Teams & Organizational Performance

Here is a conversation I have had more times than I can count.

A leader tells me about their team. They walk me through the people one at a time. This one reads a client better than anyone in the building. This one builds the cleanest analysis you will ever see. This one has the deepest network in the region and a phone that never stops. The leader is proud of these people, and the pride is earned. Then they stop, and their voice changes, and they ask some version of the same question. So why is this not working?

That question deserves a real answer, because the roster is not the problem. That is exactly what makes it maddening. On paper, the leader did everything the playbook told them to do. Hire the best people you can find. Give them resources. Get out of their way. And the team still delivers less than the people inside it are worth.

I spent three decades in wealth management, most of them leading teams of advisors, and I watched this pattern hold across every level of talent I ever assembled. Great people, ordinary results. For years I treated it as a mystery. It is not a mystery. It is arithmetic, and almost nobody teaches it.

Individual talent means nothing until collective execution turns it into a business outcome. That sentence is the whole pillar. Everything below unpacks it.

What team performance actually is

Team performance is what a group produces together, which is a different number from what its members produce individually.

Read that again, because the entire industry blurs it. When an organization says it has a team problem, it usually means one of two things. It means the people are not good enough, so the fix is hiring. Or it means the people do not like each other enough, so the fix is an offsite. Both diagnoses skip the actual variable, which sits between the people, not inside them.

A team is an operating system, not a roster. A roster is a list of capabilities. An operating system decides how those capabilities combine, in what order, under whose authority, against which standard, with what happens when something goes wrong. Two organizations can hold identical rosters and produce wildly different results, because they are running different systems. You have seen this in sports, in markets, in your own company. You have probably lived it.

So when your team underperforms, the honest first question is not who. It is how. How does the work actually move between these people, and does anything in this system help it move well?

The assumption almost every buyer arrives with

Most leaders arrive at a team problem holding a talent hypothesis. It goes like this: performance equals the sum of the people, so a shortfall means the sum is too small. Add talent, subtract the weak link, and the number goes up.

That hypothesis is intuitive, nearly every leader holds it, and the research has been dismantling it for fifty years.

Ivan Steiner wrote the equation in 1972, and it still holds. Actual productivity equals potential productivity minus process losses. Potential is your roster: everything your people could produce if every one of them performed at full capability and every interaction between them worked perfectly. Process losses are what the system takes off the top. Steiner's point was that groups rarely reach their potential, because faulty group processes eat the difference.

Sit with the shape of that equation for a second. Hiring raises the first term. It does nothing to the second. A leader who keeps adding talent to a team with heavy process losses is pouring water into a bucket and never looking at the hole. The bucket gets more expensive. It does not get more full.

And the losses are measurable. In the 1880s, a French engineer named Max Ringelmann asked men to pull on a rope, alone and in groups, and measured the force. A single man averaged about 85 kilograms of force. In a seven-man group, the average dropped to 65 kilograms per man, and relative output per person fell to roughly half in groups of eight. Adding people added force. It also added loss, and the loss grew faster than anyone expected.

Bibb Latané, Kipling Williams, and Stephen Harkins picked the question back up in 1979 and separated the two mechanisms. Some of the loss came from coordination, people failing to synchronize. But some of it came from motivation: individuals simply exerting less effort when their contribution stopped being identifiable. They named it social loafing, and they showed the fix runs through accountability rather than inspiration.

Notice what that finding does to the standard playbook. The problem was never that the men on the rope lacked strength. Every one of them could pull 85 kilograms. The system took the rest.

What the research says about stacking stars

Two studies changed how I think about roster building.

The first came out of MIT and Carnegie Mellon in 2010. Anita Woolley, Christopher Chabris, Alex Pentland, Nada Hashmi, and Thomas Malone ran 699 people through groups of two to five and tested whether groups have something like general intelligence. They found it, and they called it the c factor. Here is the part that should stop you. The c factor did not track the average intelligence of the group's members, and it did not track the intelligence of the smartest member either. It tracked how well the members read each other socially and how evenly the group distributed its speaking turns. In the original data, that factor explained a large share of the variance across group tasks and predicted performance on a complex criterion task strongly, while average member IQ correlated with it only weakly.

