The myth of the united team (why assembling stars isn’t enough)

Fourteen people in a room

A professional-services firm has a serious problem and decides to attack it the way you attack what matters: it puts the best people on it. It sets up a committee of fourteen —the brightest from each area, the ones who "can't be left out"— and asks them for a solution in ninety days. The first meeting is electric: there is energy, ideas, impeccable résumés around the table. Everyone leaves convinced that, with a group like this, the result is assured.

Three months later, the committee delivers a lukewarm document, late, that no one quite owns. No one was incompetent. No one sabotaged it. And yet the whole performed worse than almost any of its members would have on their own. At the closing meeting, someone sums it up with the sentence heard in every latitude: "we had the best people, I don't understand what happened".

What happened is the most studied and least understood thing in organizational life: gathering talent is not the same as building a team. A group of excellent people can produce a mediocre result with a facility that is frightening —and it does so all the time, in committees, task forces, boards and "teams" that share nothing but the name—.

This article —the first of a series of three— is about why that happens, and about the evidence, accumulated over half a century, that dismantles the most expensive myth of teamwork: that it is enough to gather the right people and wait for chemistry to do the rest. The conclusion, given away up front, is harsh and liberating at once: high performance is not a lucky accident or a trait of collective personality. It is an artifact of design.

The team that performs worse than its parts

Let's start with the most counterintuitive and best-documented fact: a group usually produces less than the sum of what its members would produce working alone. It is not office pessimism; it is a regularity measured more than a century ago.

In the late nineteenth century, the French agricultural engineer Maximilien Ringelmann had men pull on a rope, alone and in groups, and measured the force with a dynamometer. An isolated individual exerted on average some 85 kilos of pull. Placed in a group of seven, each person contributed 65; in one of fourteen, 61. Translated into percentages, the classic series is eloquent: effort per person falls to around 93% in a pair, 85% in a trio and just 49% in a group of eight. Half the muscle evaporates along the way.

Part of that drop is coordination; the other part —the one that matters— is motivation. When individual effort dissolves into the collective result, people ease off without meaning to. It is social loafing, and it is not an anecdote: the meta-analysis by Karau and Williams, which pooled 78 studies, found an effect of moderate magnitude —a Cohen's d of around 0.44— consistent across tasks, countries and people. It also eases when each person's contribution is visible and the task matters: the first hint that we are dealing with something designable, not with a fatality.

Decades earlier, Ivan Steiner put a formula to it: actual productivity = potential productivity − process losses. The distance between what the group could achieve and what it does achieve is the meetings that decide nothing, the information that does not circulate, the efforts that step on each other. The management question is not whether those losses exist —they always do—, but how far they are reduced.

And there is a modern aggravator. According to Rob Cross, Reb Rebele and Adam Grant, time devoted to collaborative activities grew by 50% or more over the last two decades, to the point that in many companies people spend close to 80% of their day collaborating. Worse: the load concentrates —between 20% and 35% of the valuable collaboration comes from just 3% to 5% of the people, the involuntary bottlenecks everyone turns to until they burn them out—.

A group and a team are not the same thing

A distinction is worth making, one Jon Katzenbach and Douglas Smith nailed in 1993 after studying more than fifty teams across thirty companies. Not every group that works together is a team. A working group comes together to share information; its accountability and its products are individual. A team produces collective products —things that only exist if several people do real work together— and answers mutually for them. That mutual accountability is the line that separates a motivating label from a performance unit.

Table — Working group or team: how to tell them apart. Adapted from Katzenbach and Smith, 1993.

The expensive mistake is not having working groups —they are needed—, but calling everything a "team" and expecting team performance without setting up any of the conditions that would make it possible. The fourteen in our room were, strictly speaking, a group asked to function as a team without being given anything a team needs.

Chemistry is an expensive myth

And what does a team need? J. Richard Hackman, of Harvard, studied teams for more than forty years —from flight crews to symphony orchestras—, and his central finding defies the dominant intuition: what makes or breaks a team is not the personalities or styles of its members, but a handful of enabling conditions that can be set up in advance. The question "do we have the right chemistry?" is the least important.

The most famous demonstration came not from academia but from Google. Between 2012 and 2015, in an internal study they called Project Aristotle, they analyzed 180 teams (115 in engineering and 65 in sales) against more than 250 variables. The result baffled their own analysts: who was on the team mattered much less than how they worked together. When they isolated what distinguished the best, a hierarchy of five dynamics appeared —and one, by far, above the rest—.

The five factors —psychological safety, dependability, structure and clarity, meaning and impact— are not personality traits: they are properties of the environment the team inhabits, and all can be built. (A note of honesty is in order: Project Aristotle is a company's internal research, not a peer-reviewed study, and the "43% of the variance" figure sometimes attributed to it does not appear in the source. It is worth taking as powerful evidence, convergent with the academic literature, not as a laboratory experiment.)

Performance is decided before you start

If the conditions are set up in advance, the decisive moment of leadership is not the match: it is the pre-season. The field's researchers —Hackman and Ruth Wageman among them— sum it up in a heuristic, the 60-30-10 rule: a leader should put close to 60% of their energy into the up-front design —purpose, people, size, structure, norms—; 30% into the launch; and just 10% into on-the-fly coaching. The bulk of the result is already played out before the first working meeting.

The label matters: 60-30-10 is a practitioner heuristic, useful as a compass, not a validated statistic. But it agrees with everything else we know: teams that work were almost always well built, and those that fail carry factory defects no later coaching can offset. It is the same pattern as in capital projects: the fate is set early, in decisions almost no one looks at. The committee of fourteen did not fail on day 80; it failed the day it was decided there would be fourteen of them, with no specific purpose, no roles, no norms and no one answering for the collective result. The rest was consequence.

Bringing together the best and waiting for magic to happen is the most expensive and most common way to build a team. Collective performance does not spring from accumulated talent or good vibes, but from concrete conditions someone has to design. In Part 2 we get concrete: what they are, one by one and with the evidence in hand, the levers that actually move the needle.