The most capable people in the company often make up its worst functioning group. The research knows why.

Afterwards the real decision is made in the corridor, between Cecilia and two of the managers.

Nobody in the room is the problem. Each of them runs their own part of the business excellently. Still the sum is less than the parts, and Cecilia knows it: the leadership team meeting is the least productive hour of the week for the company's most expensive gathering.

There is plenty of research explaining why it turns out this way. This article goes through what it says, and why the mechanisms are the same as in the three previous articles in this series, only with higher stakes.

How common is it? The numbers from the largest study

The most comprehensive investigation of leadership teams specifically was carried out by Ruth Wageman, Debra Nunes, James Burruss and Richard Hackman. Over several years they studied more than a hundred senior and top teams in international companies using a standardized diagnostic instrument, and summarized the results in the book Senior Leadership Teams (2008).

One of the study's most cited findings is also its most telling. When members were asked to list who belonged to their own leadership team, only a small minority of the groups gave answers that agreed. People who met every week were not even in agreement about who was in the group.

That this matters was already well established. Donald Hambrick and Phyllis Mason showed back in the 1980s, in what has come to be called upper echelons theory, that organizations largely reflect their leadership teams: how the group is composed and how it functions shows up in the company's strategic choices and results. A leadership team that underperforms is therefore not an internal problem for eight people. It is a problem for the whole organization.

A team in name, not in practice

The first and most common failure is that many leadership teams are not teams in any meaningful sense.

Richard Hackman devoted much of his career to the question of what makes a team real. His answer: a bounded membership that is stable over time and that has a shared task where the members depend on each other to solve it. Jon Katzenbach and Douglas Smith reached the same conclusion in The Wisdom of Teams: what looks like a team on the organizational chart is often a working group, people who inform each other but solve their tasks separately.

Cecilia's Monday meeting is a textbook case. Eleven items of reporting is not joint work. It is eight separate jobs presented in turn, in front of an audience. No item requires the group to think together, and so it does not. Wageman and colleagues found this to be a consistent pattern in the weaker groups: meetings were filled with information that could have been sent in advance, while the questions that genuinely required the group's combined judgement, the strategic choices, never got the time.

A group without a shared task cannot perform as a group. There is nothing to perform on.

The information that never reaches the table

The second failure is subtler and affects even groups that genuinely try to work together.

The whole point of a leadership team is to bring together different perspectives: the CFO knows things the sales director does not, and vice versa. Yet as early as 1985 Garold Stasser and William Titus showed in a series of experiments that groups are surprisingly bad at using precisely what makes them valuable. When participants were given different pieces of information, the groups spent the discussion on what everyone already knew, the shared information, while the unique pieces never came up. The groups therefore often chose the wrong option, even though between them they held all the information needed to choose correctly.

The phenomenon has been replicated many times since, and it is reinforced by status. Repeating what everyone already knows is socially safe: nobody objects, everyone nods. Putting forward something only you know, particularly if it cuts against the CEO's line or a colleague's proposal, is a risk. In a leadership team, where everyone around the table is used to being right and where positions are guarded, that risk is greatest exactly where the value of unique information is highest.

The conflict that is missing, and the conflict there is too much of

The third failure concerns how groups handle disagreement, and here the research offers a more nuanced picture than the common notion that conflict is always harmful.

Frank de Wit, Lindred Greer and Karen Jehn brought together the results of more than a hundred studies of conflict in work groups. Two patterns emerged. Relationship conflict, where the disagreement becomes personal, about who someone is rather than what should be done, was consistently associated with worse performance. Task conflict, disagreement about the issues themselves, was on average neither good nor bad. But in leadership teams specifically the association was positive: top teams that argued about the substance performed better.

Many leadership teams manage to get it the wrong way round on both counts. Task conflict is suppressed (nobody challenges the forecast in the room, it is done in the corridor afterwards) while old relationship conflicts sit and chafe beneath the polite surface. The result is a group that is too harmonious on the surface and too infected below it.

