How to Moderate a Panel Discussion (Without Boring Everyone)

Most panels are bad. Not because the panelists are uninteresting — most aren't — but because the moderator runs the same playbook everyone runs: introduce the speakers, ask each one a soft opener, work through a list of prepared questions, take audience Q&A, thank the panel, exit. The audience claps. Nobody learns anything. The conference organiser is happy. The panelists go home. Everyone has politely wasted forty-five minutes.

This piece is the alternative. A specific method for moderating a panel discussion that the audience remembers — built around the work moderators are usually skipping. Read it if you have a panel coming up. Read it more carefully if you've moderated panels before and you want to stop doing it badly.

TL;DR

• The work happens before the panel, not on stage. Most moderators do the prep wrong.
• Design the panel as a conversation with an arc — three to five themes that build to a tension.
• The first 90 seconds of the panel decide everything. Set the room.
• The follow-up is the interview. Plan it before the panel, not in the moment.
• Close hard. Don't let the panel die in audience Q&A.

Phase 1: Research the speakers, not the topic

The first thing most moderators do is research the topic. This is the wrong starting point. The topic is given. The speakers are the variable.

For each panelist, you need to know:

• What they've publicly argued in the last 12 months. Their LinkedIn posts, their interviews, their podcasts, their conference keynotes. If they've written a book or a Substack, scan it.
• Where their public position has shifted. Look for tension between what they said two years ago and what they're saying now. Those gaps are gold.
• Who their critics are and what those critics are saying. Every public figure has critics. The criticism tells you what the figure is defensive about — and what the audience already knows about them.
• What they have not said. This is harder. It requires reading the topic landscape and noticing what each panelist has avoided. The unsaid is often more interesting than the said.

This research takes time — typically two to four hours per panelist for a serious panel. Most moderators skip it because it isn't visible work. The audience doesn't see it. But the audience absolutely sees the result: questions that feel personal and pointed instead of generic and recyclable.

If you only have one hour to prep, spend the entire hour on research. Don't write a single question yet.

Phase 2: Find the tension

Once you understand each speaker, the next question is structural: what is the actual disagreement on this panel?

If you can't name the tension, you don't have a panel. You have four people taking turns saying things.

The tension might be:

• Speed vs. caution (one speaker has shipped fast and broken things; another has shipped slowly and shipped safely)
• Inside vs. outside (one speaker has run the system; another has critiqued it from outside)
• Old guard vs. new (one speaker built the playbook; another is rewriting it)
• Theory vs. practice (one speaker has the framework; another has the data on whether the framework actually works)

Whatever it is, name it before you write a question. The tension is the story arc of the panel. Every question you ask should put pressure on it.

If your panelists agree on everything that matters, the panel is broken. Either change the panel or change the topic.

Phase 3: Design the arc

A panel is a 45-minute conversation. Most moderators design it as 12 separate questions. That's why most panels feel like a list. They are a list.

Instead, design the panel as a conversation with three to five themes, each with two to four questions, building toward the sharpest tension at the end.

Example structure for a 45-minute panel:

• Theme 1 (10 min): The setup. Establish where each panelist sits on the topic. Two questions.
• Theme 2 (12 min): The pressure. Where their positions actively conflict. Three questions, including a routed exchange ("Sarah, you just heard Marcus's position — respond to that.")
• Theme 3 (10 min): The data. Bring in evidence — case studies, numbers, public examples. Two to three questions.
• Theme 4 (8 min): The application. What should the audience actually do with this. Two questions.
• Close (5 min): One sharp closing question to each panelist. Skip audience Q&A or limit it to two questions.

Note what's NOT in this structure: opening softballs, "tell us about your journey", or any "lightning round" filler. Cut all of it. Forty-five minutes is short. Use it.

Phase 4: Write the questions, with the follow-up planned

Once you have the arc, write the questions. Use sharp architectures — tradeoff, contradiction, boundary, scenario, accountability. We covered all five and gave 25 examples in our piece on fireside chat questions. The same architectures work for panels.

The critical difference: every question must come with a planned follow-up.

