The Right Team Size: What Research on Small Groups Says

Why teams in a growing company become too large unnoticed

In a company of six, everyone knows what the others are doing. With twelve people this is no longer a given, with twenty certainly not. Yet the organization often stays the same: one team, one meeting, the owner in the middle. The meeting gets longer, decisions take time, and some colleagues no longer say anything in the round. Nobody decided that the team is too large. It simply grew.

This leads to the guiding question of this assessment: How large should a team in a small company be, and how can you tell that it should be split? To answer it, we first clarify the terms and introduce Steiner’s model of process losses as a thinking model. Then we review what studies show about team size and performance, weigh two perspectives, and derive check questions and a procedure for splitting.

A team does not become too large because it has many people, but because everyone has to talk about everything.

Terms: team, team size, span of control

A team here is a group of people who are responsible for a shared result and work together regularly to achieve it. It differs from a department, which is organized by specialty but whose members do not necessarily work on a shared result. Team size is the number of people who belong to a team and take part in its coordination. The span of control is the number of people who report directly to one manager.

Team size and span of control are often confused. An owner can lead twelve people working in three teams of four. Then their span of control is large and the teams are small. Conversely, a team of ten can have a single team lead whose span is ten. For the question of this article, team size matters most, because it determines how much coordination the work requires.

The thinking model: Steiner’s process losses

In 1972, the social psychologist Ivan Steiner presented a model that still shapes research on groups today. In one sentence: a group’s actual performance is its potential performance minus process losses. Potential performance grows with every person, because more knowledge, hands, and time are available. But process losses grow too, and faster.

Steiner distinguished two kinds of losses. Coordination losses arise because contributions must be aligned: who does what, when, and how does it fit together? Motivation losses arise because the individual is less visible in a larger group and their contribution counts for less. Research calls this social loafing. Around 1900, the French agricultural engineer Maximilien Ringelmann had already observed in rope pulling that people exert less force in groups than alone. The image is still apt: everyone pulls on the rope, but nobody knows who is pulling how hard.

Coordination losses can be made tangible with a simple calculation. In a team of n people, there are n times (n minus one) divided by two possible connections between any two people. This is the reason for the observation in the article on thinking in roles rather than people that the number of coordination paths rises faster than the number of people.

What studies show about team size and performance

Mueller (2012) tested Steiner’s model in a field study of 212 knowledge workers in 26 teams ranging from three to 19 members. Individuals performed worse in larger teams. The cause lay not only in coordination and motivation but in a third kind of loss she calls relational loss: as team size grew, employees perceived less support from colleagues. This support otherwise buffers the stress of work. The effect persisted when motivation and perceived coordination problems were controlled for.

The meta-analyses paint a more differentiated picture. Stewart (2006) summarized 93 studies on team design and performance. Project and management teams achieved slightly better results with somewhat more members, presumably because more knowledge and contacts came together. Byron, Keem, Darden, and two further authors (2023) analyzed 134 field studies with 11,353 teams on team creativity and innovation. Team size was unrelated to it. More important were task interdependence, shared goals, and leadership that supports autonomy.

Small is better or more minds? Two perspectives

The first view says: small teams are almost always better. Everyone knows everyone, decisions are made quickly, and nobody can hide. This view is supported by Steiner’s process losses, Ringelmann’s rope, and Mueller’s finding that support dwindles in larger teams.

The second view says: some tasks need more minds. A project that has to bring together engineering, purchasing, and the customer will not succeed with three people from one specialty. This view is supported by Stewart’s finding that project and management teams with more members did somewhat better, and by the finding of Byron and colleagues that size matters less for innovation than the nature of the collaboration.

I consider both views compatible if you look at the nature of the work. Where contributions interlock tightly, as in an installation team or a quoting team, coordination losses grow with every person, and the team should stay small. Where contributions run side by side and simply add up, as in a field sales force with separate territories, the team may be larger, because not everyone has to coordinate with everyone. The obvious question about the right number is therefore the wrong one. The better one is: how many people really have to coordinate with each other for this task?

How to tell that a team should be split

As a rule of thumb, not a rule, I consider four to eight people manageable for teams with tightly interlocking work. With eight people there are 28 possible connections, with twelve already 66. The article on OKR, Kanban, and weekly planning sets a similar limit at around ten people, up to which everyone talks to each other daily anyway. More important than the number, however, are the signals in daily work.

If three of the five signals apply, splitting pays off. The long meeting is often the first visible sign. How meetings with few participants lead to better decisions is described in the article on fewer meetings, better decisions. The agenda for a 30-minute weekly meeting is in the article on weekly planning in the team.

Splitting a team in four steps

Splitting a team is more than a new list. It changes who talks to whom and who is responsible for what. For the split to reduce process losses rather than merely shift them, it belongs in a fixed sequence.

1. Cut along the work

   Split by customers, orders, or processes, not by sympathy or seniority. Each new team should have a result it is largely responsible for itself.

2. Name the interfaces

   Write down where the teams still depend on each other, and set a fixed rule or a short meeting for each interface. That way one rule replaces many individual conversations.

3. Clarify roles and responsibilities

   Each team gets a person who leads the team meeting and orders priorities. Who is involved in decisions is set out in a RACI matrix, as described in the guide to responsibilities.

4. Review after three months

   Together you check whether meetings have become shorter, decisions faster, and duplicate work rarer. If not, the cut is corrected.

For the third step, the guide to clarifying responsibilities with the RACI matrix helps. It distinguishes who carries out a task, who is ultimately accountable, who is consulted, and who is informed. The group of those informed in particular should shrink after a split; otherwise the coordination effort stays the same. How tasks change owners with a clear handover is described in the guide to delegating tasks.

Conclusion: how large should a team be?

The question at the start was
large a team in a small company should be and how to tell that it should be split. Steiner’s model gives the answer in principle: with every person, potential performance grows, but process losses from coordination, declining effort, and dwindling support grow faster. The number of possible connections rises from ten with five people to 66 with twelve. The studies confirm the costs of size for the individual but also show that some tasks benefit from more minds.

There is therefore no universally right team size. Teams with tightly interlocking work should stay small, in my assessment at four to eight people; teams whose work runs side by side may be larger. Splitting is decided by the signals in daily work, not by the number. The sober observation is: a team is the right size when everyone knows whom they need to coordinate with, and with everyone else they do not.

The next step

• For each team, count the people who take part in its weekly meeting and calculate the possible connections.
• Check the five signals of a team that is too large and note which apply.
• For the largest team, draw who coordinates with whom in a typical week.
• Decide whether the work interlocks tightly or runs side by side, and derive from that whether the team must stay small.
• If three signals apply, plan the split along the work and set a review date after three months.