How to Keep a Conference or Symposium on Schedule

Conferences rarely run late because one talk is long. They run late because nobody owns the minutes between talks: the laptop swap, the extra question, the coffee break that ends five minutes after it was meant to. Keep a conference on schedule by designing slack into every session, giving each room a timekeeper with the authority to use it, and protecting published start times rather than published talk lengths.

The mechanics of a running order (columns, cue sheets, rehearsal) are in the run of show guide.

Where conference time actually goes

Look at any session that finished 15 minutes late and the overrun is usually spread across small leaks:

• Changeovers. Unplugging one laptop, finding the next deck, testing the clicker, clipping on a microphone.
• Q&A drift. "One last question" becomes three, and the chair doesn't want to be rude.
• Late starts after breaks. People are still in the coffee queue at the scheduled restart, so the chair waits.
• The keynote. Invited speakers are the least briefed and the hardest to stop.
• Housekeeping. Announcements, sponsor thank-yous and a chair introduction that turns into a biography.

How long should conference sessions and slots be?

There is no universal standard, so copy what well-run meetings in your field do and write it down for every speaker. As one example, the 2024 International Meeting of the Psychometric Society (IMPS) told chairs that in a typical 75-minute concurrent session "each speaker has 15 minutes total, of which 12 minutes are intended for the talk", with the other 3 for questions and moving to the next speaker (IMPS 2024 session chair information).

These are the lengths and buffers I recommend as a starting point. Treat them as defaults to adjust, not rules:

Slot type	Typical slot	What to build in
Contributed paper	12 to 15 min	2 to 3 min Q&A inside the slot; changeover time on top unless slides are preloaded
Lightning talk	5 min	No Q&A; all slides in one deck so there is no changeover
Invited or plenary talk	30 to 45 min	Q&A as a separate, visible block of 10 to 15 min
Panel	45 to 60 min	Moderator owns the clock; see moderating a panel to time
Coffee break	20 to 30 min	The restart time is announced, not the break length
Room change between sessions	5 to 10 min	More in large venues or across buildings

The single biggest saving is getting every deck onto the room computer before the session starts. The DRUID conference's guidance (an undated page) asks chairs to make sure each presentation has been uploaded before the session (DRUID, hints for the session chair). A preloaded deck removes most of the changeover.

Worked example: fitting papers into a 90-minute session

Inputs:

• Session length: 90 minutes
• Chair's opening and closing remarks: 3 minutes in total
• Paper slot: 15 minutes (12 talk, 3 questions)
• Changeover between papers: 1 minute, because slides are preloaded
• Reserve kept unallocated: at least 5 minutes

Working: each paper costs 15 + 1 = 16 minutes. Five papers cost 80 minutes, plus 3 minutes of chair time, which leaves 7 minutes of reserve. Six papers would need 96 + 3 = 99 minutes, which doesn't fit.

So the session holds five papers with 7 minutes spare. If someone asks for six, the honest answer is a 12-minute slot (10 talk, 2 questions): 6 × 13 = 78, plus 3 = 81, leaving 9 minutes. Run the same calculation before you publish any session, because once speakers have their slot in writing it's very hard to shorten.

The session chair is your timekeeper

In a multi-room conference you can't be in every room, so the chair (or a dedicated timekeeper sitting beside them) runs the clock. Give them three things: the published times, a way to signal, and explicit permission to stop people.

Time cards are still common. IMPS 2024 put cards for "5 minutes" and "1 minute remaining" in every room and told chairs to hold each card up until the speaker acknowledges it. DRUID uses a three-card sequence, five minutes, two minutes, then a red card meaning stop, and tells chairs to be "polite but firm". A screen-based countdown does the same job without the speaker needing to spot a card; the trade-offs are covered in what to show on a speaker timer.

A script for session chairs

Opening the session (about 30 seconds):

"Good morning, and welcome to the session on [topic]. We have five papers, each with twelve minutes to present and three for questions. I'll show a card at five minutes and one minute, and I'll stand up when time is up. Please hold questions until the end of each talk."

When a speaker passes their time:

Stand up, move toward the lectern and wait for a natural pause. "Thank you, that's our time. We have room for one quick question."

