Abstract
Knowledge-work teams face a cluster of problems that feed into one another: invisible output, work that passes back and forth between specialists, membership in several teams at once, leaky boundaries, distance, and collaboration overload that falls on a few people. Organisations usually respond by adding people, meetings and tools, and this article explains why those responses tend to deepen the losses.
It starts with the oldest measured finding about teams: Ringelmann’s discovery that fourteen people pull with the force of about ten, and follows the questions a leader would ask: where the effort goes, why each new person costs more than the last, and why the best people burn out.
The effort goes mostly into the gap between what people do and what anyone can see, a gap that widens with size and distance. Each new person costs more than the last because potential connections grow faster than headcount, most knowledge work depends on everyone, and collaboration concentrates on the able and willing until they become bottlenecks.
The longest part, drawing on the author’s doctoral research with 48 knowledge-work teams and on research published to 2024, explains why the same mix of personalities thrives in one team and fails in another: how evenly a trait is spread matters as much as how much of it a team has, and the work a team does at its edges can turn a helpful mix into a harmful one. The closing chapters set out the design levers a leader can use, with the strength of the evidence for each.
Keywords: Ringelmann effect; social loafing; process loss; coordination loss; motivation loss; moral disengagement; team size; collaboration overload; network centrality; task interdependence; task complexity; knowledge work; team personality composition; Big Five; trait activation; boundary management.
PART I. BUSY, CONNECTED AND STALLED
Chapter 1: The Problem Nobody Designed
1.1 Fourteen People on a Call
Picture a programme you may well have worked on; this one is a composite, and it will follow us through the article. It is the third steering group of the quarter on a programme meant to take six weeks; fourteen people are on the call, nine with cameras off, and the update slides say the work is ninety per cent complete, as they said a month ago. Every workstream lead can describe what they have done, and every one of them is waiting on somebody else. The two people everyone needs, an architect and a commercial lead, have between them forty unread requests and a calendar with no gap longer than half an hour until the following Thursday.
Nobody in the meeting is idle. Most of them worked late the night before. The programme director’s response, on the evidence of the last two quarters, will be to ask for three more people, and the rest of this article explains why that is likely to make the programme slower, not faster.
1.2 Nine Problems That Feed Each Other
Most leaders of knowledge workers will recognise some version of that scene, and it is so common because managing knowledge work means managing several hard problems at once, each of which makes the others worse. Peter Drucker argued at the end of the last century that raising the productivity of knowledge workers would be the central management challenge of the twenty-first century, precisely because their work could not be measured the way manual work had been (Drucker, 1999). A quarter of a century later, the challenge has multiplied rather than eased.
- Output you cannot see. Knowledge work produces analysis, designs, decisions and advice, which are intangible and hard to attribute to any one person (Drucker, 1999; Keca, 2019). When nobody can see who contributed what, effort becomes easy to withhold and hard to reward.
- Work that bounces between specialists. Most knowledge work passes back and forth between people with different expertise, in loops rather than in sequence (Thompson, 1967), so every task depends on several people being available at the right moment.
- Complexity and uncertainty. The work involves many pieces of information whose relationships change as it progresses (Wood, 1986), so the plan is always being renegotiated.
- People on several teams at once. At least 65 per cent of knowledge workers are members of more than one team at a time, and one survey put the figure as high as 95 per cent (O’Leary, Mortensen & Woolley, 2011), so every team is competing with others for the same people’s attention.
- Boundaries that leak. Knowledge-work teams depend on information, resources and decisions from outside the team, and much of their work happens at its edges (Ancona & Caldwell, 1992; Faraj & Yan, 2009).
- Distance. When Microsoft moved more than 61,000 employees home in 2020, collaboration with people in other groups fell by about a quarter, and groups became more inward-looking (Yang et al., 2022).
- Collaboration overload. The time managers and employees spend on collaborative work has grown by 50 per cent or more over two decades, and Rob Cross and his colleagues estimate that at many companies meetings, calls and email now take around 80 per cent of working time (Cross, Rebele & Grant, 2016).
- A load that falls on the few. In many organisations, 3 to 5 per cent of employees provide 20 to 35 per cent of value-added collaboration, and they end up among the least engaged (Cross et al., 2016).
- Team mixes nobody chose. Teams tend to attract and keep people like themselves (Schneider, 1987), so the combination of personalities in a team is usually an accident of history rather than a design decision.
None of these problems sits on its own. Invisible output makes collaboration overload harder to see, multiple team membership feeds the load on the few, distance shrinks the networks that would spread that load, complexity raises the cost of every handover, and the mix of people in the team decides whether the pressure is absorbed or amplified. That is why fixing one at a time so rarely works.
1.3 Why the Usual Fixes Make It Worse
Most organisations reach for reasonable responses, and most of them make the problem worse. Adding people to a late piece of work adds communication paths and pulls the existing team away from the work to bring the newcomers up to speed (Brooks, 1975). Adding meetings, channels and tools adds interruptions, and every switch between tasks leaves part of a person’s attention behind on the last one (Leroy, 2009). Leslie Perlow, who spent nine months with a software team developing a new product, called the result a time famine, in which collaborative demands crowded out the concentrated work the product also needed and people worked longer hours to make up for time the group itself consumed (Perlow, 1999).
Exhortation fails for a different reason. Campaigns to raise engagement assume that the people who are holding back have chosen to, but, as Chapter 3 shows, much of the effort a team loses is lost by people who would pull just as hard alone. Hiring stars further concentrates the load on the people already most in demand, and profiling a team’s average personality tells a leader surprisingly little about how the team will perform (Han, Krieger, Kim, Nixon & Greiff, 2024).
1.4 What This Article Argues
Organisations have built knowledge work around teams and expect larger, better-connected teams to deliver more, but every person added loses some effort to invisibility and some time to coordination; the load piles up on a few willing people, and the way members’ personalities combine can amplify each of those losses. Therefore the struggle of knowledge-work teams is not mainly a problem of motivation, culture or individual time management, but a problem of design, and it has a long and surprisingly well-measured history.
I came to this question from operations leadership rather than from the laboratory. In my experience, across senior operations roles in small, medium and multinational organisations, the way teams were composed and managed was strikingly inconsistent, particularly in matrix structures, and that puzzle became the starting point for my doctoral research (Keca, 2019).