What decides whether a group can handle task conflict without it turning into relationship conflict has a name, given to it by Amy Edmondson: psychological safety. In her study of 51 work teams at a manufacturing company, it was the shared belief that interpersonal risks could be taken (questioning, admitting mistakes, asking for help) that explained why some teams learned and improved while others stood still. Without that safety every objection becomes a personal attack, and then sensible people choose silence.

The same mechanisms as earlier in the series, but with higher stakes

Anyone who has read the three previous articles will recognize the pattern, because a leadership team is a concentration of everything they were about.

Everyone around the table lacks reliable insight into their own behavior, we cannot rate ourselves from the outside. Everyone has the defensive routines Chris Argyris described, and they are strongest in successful people who have rarely failed, which is an apt description of most of those who reach a leadership team. Argyris incidentally developed his ideas working with company leadership groups, and he noted something every leadership team recognizes: there are questions nobody raises, and the fact that nobody raises them is itself something nobody raises.

And the feedback that could break the pattern does not arrive. Managers at this level receive less honest feedback than anyone else in the organization: their reports do not dare, the colleagues around the table are competing, and the board only sees the results. The gap between how each of them believes they act in the group and how they are actually perceived by the others can therefore grow for years without anyone saying anything.

A leadership team is therefore not a place where the individual mechanisms are weakened by experience and competence. It is a place where they are amplified by status, competition and silence.

What the functioning groups do differently

Wageman and colleagues found that the high-performing groups differed from the rest on a few concrete points, and that none of them had to do with the members' ability.

They are real teams. A clear and stable membership, where everyone knows who belongs and why those particular people are there.

They have a task that requires the group. A small number of questions that can only be solved jointly (strategy, priorities, direction) instead of an agenda of reporting. What can be sent in advance is sent in advance.

Composition is an active decision. The CEOs who led the best groups had taken a position on who should be part of them, and dared to change it when it was not working, rather than letting the organizational chart decide.

The interaction itself is maintained. The best groups worked on how they worked: they obtained feedback on their own behavior as a group and took outside help to develop it. Performance was not treated as an automatic consequence of the members' competence, but as something the group had to build.

Add Edmondson's findings and a clear picture emerges: what separates functioning leadership teams from the rest is structural choices and behavioral patterns. Both can be changed. But behavioral patterns can only be changed by those who see them, and as the earlier articles in the series showed, that is precisely what the members cannot do on their own.

A shared language for the interaction

This is where IDI is used in team development. The starting point is the same as throughout the series: what matters is not how each person in the group describes themselves, but how they are perceived by the people around them, including by each other.

In team development work with IDI, each member receives a behavioral profile built on the observations of others, set against their own self-image. When the profiles are then placed side by side, the group sees its own pattern: where the styles collide, where they complement each other, and why certain discussions always get stuck between the same people.

Two properties of the method make it useful for exactly what the research points to. No behavioral style is better than any other, the profile describes, it does not grade. That allows the group to talk about its interaction in substantive terms, as behavior that can be adapted, rather than as judgements about people. That is precisely the difference between task conflict and relationship conflict: a shared, neutral language for behavior makes it possible to disagree about the issues without it becoming about who someone is.

The work is led by a certified IDI facilitator. For Cecilia's group that would mean that what today sits beneath the surface (why the pricing question never fits, why the decisions are made in the corridor) becomes something the group can see and talk about, in words that do not single anyone out.

Competence is not the problem

Leadership teams do not underperform because the wrong people were recruited or because the members know too little. They underperform because a collection of capable individuals does not automatically become a capable group, and because the mechanisms that prevent it, from unclear tasks to invisible behavioral patterns, are rarely visible from the inside.

On that point the research is hopeful. What separates the best groups is not ability but structure and interaction, and both can be worked on. But the first step is the same for a group as for an individual: to obtain an outside view of how you actually function.

Cecilia's group does not need to get smarter. It needs to see itself.