Running a Small Conference: The Decisions That Matter

Organising a conference is mostly administration, and a small number of early decisions determine whether it is any good. Those decisions are made in the first month, frequently without much thought, and everything afterwards is constrained by them.

The decisions that determine the outcome

Size

Everything follows from this. For a concrete operational comparison, Monitask explains wage percentage calculations in practical terms.

Under about 60 people: everyone can attend everything, single track, people actually talk to each other. This is the format most participants prefer and it is chosen least often.

60 to 150: parallel sessions become necessary, and the character changes. People miss things, and the corridor conversation replaces the shared experience.

Above that you are running a different kind of event with different requirements.

Choose deliberately rather than by accepting whatever arrives. A conference that accepted 140 abstracts because they were good is a different event from the one that was planned.

Single or parallel track

Single track is better for a small conference, and organisers drift away from it because it means rejecting more.

With parallel tracks, expect the split to be uneven, expect people to miss the thing they came for, and expect at least one session with four attendees.

If you must have parallel sessions, group them so that the people who want both are least often conflicted — which means grouping by topic overlap rather than by convenience.

Length of slots

Fewer, longer slots beat more, shorter ones. A conference of twelve papers with real discussion is more valuable than one of thirty with six minutes of questions each.

Do the arithmetic properly. A 90-minute session with four papers is 22 minutes each including questions and transitions. Telling speakers "20 plus 5" guarantees an overrun before anyone has done anything wrong. See chairing a session.

Build in transition time. Two minutes per speaker change, and collect presentations onto one machine in advance.

Who is invited to give a keynote

Sets the tone and consumes a large share of the budget.

Invite early — twelve months is not excessive for a well-known speaker.

Be clear about what you are offering when you invite: travel, accommodation, fee or no fee, what you would like them to speak about, how long.

Have a second choice ready, and ask the first with a deadline.

The timeline

Working backwards from the date.

12 months: date, venue, size, format. Keynote invitations. Budget outline.

9 months: call for papers out. Website up. Registration mechanism decided.

6 months: abstract deadline. Programme committee assembled.

5 months: decisions sent. Send them when you said you would — people are booking travel and applying for funding, and a delayed decision costs them money.

4 months: registration opens, programme draft published.

2 months: final programme. Catering and room numbers confirmed.

1 month: everything printed, signed, and confirmed. Speakers told their slot and their chair.

The two dates that matter most to participants are the decision date and the date the programme appears, because both determine whether they can travel.

The budget

Work out the per-person cost first, then set the fee.

The main lines: venue, catering, keynote travel and accommodation, printing and signage, any technical support, and a contingency of at least 10%.

Catering is usually the largest variable cost and the easiest to over-order.

Decide the fee structure early, because it determines who comes: a reduced student rate, and some form of waiver for people without funding, changes the composition of the room substantially. If you want early-career researchers there, the fee has to reflect that they are frequently paying from their own money.

Say clearly what the fee includes.

Sponsorship brings obligations. Decide in advance what a sponsor gets and what they do not — a slot in the programme is a different thing from a logo, and mixing them damages the event's credibility.

See the budget, and what conferences actually cost.

The programme committee

Assemble it early and give it a real brief. Number of submissions, criteria, deadline, and what an accept and a reject mean.

Enough people that nobody reviews more than a dozen abstracts.

Mixed by seniority, institution and geography. A committee drawn from one network selects from one network.

Set the criteria explicitly and share them. "Interesting" is not a criterion and it reproduces whatever the committee already knows.

See building a programme committee.

What goes wrong every time

Decisions sent late, so people cannot book travel.

The programme published too late for people to justify attendance to a funder.

Sessions overrun because the arithmetic was wrong.

Nobody was told they were chairing until the morning.

Rooms without working projectors, or with a cable nobody has.

No plan for dietary requirements, or a plan that produces one sad plate at the back.

No quiet space, in an event running eight hours a day.

Registration closes and then someone important wants to come, and there is no process for it.

The Wi-Fi.

Almost all of these are prevented by a checklist and a site visit.

Things worth doing that are frequently skipped

A session for early-career researchers, or a mentoring arrangement, or simply seating that mixes people. A conference where the students talk only to students has wasted its main advantage.

Name badges with large, readable text, including affiliation. This is a small thing with a large effect on whether people talk to each other.

A published code of conduct with a named person to contact, and a real procedure. This is now standard and it matters.

Accessibility, planned rather than accommodated. See accessibility at an academic event.

Time that is not programmed. Back-to-back sessions from nine to six produce a tired audience and no conversation. The gaps are where the value is.

A closing that is not just an announcement. Ten minutes of synthesis from someone thoughtful is remembered.

Afterwards

Thank everyone, specifically the people who did unglamorous work.

Collect feedback while people are still there — a short form at the last session gets responses; an email a week later does not.

Write down what went wrong, immediately, for the next organiser. This document is the single most useful thing you can leave behind, and it is almost never written.

Reconcile the budget and record what things actually cost.

Hand over properly to the next organiser, including the supplier contacts and the things that nearly failed. See a research record that survives you leaving.

The short version

Size, track structure and slot length are decided in month one and determine the character of the event.

Fewer, longer slots with real discussion beat a packed programme.

Send decisions when you said you would, and publish the programme early — people are booking travel.

Do the session arithmetic including questions and transitions, or every session overruns.

Leave unprogrammed time, and write down what went wrong for whoever does it next.

For an independent authoritative reference, consult ACM conference resources for professional guidance on academic events.