The article follows the questions a leader in the programme director’s position would ask. Part II asks where the effort goes, Part III why every new person costs more than the last, Part IV why the best people burn out, Part V why the same mix of people thrives in one team and fails in another, which is where my own research sits, and Part VI what a leader can change on Monday morning. The story starts in the nineteenth century, with a French agricultural engineer, a rope and a gauge.
Chapter 2: Fourteen People, the Force of Ten
2.1 The Rope
Put one person on a rope attached to a gauge and ask them to pull as hard as they can, then put seven people on it, then fourteen. Common sense says fourteen people should pull fourteen times as hard. In controlled trials in the 1880s, people pulling alone exerted an average of 85.3 kg; in groups of seven, the average per person fell to 65.0 kg, and in groups of fourteen to 61.4 kg, so that fourteen people pulled with the force of about ten (Kravitz & Martin, 1986, p. 937).
That loss, the fall in each person’s contribution as the group grows, is the Ringelmann effect. It is the oldest measured finding about teams, and it is the right place to start, because the trials already contained both of the problems that make modern knowledge work so hard to manage.
2.2 The Engineer
Maximilien Ringelmann was a French agricultural engineer, a professor at the National Institute of Agronomy, interested in the mechanical efficiency of people and draught animals. He collected most of his data between 1882 and 1887 and published them in 1913, which Kravitz and Martin suggest makes his study the first social-psychological experiment, ahead of Norman Triplett’s better-known study of competitive cycling (Kravitz & Martin, 1986; Triplett, 1898).
2.3 The Numbers on the Slide Are Not His
If you have seen the Ringelmann effect on a training slide, you have probably seen a different set of numbers: 100 per cent of individual effort alone, 93 per cent in pairs, 85 per cent in threes and 49 per cent in groups of eight. They do not come from the controlled rope trials but from a much less well-documented set of field observations, which reached English-language psychology second-hand through Moede (1927). Kravitz and Martin (1986, p. 940) note the irony that these were the data that had such an impact on social psychology. For decades, the secondary literature also assumed Ringelmann was a German psychologist working in the 1920s.
The difference matters to anyone planning a team. The famous series falls in an almost perfectly straight line, inviting the idea that every extra person costs the same and leading to the familiar slide that extrapolates teams into collapse, but the controlled data do not behave that way. They bend. The losses are steepest with the first few additions, then the curve flattens, as later, more rigorous research also found.
Figure 1 (proposed): the widely quoted straight-line series set against the bending pattern in Ringelmann’s own controlled data and in Ingham and colleagues (1974).
2.4 Two Explanations He Already Had
Ringelmann attributed the losses mainly to what he called a lack of simultaneity, meaning people failed to pull at exactly the same moment and in the same direction, a loss of coordination. He also recorded a second possibility, in an anecdote about prisoners turning the crank of a flour mill, in which each man relied on his neighbour to provide the effort and contented himself with following the movement of the handle, which is a loss of motivation (Kravitz & Martin, 1986, pp. 937 to 938). The founder of the field saw both mechanisms in the 1880s. It took psychology roughly ninety years to separate them, and the experiment that did it is the best answer yet to the question of where a team’s effort goes.
PART II. WHERE DOES THE EFFORT GO?
Chapter 3: Pulling With Nobody Behind You
3.1 Two Ways to Lose
In 1972, Ivan Steiner gave group research its most useful piece of arithmetic. He argued that what a group actually produces is what its members could produce in principle, minus whatever the process of working together uses up. He split that loss into two parts: coordination loss, from imperfectly combining members’ efforts, and motivation loss, from members not giving their full effort (Steiner, 1972). The sum is simple, and that is its strength, because it tells a leader that an underperforming team is not necessarily a team of underperformers, and that the same people, arranged differently, could produce more.
Figure 2 (proposed): Steiner’s sum drawn as a bar, with potential at full height and coordination loss and motivation loss taken off in turn to leave what the group actually produces.
3.2 The Blindfold Experiment
Two years later, Alan Ingham and his colleagues found a way to tell the two losses apart (Ingham, Levinger, Graves & Peckham, 1974). Their first study replicated Ringelmann with real groups: individual effort fell by about 9 per cent with one co-worker and 18 per cent with two, and after that adding people made little further difference, with groups of three pulling at 82 per cent of their members’ average solo effort and groups of six at 78 per cent.
The second study is the clever part. Participants were blindfolded and placed at the front of the rope, and behind them trained confederates, who were members of the research team posing as fellow participants, went through the motions of pulling but applied no force at all, so each participant was in fact pulling alone, and there was no coordination to lose. Even so, effort fell when people believed they were part of a group, to 85 per cent of their solo effort with three perceived co-workers and 86 per cent with six (Ingham et al., 1974, pp. 377, 380).
The gap between the two studies gives a rough estimate of coordination loss; what remains when coordination is removed is motivation loss, which is larger. Nobody in the second study could see anyone else. They simply believed others were there, and that belief alone was enough to reduce their effort by a seventh.
3.3 What That Means for the Programme
The programme team in Chapter 1 is not a team of shirkers, and the blindfold experiment explains why that does not help. Most of the people on that call would work just as hard as anyone else if the work were theirs alone, but each of them is one of fourteen names on a status report, contributing to outputs that nobody can trace back to them, and that is exactly the situation in which effort quietly drains away.
Chapter 4: Why Good People Pull Less
4.1 Three Switches
Bibb Latané, Kipling Williams and Stephen Harkins gave the phenomenon its name, social loafing, in experiments in which people shouted and clapped as loudly as they could, alone or in groups (Latané, Williams & Harkins, 1979). Later work identified three conditions that switch it on, and each of them is a feature of ordinary knowledge work.
The first is anonymity: people loaf less when their individual output can be identified and evaluated, and the loss largely disappeared in the cheering experiments once each voice could be picked out (Williams, Harkins & Latané, 1981). The second is redundancy, where people hold back because they believe their contribution makes no difference when someone else will provide it, which is the free-rider effect (Kerr & Bruun, 1983). The third is the fear of being taken for a fool, in which people withdraw effort to avoid being exploited by colleagues they suspect of loafing, which Norbert Kerr called the sucker effect (Kerr, 1983).
Steven Karau and Kipling Williams brought these together in a review of 78 studies and a model with a simple core: people work hard on collective tasks when they expect their own effort to make a difference to an outcome they value (Karau & Williams, 1993). Anything that breaks the chain from my effort to the group’s performance to something I care about invites withdrawal, and knowledge work breaks that chain at every link.
