Facilitating two hundred people without losing the room

Somewhere between sixty and eighty participants, a facilitator stops being able to run a room and starts having to run a system. Almost everything that worked at twenty fails, and it fails in the same order every time.

The failures are structural, which is the good news: nearly all of the work happens before anyone arrives.

What actually breaks

You can no longer see the room. At twenty, you know which table is struggling because you can see it. At two hundred, a table can sit stuck for eleven minutes and you will not know unless the design tells you.

Instructions stop propagating. A briefing that lands cleanly with twenty people will be misheard by a predictable fraction of two hundred, and the misheard version spreads table to table faster than you can correct it.

Time stops being uniform. With ten tables, the spread between the fastest and slowest is a few minutes. With forty tables, it is the length of a round, and a round that ends while a third of the room is mid-decision produces resentment rather than insight.

The debrief has no natural form. You cannot take contributions from the floor. You cannot hear the back. The move that carries the learning at twenty simply does not exist at two hundred.

Table architecture is the main instrument

Decide the table size from what you want to be visible, not from the venue's furniture.

Five to six people per table is the working range for most designs. Below four, a single dominant participant controls the table. Above seven, two people will disengage completely and you will not see it happen.

Round tables beat long ones. On a long table the ends cannot hear the middle, and a group of eight at a rectangular table is functionally two groups of four having separate conversations about the same task.

Aisles matter more than they look. You need to be able to reach every table within about fifteen seconds without asking anyone to move a chair, because the whole recovery mechanism at scale is physical proximity.

The facilitator ratio

One facilitator can hold roughly six to eight tables — thirty-five to fifty participants — if the design is doing its share. Past that you need a second, and the second is not an assistant. They run their half of the room with the same authority, or the room learns very quickly which one to ask.

There is a cheaper intermediate step that works well: a table host per table, recruited from the client, briefed for thirty minutes the evening before. They are not facilitating. They have one job, which is to raise a hand when their table is stuck or has finished, and that single signal restores the visibility you lost at scale.

Instructions have to be given three ways

Say it, show it, and leave it on the table.

Said, briefly, at the front, in the same words every time. Shown, as one slide with the same words on it, left up for the duration of the round rather than advanced. And printed on a card at each table, so a participant who missed it can recover without asking a neighbour who also missed it.

Redundancy feels excessive when you write it. It is the difference between a round that starts cleanly and eleven minutes of correction.

Timing: announce the end twice, and be early

Give a warning at the point where a table that has not started deciding still can — usually around a quarter of the round remaining — and give it as a number, not as "nearly done".

Then end when you said you would. The first time you extend a round because a few tables are not ready, you have taught the room that your times are negotiable, and every subsequent round runs long.

The design should tolerate some tables not finishing. That is a feature. A table that ran out of time has produced exactly the material the debrief needs.

Debriefing a room you cannot hear

Three moves, in order.

Table first. Ask the question and give it to the tables for three minutes. Everyone speaks, nobody performs, and it costs you three minutes.

Then sample, deliberately. Take four tables, not twelve. Choose them — the one that finished first, the one that did not finish, the one you watched abandon its plan, and one at random. Say why you are choosing them. This is not a poll; it is a selection, and the room can see the reasoning.

Then land it from the front. At scale the facilitator has to close. You cannot build a conclusion out of contributions from a room of two hundred; you can build it in advance from what you watched and let the sampled tables supply the evidence.

The one thing to protect

Whatever is cut for time, do not cut the table-level conversation at the start of the debrief. It is the only point in a large session where every individual speaks. Lose it and two hundred people spend the day watching, which is a conference, not a simulation.