When Q&A runs on:

"Let's take that conversation to the break. Please join me in thanking [name]. Our next speaker is [name]."

Saying the rules out loud at the start matters more than the wording. Speakers accept being stopped when everyone heard the limit announced.

Strict paper slots and parallel tracks

When several sessions run at once, attendees plan their day around specific talks and walk between rooms to catch them. That changes the rule for no-shows. The ICDAR 2026 instructions for chairs say: "If a presenter does not show up, do not move the following presentation forward", and to start the next talk at its scheduled time, because "participants may move between parallel sessions to attend specific talks" (ICDAR 2026, instructions for session chairs).

Not every meeting agrees. IMPS 2024's policy told chairs not to leave a break for an absent speaker and to start the remaining talks immediately after the preceding one (a sample announcement elsewhere on the same page describes leaving a break, so its guidance is not fully consistent). Starting early works when a session is self-contained, but breaks the published times room-hoppers rely on.

Pick one policy, put it in the chair briefing, and use the gap well: extend questions for the previous speaker, open a general discussion, or give the room a short break. For the logistics of staffing several rooms at once, see running multiple rooms and tracks.

Keynotes: agree the end time in writing

Keynote speakers overrun more than anyone, mainly because nobody feels able to stop them. A few habits help:

• Confirm the end clock time, not just the length. "You're on from 9:15 to 9:55, with questions until 10:05" is harder to stretch than "40 minutes".
• Separate the talk from the Q&A. If the talk overruns, the Q&A shrinks, not the next session.
• Decide in advance who stops them. Usually the host or chair, with a specific cue the speaker has agreed to.
• Place keynotes before a break where you can. An overrun then eats into coffee rather than into the next speaker's slot. The cost is a shorter break, so announce the restart time clearly.

The broader toolkit for speakers who drift is in how to stop speakers running over time.

Breaks: announce the restart, then keep it

A break only works if the next session starts on time. Announce the restart as a clock time ("we start again at 11:00"), give a call from the doors a few minutes before, and start at the time you said even if the room is half full. If chairs wait for stragglers, everyone learns that restart times are soft, and every break after that gets longer.

When the day is already running late

Decide early. A ten-minute slip at 10:00 is easy to absorb in a break; the same slip discovered at 16:00 usually means cutting someone's talk. Have the chair or producer check the clock at every break and make the call there. The full playbook, including where to claw back time and how to tell speakers, is in what to do when an event runs late.

Timing guides for other kinds of events

The same principles apply elsewhere, with different pressure points:

• Church services and their livestreams, where the preacher sets the pace and online viewers need a clear start.
• Town halls, all-hands and product launches, where the most senior speaker overruns and the Q&A gets squeezed.
• Award ceremonies and galas, where the walk to the stage costs as much as the speech.
• Hackathons, where judging and demos decide whether the closing starts on time.

Frequently asked questions

How early should session chairs and speakers be in the room?

IMPS 2024 asked chairs to be in the room at least 15 minutes before the session starts. I suggest asking speakers to arrive by the break before their session, so slides can be loaded and the clicker tested while the previous session is still running.

Should the timekeeper be the chair or a separate person?

For sessions of short papers, a separate timekeeper beside the chair works better: the chair can listen, think of questions and introduce speakers while someone else watches the clock. For a single keynote, the chair can do both.

Do hybrid and online attendees change the timing?

They make published start times even more important, because remote viewers join at the time on the programme and leave if nothing happens. See timing webinars and hybrid meetings for the online side.

Key takeaways

• Most overruns come from changeovers, Q&A and late restarts, not from the talks themselves. Preload slides and announce restart times.
• Do the slot maths before you publish: talk plus questions plus changeover, with a reserve of at least five minutes per session.
• Give every room a chair or timekeeper with cards or a visible countdown, a script, and permission to stop people politely.
• In parallel tracks, protect published start times. Decide your no-show policy in advance and brief chairs on it.
• Agree keynote end times as clock times, keep their Q&A separate, and decide who stops them before they walk on.

Sources

• IMPS 2024 session chair information, Psychometric Society
• Instructions for session chairs, ICDAR 2026
• Hints for the session chair, DRUID conference (undated page, accessed October 2026)