4.2 It Happens at Work, Not Just in Laboratories
Most of this evidence comes from laboratories, but it holds up in real workplaces. In 23 intact work groups across two organisations, Liden, Wayne, Jaworski and Bennett (2004) found that managers rated social loafing higher in larger and less cohesive groups, and higher among people who saw their work as less visible to their manager, who felt less fairly treated, and who saw their work as more interdependent with that of colleagues. That last finding runs against intuition, because interdependence is usually sold as what binds a team together, and Chapter 7 returns to it.
The same study produced a hopeful anomaly. People who saw their colleagues loafing loafed less, not more, and the authors read this as social compensation: when the work matters to them, some people pull harder to make up for colleagues they think are not pulling their weight (Liden et al., 2004). The sucker effect and social compensation are two answers to the same situation, and which one appears depends on how much the person cares about the outcome.
There is also a mirror image of loafing, in which the weakest member of some groups works harder because the group’s result depends on them and their effort is plainly indispensable. This motivation gain is known as the Köhler effect, and it is the clearest evidence that group work does not have to sap effort (Hertel, Kerr & Messé, 2000).
4.3 What Bigger Teams Do to People
The losses are not confined to effort. In a study of individuals working in teams, Mueller (2012) found that people in larger teams performed worse partly because of what she called relational loss: as teams grow, members feel less sure support will be there when they need it. In a study of work teams, Aubé, Rousseau and Tremblay (2011) found that team size harmed the quality of members’ experience of the group, and that this was carried by counterproductive behaviour.
The largest synthesis to date puts the point sharply. Pooling 208 samples covering 21,435 teams, Bernerth, Beus, Helmuth and Boyd (2023) found that larger teams were more likely to show deviant behaviour and passive withdrawal, the kind of quiet disengagement that rarely appears in a status report, and that they held slightly less favourable attitudes towards their team. The nine cameras that stayed off during the programme call aren't proof of anything, but they show what withdrawal tends to look like from the outside.
Chapter 5: When Coasting Becomes Easy to Justify
5.1 Switching Off the Inner Critic
Loafing sounds harmless, like a moment of slack, but the psychology that allows it can go further. Albert Bandura described moral disengagement as a set of mental moves that let people switch off the self-criticism that normally keeps their behaviour in line with their own standards, and two of these moves are directly relevant to teams (Bandura, 1999).
The first is the diffusion of responsibility: when everyone is responsible, no one feels responsible. The second is displacing responsibility onto something else, which in organisations has a familiar sound: it’s the process, it’s the leadership, it’s the other team. Both are close relatives of the conditions behind social loafing, and both grow more available as a team grows and its work becomes harder to attribute.
5.2 Distance Makes It Easier
Distance makes that story easier to tell. In technology-supported teams, larger and more geographically dispersed teams showed more social loafing, and moral disengagement carried the link (Alnuaimi, Robert & Maruping, 2010): the structural features of distributed teams appeared to make holding back easier to justify. The teams were students reporting on themselves, so the finding still needs testing in workplaces, but it fits everything else in this part of the article.
More broadly, the evidence that moral disengagement matters is now cumulative. A synthesis of the workplace research found that people higher in moral disengagement engaged in more misconduct, helped colleagues less, performed their own tasks less well and were more likely to intend to leave (Ogunfowora, Nguyen, Steel & Hwang, 2022). The same review found that abusive supervision and office politics strongly enabled it, while ethical leadership and fair treatment were relatively weak deterrents; earlier work also showed that moral disengagement predicts a range of unethical behaviour at work (Moore, Detert, Treviño, Baker & Mayer, 2012).
5.3 A Story People Tell Themselves
None of this makes coasting a moral failing, and most people who ease off in a large team are not making a moral choice at all. The point is subtler: the same structures that make individual contribution invisible also make it easier to tell yourself a story in which your contribution does not matter, and the research on moral disengagement describes exactly how that story is told.
My own suspicion, which nobody has yet tested in knowledge-work teams, is that once that story takes hold it does not always stay confined to effort, and that the corner-cutting it licenses can spread.
Figure 3 (proposed): the chain from invisible contribution, to feeling dispensable, to diffused responsibility, to moral disengagement, with the last link marked as inference.
5.4 Where Part II Leaves Us
The effort a team loses mostly goes into the gap between what people do and what anyone can see them do, and it widens as teams grow, disperse, and blur responsibility. That is only half of Steiner’s sum. The other half, the time and attention lost to coordinating more and more people, is where knowledge work differs most from a rope, and it explains why the programme director’s three extra people will cost more than they add.
PART III. WHY DOES EVERY NEW PERSON COST MORE THAN THE LAST?
Chapter 6: Six Relationships, Then Ninety-One
6.1 The Arithmetic of Pairs
If motivation loss is about invisibility, coordination loss in knowledge work is about the sheer volume of interaction. In 1945, the family sociologist James Bossard set out what he called the law of family interaction: as a group gains members one at a time, the number of members grows by one each time, but the number of relationships between pairs grows according to the triangular numbers, so that a family of four has six relationships and a family of twelve has sixty-six (Bossard, 1945). The fourteen people on the programme call have ninety-one, and the three extra people the director wants would take that to one hundred and thirty-six.
Bossard illustrated the point with Benjamin Franklin, one of seventeen children, and Frederick Brooks later applied the same pairwise logic to software. Brooks’s law, a veteran’s rule of thumb rather than a measured result, says that adding people to a late software project makes it later, partly because each new person adds communication paths and partly because the existing team must stop to bring them up to speed (Brooks, 1975).
6.2 Subgroups, Managers and the Explosion of Links
Pairs are only the beginning, as William Kephart pointed out in 1950, because people also relate to subgroups, such as parents, engineers or the London office, and subgroups relate to each other. Counting all of these gives 25 potential relationships in a group of four, 90 in a group of five, 966 in a group of seven and 3,025 in a group of eight; in his example, adding one in-law to a household of five creates 211 new potential relationships (Kephart, 1950, p. 548).
Earlier still, in 1933, the management consultant V. A. Graicunas had counted the relationships a manager must attend to: with each subordinate directly, between subordinates, and with groups of subordinates. His count gives 44 relationships for four direct reports, 100 for five, 222 for six, and 1,080 for eight; he recommended no more than five reports, and probably four (Graicunas, 1933/1937, as reported in Nickols, 2011). He reportedly described adding a fifth report as “an increase in complexity of 127 per cent in return for a 20 per cent increase in working capacity” (quoted in Nickols, 2011, p. 3) and noted that the heads of closely integrated divisions are harder to manage than the heads of independent ones, which is the earliest version of the interdependence argument in Chapter 7.
