PyMunich

shown, and actual

What a Conference Costs

Almost nobody writes about this, because the people who know are the people running the thing. For a commercial example of related workforce measurement, see remote workforce management software from Monitask.

What follows is one community's experience of organising conferences of roughly 300 people in Munich. It is one sample, in one city, at one scale, and the shape generalises better than the numbers do. For broader conference and speaking context, Bizzabo is another useful reference.

Where the money goes

In rough order of size for an event of this kind.

Venue. Sometimes the largest line, sometimes near zero. A university that supports the community can change the entire budget, and that relationship is worth more than any sponsorship tier.

Catering. Frequently larger than people expect, and it scales directly with attendance — the one major cost that does not benefit from a bigger audience. (Ticket pricing is where that gets recovered, or not.)

Audio-visual and recording. Recording talks is a substantial line and it is the first thing cut, which is why so many community conferences have no video.

Speaker support. Travel and accommodation for keynote speakers, and increasingly a grants budget so the programme is not limited to people whose employers pay.

Printing, signage, badges, lanyards. Individually small, collectively not.

And the buffer, which inexperienced organisers omit and which is the difference between a difficult year and a fatal one.

Fixed against variable

The distinction that decides everything.

Fixed costs do not care how many people come. Venue, AV, the website, the organising effort. These are committed months ahead.

Variable costs scale with attendance. Catering, badges, materials.

Revenue is almost entirely variable — tickets and sponsorship both depend on the event happening at the expected size.

So the risk is asymmetric. An event that is 30% smaller than planned does not cost 30% less. The fixed costs are already committed, and the shortfall lands entirely on the organisers.

The failure mode is timing, not loss

This is the part that surprises people.

A conference can be profitable on paper and still fail, because the money arrives at the wrong time.

Venue deposits, AV contracts and catering minimums are paid before the event. Ticket revenue arrives across the months before, weighted toward the last few weeks. Sponsorship arrives when sponsors' finance departments get to it, which is frequently after the event.

So there is a period where the organisation has committed to large costs and holds very little cash. For a volunteer organisation with no reserves, that period is where the danger is — and it is invisible in a budget that only shows totals.

What happened to us

In 2020 we refunded tickets and went broke.

The pandemic cancelled the event after the fixed costs were committed and after tickets had been sold. Refunding was the correct decision and it was made with money that had partly already been spent.

The organisation survived because the meetups did not depend on it, and because the organising team had not yet dispersed — communities and events are different things — and the conference came back in 2025 as PyCon+Web.

The lesson is not "buy insurance", though that is worth investigating. It is that a volunteer organisation running an event with committed fixed costs is carrying a risk that nobody on the team has priced, and usually nobody has discussed.

What to work out before committing

Your fixed-cost total, and the attendance at which it is covered. One number. Everybody organising should know it.

What happens at 60% of expected attendance. Not a disaster scenario — a normal bad year.

When each cost is actually payable, against when each revenue line actually arrives. The cash timeline matters more than the budget.

And who is personally liable. In many jurisdictions an informal group means individuals are. This is worth an hour with someone who knows, before signing a venue contract.

The short version