Say that plainly. You can staff a group with the smartest people available and still build a group that thinks poorly, because group intelligence lives in the interaction, not in the members. Woolley and her colleagues put it directly: this kind of intelligence belongs to the group itself, not just to the individuals in it.

The second study came from Roderick Swaab, Michael Schaerer, Eric Anicich, Richard Ronay, and Adam Galinsky in 2014. They ran five studies on the relationship between talent and team performance. First they established what people believe, and people believe the line runs straight: more talent, better results, forever. Then they tested it against real performance data. In sports that demand heavy interdependence, football and basketball, talent improved performance up to a point, and past that point the benefit reversed as coordination broke down. The researchers named it the too-much-talent effect. In baseball, where players operate far more independently, the effect did not appear. Talent kept adding.

Interdependence is the whole hinge. When the work is independent, stacking talent works exactly the way your instincts say it should. When the work requires people to hand off, coordinate, and depend on each other under pressure, stacking talent starts to cost you, because star players begin competing for status instead of coordinating. That competition for status is the mechanism the researchers identified.

I will be straight with you about the state of this evidence, because &FOX does not oversell research. A later analysis by Bartosz Gula and colleagues re-examined the NBA data and argued the curve fits better as a diminishing return than as a reversal. The academic argument continues. What survives the argument, and what every model in this field agrees on, is the mechanism underneath: coordination costs rise with interdependence, and past some point they eat the benefit of the talent you added.

So ask the question that actually determines your team's ceiling. How interdependent is our work? If your people succeed or fail on their own, hire stars and stop reading. If your people cannot deliver the outcome without each other, then coordination is your product, and talent alone will not buy it.

The finding nobody wants to hear

J. Richard Hackman spent forty years at Harvard studying teams in hospitals, orchestras, and airline cockpits. He reached a conclusion that the entire teamwork industry would prefer to ignore. Research consistently shows that teams underperform despite all the extra resources they have. The reason he gave: problems with coordination and motivation typically chip away at the benefits of collaboration. He was blunt about the implication. Having a team is often worse than having no team at all.

Hackman was not against teams. He spent his life on them. He was against the assumption that gathering capable people produces a team, and his data kept showing how rarely it does. Most of the time, his research found, team members do not even agree on what the team exists to do. He also found that as a team grows, the effort required to manage the links between members rises almost exponentially, which is why he pushed leaders to keep teams small.

Here is why that finding is good news rather than bad. If teams underperformed because people are flawed, you would have no move. People are what they are. Teams underperform because of coordination and motivation, and both of those live in the system. Systems get designed. Which means the gap between your team's potential and your team's output is not a fact about your people. It is a decision nobody has made yet.

The three layers of a team operating system

Everything the research points to sorts into three layers. They stack. Each one sits on the one below it, and skipping a layer costs you the layers above.

The design. This is the structure your team runs inside before anyone shows up to a meeting. Who is actually on this team and who is not. What the team exists to produce. How the work divides. What the scoreboard measures. How the organization pays people, which is the loudest instruction any team ever receives. Hackman's five conditions in Leading Teams start here: a team has to be real, with clear boundaries, and it has to have a compelling direction that members know and agree on. Design is where the leverage concentrates, and it is where almost nobody looks, because design work happens before the problem becomes visible.

The standard. This is the shared picture of what "right" looks like. Not the values on the wall. The operating standard that lets one person predict what another person will do without asking. When the standard is genuinely shared, coordination stops needing narration and the team gets quieter and faster. When it is absent, the team compensates with meetings, and the meetings never close the gap, because the meetings are treating a symptom.