6.3 Potential, Not Actual
These numbers are striking, and they appear in countless slide decks about team size, but the people who calculated them knew they count potential relationships, not actual ones. Kephart said so explicitly: he expected the great majority of potential relationships in larger groups to stay “nonoperative”, proposed that a group be described by the ratio of the relationships it actually uses to those it could use, and warned that no causal link to efficiency should be assumed (Kephart, 1950, pp. 548 to 549).
Real teams use only a fraction of the links available to them, which is consistent with the flattening in Ingham’s data. Kephart still drew a clear conclusion, and it is the best historical line on the subject: “primary groups must remain small”, he wrote, and perhaps smaller than had previously been supposed (Kephart, 1950, p. 548).
Figure 4 (proposed): potential relationships by team size under Bossard, Kephart and Graicunas, captioned “potential, not actual” and annotated with Kephart’s ratio of actual to potential relationships.
6.4 So Is Smaller Always Better?
The modern evidence says no, and the reason is useful. Hackman and Vidmar (1970) found that members judged teams of about four to five to be the right size, and Hoegl (2005) found that smaller project teams showed better teamwork quality, but the largest synthesis complicates the picture: across 208 samples and 21,435 teams, team size had no average relationship with team task performance, and the relationship varied enormously from one setting to another (Bernerth et al., 2023).
The variation was not random. Larger teams performed better on complex tasks because they brought more knowledge and skill to bear, and worse on simple tasks that still required extensive coordination, which is Steiner’s process loss in its purest form (Bernerth et al., 2023). Wu, Wang, and Evans (2019), studying more than 65 million papers, patents, and software projects, found a related trade-off: large teams tend to develop existing ideas, while small teams are more likely to disrupt them.
Size buys expertise at the cost of coordination, and it also costs commitment, as Chapter 4 showed. What the arithmetic does capture is that the coordination cost grows faster than headcount, and in knowledge work that cost is paid in the scarcest resource of all, which is attention: when people switch between tasks, part of their attention stays on the previous one, and this attention residue lowers performance on whatever comes next, so that every extra thread, channel and meeting fragments attention a little further (Leroy, 2009).
Chapter 7: When the Work Depends on Everyone
7.1 Why a Rope Is the Easy Case
Whether adding people helps or hurts depends on the task, and the rope is the easiest task. Steiner (1972) classified group tasks by how members’ contributions combine and named five kinds.
- In an additive task, the group’s output is the sum of what members put in, as with pulling a rope or working through a backlog of independent cases.
- In a compensatory task, the group’s output is the average of members’ judgements, as when independent estimates are combined into a forecast.
- In a disjunctive task, the group succeeds if its best member succeeds, as when one person spots the root cause of a fault.
- In a conjunctive task, the group is only as good as its weakest member, as with a release that needs every sign-off, or a roped climbing party.
- In a discretionary task, the group decides for itself how to combine what members contribute, which describes most real project and strategy work.
Ringelmann’s rope is additive. Most knowledge work is conjunctive or discretionary, which means the basic losses compound: weak links, waiting time, and the cost of agreeing how to combine the work all add on top of them. The programme in Chapter 1 cannot go live until every workstream signs off, so it moves at the speed of its slowest handover.
7.2 How Work Passes Between People
James Thompson (1967) described three ways in which work can depend on other work. In pooled interdependence, people work independently and combine their outputs; in sequential interdependence, one person’s output becomes the next person’s input. In reciprocal interdependence, work flows back and forth in iterative loops, as it does in most design, analysis and advisory work.
Van de Ven, Delbecq and Koenig (1976) added a fourth, team interdependence, in which the whole unit works on the task together at the same time. They showed that as interdependence rises, organisations shift from impersonal coordination, through rules and plans, to personal and group coordination, through meetings and mutual adjustment, and that shift is exactly the interaction volume described in Chapter 6.
7.3 How Complexity Raises the Bill
Robert Wood (1986) described task complexity in three parts: the number of distinct acts and pieces of information involved, the timing and sequencing relationships between them, and how much both change while the work is under way. Knowledge work tends to be high on all three, and Jay Galbraith (1974) drew the organisational consequence. As complexity and interdependence rise, an organisation must either increase its capacity to process information, with more meetings, more managers and more channels, or reduce its need to, by creating slack or self-contained units.
7.4 Interdependence Cuts Both Ways
Interdependence is not good or bad in itself, and the evidence shows why. Wageman (1995) found that it has to fit both the task and the reward system, and in her study of service teams, groups with mixed designs, partly individual and partly collective, performed worst. A later synthesis of the research confirmed that the way work is structured to depend on others shapes both how teams work together and what they achieve (Courtright, Thurgood, Stewart & Pierotti, 2015).
In the field, as Chapter 4 noted, people who saw their work as more interdependent loafed more, and the authors argue that interdependent effort is harder to tell apart, though they did not test that directly (Liden et al., 2004). Reward design can help: combining individual and shared rewards reduced loafing in interdependent teams in one laboratory study (Pearsall, Christian & Ellis, 2010), and in virtual teams, deliberately designed goal, task and reward interdependence predicted effectiveness, largely through members’ sense that their own contribution mattered (Hertel, Konradt & Orlikowski, 2004).
Figure 5 (proposed): a two-by-two of interdependence against complexity, presented as an organising frame rather than a tested model, with the rope in the low-low corner and most knowledge work in the high-high corner.
7.5 Where Part III Leaves Us
In knowledge work, interdependence is a double-edged design choice, necessary for integration but obscuring individual contribution and multiplying interaction. Every person added to interdependent, complex work adds more potential connections than the last, and those connections have to run through somebody. In practice they run through a very small number of people, and that is where the damage is most visible.
PART IV. WHY ARE MY BEST PEOPLE BURNING OUT?
Chapter 8: The Architect and the Commercial Lead
8.1 Forty Unread Requests
Go back to the two people on the programme call with forty unread requests between them. They are not badly organised. They are what network researchers call central: they have the most direct ties, they sit on the shortest path between other people, and they can reach everyone else quickly (Freeman, 1978). In knowledge work, centrality is what happens to people who are both able and willing to help, and it is a trap.
