If you can, agree beforehand on roughly what a strong answer looks like for the harder questions — not a rigid script, just enough that two panelists don't walk away with wildly different bars for the same response. This matters more than most people think: without it, the panelist who talks most in the debrief tends to set the standard, regardless of whether their read was actually the sharpest one.

During the interview

Pick one person to open and close — introductions, what the role is, how the time is split, and at the end, next steps and timeline. This is usually whoever owns the hire, not necessarily the most senior person in the room.

A few things that make the middle go better:

• Stick to the order you agreed. If Amira's section runs long, say so out loud ("let's move to Jun's questions, we can come back if there's time") rather than letting it drift.
• Don't interrupt each other's sections. If something interesting comes up outside your assigned area, write it down and raise it in the debrief, not mid-interview.
• Take your own notes, on paper or a shared doc, but don't compare in real time. A raised eyebrow or a quick "good point" from a colleague mid-interview changes what the next person asks, and now you've lost the independence that made a panel worth having in the first place.

After: score before you talk

This is the step almost everyone skips, and it's the one that actually matters. Each panelist writes down their own read — strong / borderline / no, plus one line of why — before the group discusses anything. Then compare.

If everyone lands in the same place, the debrief is five minutes: confirm, agree on the offer or the pass, move on. If people land in different places, that's useful information, not a problem to smooth over. Find out specifically what each person saw that the others didn't. Often it's not a real disagreement about the candidate — it's that one panelist was judging communication style and another was judging technical depth, and nobody had agreed those were both in scope.

Watch for the most senior person in the room speaking first in the debrief. It's natural, and it quietly ends the conversation — everyone else calibrates to that opinion instead of stating their own. Ask the newest or most junior panelist for their read first, every time.

When the panel genuinely can't agree

Sometimes two people have a real, substantive disagreement — not a communication mismatch, an actual different read on whether the person can do the job. That's a sign you need a third data point that isn't another opinion: a work sample, a short structured test, something that measures the specific thing you're actually unsure about rather than adding a fourth person's gut feel to the pile. AssessFit exists for exactly that moment — a skills-based tiebreaker when the room is split and everyone's already given their honest read.

A panel interview isn't a way to spread the decision across more people so nobody's individually responsible for it. It's a way to get three specific, independent reads on three specific things. Plan the parts, score before you talk, and let the newest voice in the room speak first. That's most of the job done.