Most moderators write a list of opening questions and assume the rest will be improvisation. This is why panels die. The opening question gets a rehearsed answer. The moderator, with no follow-up planned, moves on. The rehearsed answer becomes the answer.

For every opening question, write down what the dodge will be. ("They will retreat into the framing.") Then write the follow-up that closes the dodge. ("That's the framing — what's the actual decision?") If you can't write a follow-up that gets past the rehearsed answer, the question isn't sharp enough. Rewrite it.

Phase 5: Open the panel correctly

The first 90 seconds of a panel decide what kind of panel it will be. Most moderators waste them on housekeeping and introductions.

Here's a stronger opening:

1. One sentence introducing the panel — not the panelists, the question. "Today we're trying to answer one specific question: [the actual question the panel is about]."
2. 30 seconds introducing each panelist — by what they've actually argued, not their CV. "Sarah Chen has spent the last decade arguing that enterprises should slow down on AI. Marcus Webb has spent the same decade arguing the opposite."
3. One line setting the tone — "I'm going to push hard. They've all agreed to that. So let's not waste the first ten minutes warming up."
4. Straight into the first question. No softball. The first question should be the sharpest question of the first theme.

The audience knows in 90 seconds what kind of panel this will be. If you spend the first ten minutes on warm-up, you've signalled that this is going to be a polite, forgettable panel. They will treat it accordingly.

Phase 6: Run the body well

The middle of the panel is where most moderators get tested. Three things matter:

Listen aggressively. Most moderators think their job is to ask the next question. It isn't. Their job is to listen to the answer and decide what should happen next. If the answer was sharp, follow up. If the answer was a dodge, push. If the answer surfaced something none of the other panelists has addressed, route the next question there. The list of prepared questions is a reference — not a script.

Interrupt when necessary. If a panelist runs long, you must interrupt. Politely, at a sentence break: "Let me pause you there — I want to bring Marcus in on this." The audience will thank you. The panelist won't be offended; they were warned during the prep call. (You did do a prep call, right? If not, see Phase 1.)

Refuse the rehearsed answer. This is mistake 4 from our piece on what most moderators get wrong. When the dodge happens, name it: "That's the version you've given before. What's the part you usually leave out?" This is the single most powerful tool a moderator has. Most never use it.

Phase 7: Close hard

Most panels die in audience Q&A. The audience asks worse questions than the moderator, the panelists give vaguer answers, the energy drains, and the panel fizzles out.

Don't let it.

If you take audience Q&A, take two questions maximum, and you choose them. "We have a lot of hands. Let me pick two." You're not being undemocratic. You're protecting the panel.

Then close with a sharp one-question round. Each panelist gets the same closing question — phrased to force a real answer. Examples:

• "In one sentence — what's the single thing you most want the audience to disagree with?"
• "Twelve months from now, what's the one number we should hold each of you accountable to?"
• "What did you almost say tonight that you held back?"

The closing question is the moderator's last chance to make the panel memorable. Don't waste it on "any final thoughts?" Final thoughts are forgettable. Specific provocations are not.

A 60-minute prep checklist (when you're short on time)

If you have a panel tomorrow and one hour to prep:

• Minutes 0–15: Read the most recent public position from each panelist. One thing each.
• Minutes 15–30: Identify the tension. Write it as a single sentence.
• Minutes 30–50: Write 6 questions that put pressure on the tension. Two openers, two pressure questions, one routed exchange, one closer.
• Minutes 50–60: For each question, write the dodge you expect and the follow-up that closes it.

That's it. You will run a more useful panel than 95% of moderators with that hour alone.

The takeaway

Most moderators run panels the way most people give wedding speeches — they prepare the script, deliver the script, and call it a panel. The audience claps because that's the social contract. But nobody actually leaves with anything.

The alternative is also a script — but a different kind. One built around the speakers, the tension, and the moments where the dodge will happen. One that treats the moderator as an editor, not a host. One designed for memorability, not pleasantness.

It takes more work. The panel has more edge. The audience remembers. The conference invites you back.

That's the trade.

Walk in prepared.