Cross and colleagues found that, in most of the organisations they studied, 20 to 35 per cent of value-added collaboration came from only 3 to 5 per cent of employees (Cross et al., 2016). Those people are drawn into ever more projects, decisions and requests until they become bottlenecks. In the authors’ data, they ended up with the lowest engagement and career satisfaction of anyone (Cross et al., 2016; Cross, Taylor & Zehner, 2018). Collaboration overload falls hardest on high performers, who can be worked to exhaustion for marginal gains to the organisation (Cross & Gray, 2013).
8.2 Why Hubs Fail on Complex Work
The cost is not only to the people at the centre. Experiments on communication networks in the 1950s found that centralised networks, with every message routed through one hub, solved simple problems quickly and efficiently (Leavitt, 1951), and the work that followed showed that the advantage reversed as problems became complex: decentralised networks did better, because the hub became saturated with more information than one person could process (Shaw, 1964).
Field evidence points to the same conclusion. The more a work group’s advice network was centralised around a few people, the worse the group performed (Sparrowe, Liden, Wayne & Kraimer, 2001). Distance makes it worse: when people work apart, collaboration contracts inside existing groups and fewer new links form across them (Yang et al., 2022), which can make the people who already bridge groups even more central.
8.3 The Quiet Cost at the Edges
There is a second, less visible cost. When decisions route through a hub, the people on the periphery may come to see their own judgement as dispensable, which is precisely the condition that switches on social loafing in Chapter 4. That link has not been tested directly, but it would mean the two losses feed each other: overloaded hubs produce coordination loss, and the people waiting on them are at risk of motivation loss.
Figure 6 (proposed): a hub-and-spoke team next to a distributed team, with the overloaded central person highlighted.
8.4 Who Becomes a Hub
Hubs are not chosen at random. The people colleagues approach first are the able, the agreeable and the outgoing, and they are also the people least likely to refuse. Boundary spanning, the work of reaching outside the team for information and support, has been linked with role overload for sixty years, because the people who do it face simultaneous and often conflicting demands from inside and outside the team (Kahn, Wolfe, Quinn, Snoek & Rosenthal, 1964).
That brings the argument to the people themselves. So far, this article has treated team members as interchangeable, as Ringelmann did with the men on his rope. They are not, and the way their personalities combine determines whether a team absorbs every pressure described so far or amplifies it.
PART V. WHY DOES THE SAME MIX THRIVE IN ONE TEAM AND FAIL IN ANOTHER?
Chapter 9: Same Average, Different Team
9.1 Two Workstreams
Inside the programme are two workstreams of six, each put together by a manager who looked at the personality profiles and was pleased to see a solidly conscientious group. In the first, all six people sit slightly above the middle of the scale: organised, dependable, and unremarkable. In the second, three are meticulous to the point of anxiety and three are relaxed about deadlines and detail, so that on paper the two workstreams have the same average conscientiousness.
They will not behave alike. In the second, the meticulous three will notice within a month who is not pulling their weight, and Chapter 4 has already described what people tend to do next. My doctoral research found that the second kind of team is the one in trouble, and that it is in more trouble than a team that is uniformly less conscientious (Keca, 2019). To see why, it helps to know how a team’s personality can be described at all.
9.2 Five Traits, Four Ways to Count Them
The most widely used description of personality, the Big Five, places people on five broad dimensions: neuroticism, a tendency towards anxiety and vulnerability to stress, whose opposite pole is emotional stability; extraversion, sociability, assertiveness and energy; openness to experience, curiosity, imagination and a taste for novelty; agreeableness, cooperativeness, trust and an aversion to conflict; and conscientiousness, organisation, dependability, self-discipline and the drive to achieve (John & Srivastava, 1999).
A team has no personality of its own, so researchers must decide how to turn six or twelve individual scores into one description, and four standard answers describe four quite different things (Barrick, Stewart, Neubert & Mount, 1998; Mohammed & Angell, 2003). The average says how high the team is on a trait overall, the spread how much members differ, the lowest score how far the weakest link falls, and the highest whether the team has at least one person who is strong on it. The two workstreams have the same average and very different spreads, and most of what matters in this part of the article lives in that difference.
9.3 More of the Same, or Something Different
Some traits seem to work best when members are similar and high on them. Conscientiousness is the clearest case: a team that is uniformly conscientious tends to agree on high goals and high effort, a team that is uniformly low tends to agree on low ones, and a mixed team tends to argue about both (Mohammed & Angell, 2003; Keca, 2019). Researchers call these supplementary traits because each person adds more of the same.
Other traits have been expected to work best when members differ, and extraversion is the usual example, on the reasoning that a team needs someone to take the lead and that a team in which everybody wants to lead may spend its energy competing for airtime and influence (Barry & Stewart, 1997; Humphrey, Hollenbeck, Meyer & Ilgen, 2007). These are called complementary traits because one person’s high score fills a gap that others leave. As Chapter 11 shows, my results did not fit that tidy theory.
Chapter 10: Why Personality Profiles Disappoint
10.1 What Twenty-Five Years of Studies Found
The early evidence was encouraging. Teams with higher average conscientiousness, agreeableness, extraversion and emotional stability tended to perform better and to be more viable, meaning more willing and able to keep working together, and a single low scorer on agreeableness or conscientiousness could drag a whole team down (Barrick et al., 1998). Later bad-apple research developed this into a general account of how one persistently negative member can change everyone else's behaviour (Felps, Mitchell & Byington, 2006).
Subsequent syntheses found that personality composition predicts team performance (Peeters, Van Tuijl, Rutte & Reymen, 2006; Bell, 2007), with Bell finding stronger relationships in real workplaces than in laboratory teams, which makes sense if personality needs time and real stakes to show itself. The answer also depends on how traits are counted, how performance is measured and how work flows through the team (Prewett, Walvoord, Stilson, Rossi & Brannick, 2009).
The most recent work is more sobering, and it is worth knowing before spending money on profiles. A 2024 update of the meta-analyses, which pool the results of many studies, found that team members’ average traits relate only weakly to team performance, that the strongest single relationship involved how much members differed in neuroticism rather than any average, and that the relationships shift with how long the team has been together and what kind of performance is measured (Han et al., 2024). Across 24,425 teams, teams whose members differed more in personality, values and culture had more conflict and fewer of the shared positive states that help teams work (Triana, Kim, Byun, Delgado & Arthur, 2021), and a large registered review found that diversity of every kind had an average relationship with performance close to zero, with the evidence on personality diversity inconclusive either way (Wallrich, Opara, Wesołowska, Barnoth & Yousefi, 2024).