The trust. This is the layer that decides whether anyone says the hard thing while it still costs nothing to say. Trust of this kind does not come from liking each other. It comes from the reps: from working, and failing, in front of each other on purpose, until the group has evidence about how it behaves when things go wrong. Teams that never rehearse have no such evidence, so they guess, and under pressure they guess conservatively and stay quiet.

Design, standard, trust. In that order, because design determines what standard is even possible, and the standard determines what the trust gets used for. The cluster articles under this pillar take one layer each.

The line we hold

A word about how this work can go wrong, because it can.

Framing team performance as a system problem gives an organization a powerful new vocabulary, and any powerful vocabulary can get misused. I have watched "we have a design problem" become a polite way to reorganize people out of a job. I have watched "we need more interdependence" become a justification for eliminating anyone's ability to work alone. And I have watched leaders reach for the collective frame precisely when an individual accountability conversation was the honest one.

So we hold three lines. Systems thinking never becomes a hiding place for a leader who owes someone direct feedback. Design never becomes a euphemism for headcount. And nobody uses the language of the collective to spread out a responsibility that clearly belongs to one person.

The point of examining the structure first is not to absolve anyone. It is to make sure that when accountability lands, it lands on something the person actually controlled. That is fairer than what most teams do now, which is to name a person because naming a system is harder.

Why this matters more now than it did five years ago

Work got more interdependent, and it did not ask permission.

Look at what changed in the last few years. Flatter structures. Cross-functional pods. Matrix reporting. Distributed teams working across three time zones. AI absorbing the individual production tasks that used to fill a professional's day, which leaves behind exactly the work that requires judgment, coordination, and other people. Every one of those shifts moves your organization from the baseball end of the spectrum toward the basketball end. And the too-much-talent research is clear about what happens at the basketball end: coordination becomes the constraint, and the roster stops being the answer.

Meanwhile the capability to run that coordination has not kept up. Deloitte's 2025 Global Human Capital Trends survey, which polled nearly 10,000 business and HR leaders across 93 countries, found that only about a quarter of organizations rate their managers as very or extremely effective at enabling the performance of the people on their teams. Managers in the same research reported spending 13% of their time developing people. Nearly three-quarters of organizations say they need to reinvent the manager role, and 7% report making great progress on it.

Put those two curves next to each other. Interdependence is climbing. The capability to manage it is flat. That gap is the thing your team is currently absorbing, and it shows up on your P&L as talent you paid for and did not receive.

The practice for this week

Run an interdependence map. It takes an hour and it will tell you more than a survey will.

Take your team and write down what the team actually produces. Not the mission statement. The output a client or another department receives.

Now trace how that output gets made. For each step, write who does the work, who has to hand something to whom, and what has to be true for that handoff to land. Be specific enough that a stranger could follow it. Most leaders discover two things in this exercise. They find handoffs nobody owns. And they find that the number of real dependencies is far higher than the org chart suggests.

Then score it honestly. If your team's output is mostly the sum of individual work that happens to get stapled together, you have a group of individual contributors, and you should stop calling it a team and stop buying team interventions for it. Manage it as what it is. If your team's output genuinely requires people to depend on each other, then write down the three dependencies that hurt most when they fail. Those three are your operating system, and they are where the performance actually lives.

Last step, and it is the one that stings. Look at what your organization measures and rewards for each person on the team. Then look at your list of dependencies. If the scoreboard pays people for individual output while the work requires collective execution, you have found your process loss. You did not find a talent problem. You found a design that instructs your people to behave exactly the way they are behaving.

Back to the question

The leader who walks me through their brilliant roster and asks why it is not working has already given me the answer without knowing it. They described the people. They did not describe the system. Nobody ever taught them there was one.

That is not their failure. Every incentive in a professional career trains you to evaluate individuals. You get promoted for spotting talent, developing talent, and retaining talent. Nothing in that training prepares you for the moment when the talent is right there in front of you and the results are not.

Your people are probably not the problem. The space between them is. That space has a structure, whether you designed it or not, and it is producing exactly what its design calls for.

The good news is that space belongs to you.