What group size actually changes

Group size is usually treated as a logistics detail: however many people need to attend, attend. It is the single biggest determinant of what a session can produce, and it is settled before anyone discusses objectives.

The same simulation is a different intervention at twelve, forty and two hundred.

Under twelve

Everyone is visible. There is nowhere to sit out, every contribution is attributable, and the facilitator can name what each individual did.

This is the only size at which a session can do individual work — where a named person can be shown their own pattern and the room is small enough that it is a conversation rather than an exposure. For a leadership team, it is the right size.

What you lose is the dynamic between groups. With two teams there is no ecosystem, no competition worth the name, no chance to observe how information does or does not move between units. If the brief is about silos, ten people cannot show you silos.

The other risk is that one person sets the tone for the whole room. At twelve there is no second centre of gravity, and a dominant participant is not diluted.

Twelve to twenty-five

The classroom range, and the one most content is written for.

Three to five teams is enough for a genuine ecosystem: teams can trade, compete, misinform each other and form alliances, and the between-team behaviour becomes observable. One facilitator can still see everything, and the debrief can take contributions from the floor without structure.

This is the size at which the most is possible per participant, and it is the most expensive per head for exactly that reason.

Twenty-five to sixty

The awkward band, and the one most often got wrong.

It is too large for the facilitator to see everything and too small to justify the architecture that makes a large room work. Sessions in this range are frequently run as though they were twenty-person sessions, and the six or eight people who disengage are not noticed until the feedback.

Two things fix it. Add the second facilitator earlier than feels necessary — around forty rather than sixty. And move to table hosts: one person per table, recruited from the client, briefed the evening before, with a single job of signalling when their table is stuck or done.

Sixty to two hundred

Now it is a system, and it works well if it is designed as one.

Individual behaviour stops being visible and stops being the point. What becomes visible instead is something no small session can show: how a rule, an incentive or a piece of information propagates across forty tables. If the brief is about an organisation rather than a team, this is the size that demonstrates it.

The debrief changes completely — table conversation first, deliberate sampling second, facilitator landing third — and the facilitator's job shifts from responding to the room to having designed for it.

Cost per head falls sharply here, which is why large sessions get commissioned. That is a legitimate reason, as long as everyone understands they are buying a different product.

Above two hundred

Possible, with the right title, and rare.

Everything becomes logistics: materials distribution, audio, sightlines, the time it takes to get two hundred people back from a break. The design has to tolerate a wide spread between fastest and slowest tables, because at this scale the spread is guaranteed.

What you get is an event. Shared language across a large population in one day, which is genuinely valuable for a merger, a restructure or a strategy launch, and is not the same thing as development.

Choosing from the objective

Work backwards from what has to be true afterwards.

If a named individual has to change, the room has to be small enough for them to be seen: twelve or fewer.

If a team has to work differently together, the team has to be in the room as a team, and the session needs enough other teams around it to create pressure: fifteen to twenty-five.

If two functions have to stop working around each other, both have to be present in numbers: forty to eighty, with the table mix designed rather than left to chance.

If several hundred people need the same reference point, the constraint is the number, and the design should accept that it is doing one job rather than three.

The number that is usually wrong

Most briefs arrive with a number that came from a budget or a venue. It is worth one question before accepting it: what has to be different afterwards, and for whom?

Quite often the answer reveals that the forty people in the brief are actually two different populations who need two different sessions, and that running one session for both is the reason the last one did not land.