The lesson is not that personality is irrelevant to teams. It is that the average tells you little, that differences between members show up first in a team’s climate and conflict, and that everything depends on the situation, which is exactly where most profiling stops looking.
10.2 Traits Need a Trigger
Traits are tendencies, not behaviours, and they show up only when a situation offers the right cues, a concept called trait activation (Tett & Burnett, 2003). An agreeable person’s helpfulness shows when someone needs help, an anxious person’s worry shows under pressure or conflicting demands, and an extravert does not behave like one alone in an empty room. A review of the research that followed found 262 tests of the idea across 75 studies, about three in five of which held up, and named its application in teams as one of the field’s most pressing gaps (Tett, Toich & Ozkum, 2021).
The gap matters because, in a team, other members play a large role. A team high in a trait creates norms and cues that draw that trait out in everyone, while a mixed team sends weaker and more contradictory signals (Keca, 2019). Teammates even change how much an individual’s own personality pays off: the link between a person’s conscientiousness and their performance was stronger when the team as a whole was more conscientious, and weaker when members differed widely in emotional stability (Prewett, Brown, Goswami & Christiansen, 2018). The same trait in the same person paid off differently depending on who sat beside them.
10.3 Teams Drift Towards Sameness
People are drawn to others who are similar to them (Byrne, 1971), and over time organisations and teams attract, select and keep people who fit, and lose those who do not, so that they become steadily more alike (Schneider, 1987). An extraverted team tends to recruit extraverts, and a cautious team tends to recruit cautious people, whether or not that suits the work.
That drift is comfortable, and it can be costly. A team whose members are similar usually feels more at ease and may be more satisfied, and it is easy to mistake that satisfaction for effectiveness.
10.4 Too Much of a Good Thing
Most hiring assumes that more is better: more conscientiousness, more agreeableness, more emotional stability. Beyond a point, it is not. In a large study, the benefits of conscientiousness and emotional stability for job performance levelled off and then disappeared at high levels, especially in complex jobs (Le et al., 2011), and Grant and Schwartz (2011) argued that this rise-then-fall pattern is the rule, not the exception, in psychology.
Teams show the same shape. When students rated how much each teammate had contributed to a creative task, contributions rose with extraversion, agreeableness and conscientiousness, then peaked and declined at the highest levels (Curșeu, Ilies, Vîrgă, Maricuțoiu & Sava, 2019). Very high conscientiousness can turn into rigidity and perfectionism, very high agreeableness into conflict avoidance, and very high extraversion into talking over people.
Some situations also overrule personality altogether. Where demands are clear and shared, almost everyone behaves alike, and individual differences fade into the background, which is one reason a team with strong norms can act more consistently than its members would alone (Keca, 2019). What a team brings therefore depends on how much of a trait it has, how evenly that trait is spread, and what situation it finds itself in, and my doctoral study set out to test all three at once.
Chapter 11: Inside 48 Teams
11.1 The Question I Asked
I wanted to know how, when and under what conditions team personality affects what teams achieve, and how that changes with the work a team does at its edges (Keca, 2019). I chose boundary management because knowledge workers cannot avoid it, and because it places heavy demands on the social behaviour that personality governs. It has three parts (Ancona & Caldwell, 1992; Faraj & Yan, 2009).
- Boundary spanning is reaching out to secure information, resources, and support from other teams and stakeholders, which, in the programme, is the commercial lead’s whole job.
- Boundary buffering protects the team from outside disruption, pressure, and distraction, meaning saying no on the team’s behalf.
- Boundary reinforcement strengthens the team’s internal identity and shared sense of purpose.
Each of these is a situation in which traits are switched on: spanning calls for persuasion and help, buffering for firmness, and reinforcement for warmth and patience.
11.2 The Teams
The study followed 48 teams, 325 people in all, from 12 UK organisations across manufacturing, services and the not-for-profit sector. Nearly all (96 per cent) were knowledge workers; about three-quarters worked in distributed or virtual arrangements; and teams ranged from three to fifteen members, averaging a little over six (Keca, 2019). Members completed a standard 44-item Big Five questionnaire (John & Srivastava, 1999) and scales on boundary management adapted from Faraj and Yan (2009), and about three months later team leaders rated each team’s performance, cohesion and viability.
The analysis examined each trait’s average and spread, patterns that rise and then fall, and whether each boundary activity, and the interdependence of the team’s work, changed the picture, with team size and collaboration with other teams held constant.
11.3 The Headline
Team personality mattered, but rarely in a straight line, and almost always depending on the situation. More was often not better; differences between members were often worse than a low level, and in several cases a combination that helped a team became harmful once its work became highly interdependent.
Chapter 12: Five Findings, Trait by Trait
12.1 Uneven Conscientiousness Does More Damage Than Low Conscientiousness
Conscientiousness is usually the most dependable personality predictor of individual and team performance (Barrick et al., 1998; Bell, 2007), and my results confirmed its importance, with a twist. Performance was highest when the team’s conscientiousness was high and evenly spread, but performance was worse when conscientiousness was low and varied across members than when it was uniformly low (Keca, 2019). A mix of conscientious and less conscientious members, the second workstream in Chapter 9, was more damaging than a team that was uniformly less conscientious.
The explanation I find most convincing goes straight back to the rope. When conscientious people work alongside colleagues who visibly put in less, they may withdraw effort themselves to restore a sense of fairness, which is the sucker effect (Kerr, 1983; Hart, Bridgett & Karau, 2001). If so, uneven conscientiousness is the personality signature of the loafing problem, which my study did not test directly but which fits every result.
Boundary work changed the picture in three ways. Teams that buffered their boundaries heavily performed better as conscientiousness rose, but only when their work was not highly interdependent; when interdependence was also high, the same combination was linked to lower performance, as if a very thorough team guarding its edges struggles when the work demands constant give-and-take (Keca, 2019). High conscientiousness with heavy boundary reinforcement also linked to lower performance, while teams that spanned their boundaries more grew more cohesive and more viable as conscientiousness rose, perhaps because securing resources for the team rewards the drive conscientious people bring. For highly interdependent teams, moderate and even conscientiousness may be a better target than the highest scores available.
12.2 Agreeable Teams Can Be Helped to Death
Agreeable teams generally cooperate well and handle conflict constructively (Barrick et al., 1998). Again, spread mattered: performance and cohesion rose with agreeableness when members were similar and fell when they differed widely (Keca, 2019), which fits the evidence that one persistently disagreeable member can change how a whole group behaves (Felps et al., 2006).
Interdependence and boundary work added richness to the picture. When agreeableness, boundary spanning and interdependence were all high, cohesion rose, but when interdependence was high and spanning was low, cohesion fell as agreeableness rose (Keca, 2019). My reading is that agreeableness creates a desire to help, interdependence creates a demand for help, and boundary spanning provides the channel that connects the two; without the channel, the desire has nowhere to go.
There is a darker side. Highly agreeable teams are natural targets for other teams’ requests, and their members respond generously, often at the cost of their own team’s work, which creates a dilemma between being a good organisational citizen and delivering the team’s own goals (Keca, 2019). It is the team-level version of the overload that fell on the architect and the commercial lead in Chapter 8. I call it the altruism trap.
12.3 Stress Travels Along the Boundary
Emotional stability gives a team a calm climate in which cooperation and constructive disagreement can happen. In my data, performance first rose and then fell as average team neuroticism increased; higher neuroticism went with lower cohesion, and cohesion was best when team neuroticism was low (Keca, 2019). When boundary spanning was high, even modest increases in neuroticism reduced cohesion, and higher neuroticism also went with lower performance in teams that buffered or reinforced their boundaries heavily, and with lower viability in teams that spanned heavily.
The strain of working across boundaries seems to build up at team level, and teams with even moderate neuroticism are especially exposed to it. Moods spread between group members (Barsade, 2002), which helps explain why a few anxious or irritable people can shift a whole team’s climate. The more boundary spanning a team’s work involves, and the heavier its interaction load, the more emotional stability matters.
12.4 The Loudest Teams Are Not the Best, and Mixed Ones Struggle
The conventional view is that extraversion helps teams because it drives communication and energy, and that a mix of extraverts and introverts is best because it avoids competition for airtime (Barry & Stewart, 1997). My results challenged both halves of that view. Team extraversion did not predict performance, although it shaped cohesion and viability, and not in a straight line (Keca, 2019).
When members differed widely in extraversion, higher team extraversion was associated with lower viability, while among teams whose members were similar, higher extraversion was associated with higher viability. With heavy boundary spanning, cohesion and viability peaked at moderate extraversion; with heavy boundary reinforcement, the decline at very high extraversion was even steeper, as if very extraverted teams with lots of outside contact lost focus or competed for influence. Viability was best, though, when a team was highly extraverted and strongly buffered its boundaries (Keca, 2019).
So sameness at the high end of extraversion helped teams stay together, contrary to the view that extraversion should be mixed. That squares with later evidence that the assertive side of extraversion and its warm, sociable side do different things at work (Wilmot, Wanberg, Kammeyer-Mueller & Ones, 2019), and with evidence that extraverts’ status in task groups can fade over time while their more anxious colleagues’ rises (Bendersky & Shah, 2013). My questionnaire could not separate those sides of the trait, so that explanation is more likely than proven.
12.5 Curiosity Needs Room
Openness to experience had no direct relationship with any team outcome, consistent with earlier research, and mattered only in combination (Keca, 2019; Prewett et al., 2009). When boundary buffering was high, more open teams performed worse, and openness went with better performance only when both buffering and interdependence were low (Keca, 2019). A very open team that spends its time defending its edges may chase novelty at the expense of delivery, and the newest synthesis suggests that differences in openness matter more when performance means creativity (Han et al., 2024).
Chapter 13: How Personality Feeds the Losses
13.1 Too Much, Too Little, Too Different
Four lessons stand out, and each ties the people back to the problems in Parts II to IV. The first is that the question is not whether a team has enough of a desirable trait, because level, spread and situation interact, and the best mix is often moderate and even rather than maximal. In my study, too much (neuroticism and conscientiousness), too little (conscientiousness, agreeableness and extraversion) and too different (conscientiousness and agreeableness) could each harm a team (Keca, 2019; Grant & Schwartz, 2011; Le et al., 2011).
13.2 Interdependence Can Reverse the Result
Second, interdependence can turn a helpful combination into a harmful one. In several cases, a mix of personality and boundary work that helped a team became harmful once its work was highly interdependent (Keca, 2019), which is Chapter 7’s lesson from another angle: how much people depend on one another is a design choice with strong psychological side effects.
13.3 Personality Shapes Both Kinds of Loss
The third is that personality is tied to both of Steiner’s losses. Uneven conscientiousness feeds motivation loss through fairness, the sucker effect, and withdrawal, while neuroticism, extraversion, and agreeableness shape how a team copes with coordination through strain under interaction load, competition for airtime, and the pull of other teams’ requests.
Personality probably shapes who becomes a hub as well, since the agreeable, the able and the outgoing are the people colleagues approach first and the ones least likely to refuse. No study has yet followed that chain from trait to hub to burnout, but if it holds, an organisation that rewards helpfulness without managing its cost may be selecting its most cooperative people for exhaustion.
13.4 Composition Is Too Important to Leave to Chance
The fourth follows from the drift towards sameness. Left to itself, a team will tend to become more alike whether or not that suits the work (Schneider, 1987), and skill alone does not guarantee team performance. The combination of people deserves at least as much thought as the people themselves (Keca, 2019).
Chapter 14: The People You Will Mislabel
14.1 The Colleague Who Seems to Be Coasting
Every finding in this article describes what tends to happen across many people and many teams, and none of them tells you what is true of the person sitting opposite you. Start with the colleague who seems to be coasting. The blindfolded pullers, the field groups and the free-rider research all point the same way: loafing is mostly a response to a situation in which effort is invisible or seems redundant, so the colleague who appears to be coasting may be someone whose contribution the design has made dispensable, and the fix is usually to the design.
14.2 The Difficult Colleague
Bad-apple research can easily become a licence to remove anyone who disagrees. It concerns persistent negative behaviour, such as withholding effort, hostility and undermining others (Felps et al., 2006). A colleague who challenges a weak plan, refuses an unreasonable request or says what others are thinking is not that person, and in a team with low psychological safety they may be the only one keeping it honest.
14.3 The Very Conscientious Colleague
The finding that very high conscientiousness can hurt in interdependent work is not a reason to distrust diligent people. It describes a team’s thoroughness meeting work that needs flexibility, and the same people are often a team’s greatest asset on work that rewards precision.
14.4 The Quiet Colleague
The finding that high, even extraversion helped teams stay together, and the wider evidence that extraversion carries advantages at work (Wilmot et al., 2019), can be misread as saying that introverts are bad for teams. Team extraversion did not predict performance in my study, and the moderate levels that suited heavy boundary spanning suggest plenty of room for quieter members. The colleague who says little in meetings may be the one doing the focused work that interaction overload destroys first.
14.5 The Team You Built
The hardest mislabelling is of your own team. Similar people are comfortable together, so a team that feels harmonious may simply have drifted into sameness, and its satisfaction may say little about its effectiveness. That is not a reason to engineer discomfort. It is a reason to ask, from time to time, whether the mix suits the work rather than the people in it.
14.6 How Far My Findings Travel
These results come from one doctoral study of 48 teams, examined and approved but not yet published in a journal, with about three-quarters of the data from one large multi-site organisation (Keca, 2019). That is a respectable number of teams for research of this kind, but a small one for the number of combinations tested, and several patterns emerged from analyses that were not planned in advance. The short questionnaire also couldn't separate facets such as altruism within agreeableness or assertiveness within extraversion, where several of my explanations sit. Treat the trait findings as strong leads for how to look at a team, not as rules for building one.
PART VI. WHAT CAN I CHANGE ON MONDAY?
Chapter 15: Nine Levers
15.1 Design, Not Exhortation
If the problem is structural, so are the solutions. The levers below start with the ones the evidence supports most strongly and end with the ones that are sensible but untested; each shows how far the evidence goes, because research on what fixes teams is much thinner than research on what breaks them.
15.2 Make Each Person’s Pull Visible
Give each output a visible owner, make individual contributions possible to see and evaluate, and give people roles that nobody else covers. This directly addresses motivation loss by restoring the link between my effort and the result. Loafing falls when contributions can be identified (Williams et al., 1981), when they are not redundant (Kerr & Bruun, 1983), and when people believe their effort matters to an outcome they value (Karau & Williams, 1993), and the field evidence points the same way (Liden et al., 2004). This is the best-supported lever in the article, though most of the support comes from laboratories.
Visibility can tip into surveillance, which defeats the point. The aim is that people can see their own pull and know that it counts, not that every keystroke is counted.
15.3 Size the Team to the Work, Not to the Backlog
Keep core teams small and split them before adding people, and treat rules of thumb such as the two-pizza team as useful heuristics rather than findings. The costs of size are well established: small teams do better on coordination-heavy work (Hackman & Vidmar, 1970; Hoegl, 2005; Mueller, 2012), and larger teams show more deviance and withdrawal (Bernerth et al., 2023). But the same synthesis found that complex work can justify larger teams, because it needs more expertise, and that the average link between size and performance was zero. That is why the lever is to size the team to the work rather than to shrink it by default.
For the programme director, the practical question before asking for three more people is whether the work could instead be split into smaller teams with self-contained goals, which would add capacity without adding the forty-five new relationships between pairs that three more people on one team would bring.
15.4 Match Interdependence and Rewards to the Task
Remove interdependence where the work does not need it, and where it does, reward both individual and shared contribution. Interdependence that does not fit the task or the reward system costs performance (Wageman, 1995); structural interdependence shapes how teams work and what they achieve (Courtright et al., 2015); and hybrid rewards reduced loafing in interdependent teams in the laboratory (Pearsall et al., 2010). The catch is that Wageman’s mixed designs performed worse, so a hybrid must be designed with care rather than assembled by adding a team bonus to individual targets.
15.5 Take the Load Off the Hubs
Map who is carrying the collaboration load, redistribute requests, and turn knowledge held by one person into shared resources. The problem is well documented (Cross et al., 2016; Cross & Gray, 2013), centralised advice networks go with weaker group performance (Sparrowe et al., 2001), and centralised communication fails on complex problems (Leavitt, 1951; Shaw, 1964). However, the specific remedies are practitioners’ recommendations, and I have found no trial testing whether redistributing requests improves a team’s performance. Remote and hybrid working make the lever more urgent, because collaboration across groups contracts when people work apart (Yang et al., 2022).
15.6 Protect Focused Work
Cut unnecessary synchronous coordination, batch interruptions, and protect blocks of uninterrupted time. The mechanism is well understood: collaborative demands crowd out individual work (Perlow, 1999), and switching between tasks leaves residue that lowers performance on the next one (Leroy, 2009).
Perlow tried the remedy with one software team of seventeen engineers. In the first phase of protected quiet time, 59 per cent rated their own productivity above average; in a phase organised around fixed hours for interaction the figure was 41 per cent, and when quiet time returned it rose to 65 per cent, though the change largely lapsed once the study ended (Perlow, 1999). It is a promising result from one team, and nobody has yet tested it at scale.
15.7 Compose Deliberately
Aim for high and even conscientiousness and agreeableness, watch for teams drifting into sameness on extraversion, ensure a highly extraverted team with heavy outside contact stays focused, and avoid putting stress-prone people into heavy boundary-spanning roles without support. When forming several teams at once, a leader can use what Humphrey and colleagues (2007) call seeding, placing particular members across teams to increase or reduce how much they differ on a trait.
The value of high average conscientiousness and agreeableness, and the damage one very low scorer can do, rest on several studies and syntheses (Barrick et al., 1998; Bell, 2007; Prewett et al., 2009), though the newest synthesis finds those relationships weaker than earlier ones did (Han et al., 2024). The finding that uneven conscientiousness does more harm than low conscientiousness, and the reversals under interdependence, come from my single study (Keca, 2019). Removing people is rarely the answer; it is usually better to place people in sub-groups where their profile complements others and supports their development.
15.8 Develop, Don’t Just Select
Use personality information for self-awareness and team development as well as for selection, because personality is more changeable than was once believed. A systematic review found meaningful trait change through intervention, especially in emotional stability, often over relatively short periods (Roberts et al., 2017). In a randomised trial of a three-month smartphone coaching programme, people who wanted to become more conscientious or more extraverted changed in ways their friends and family also noticed, and the change lasted at least three months after the programme ended (Stieger et al., 2021). Nobody, as far as I know, has yet tested whether developing members’ traits changes how their teams perform.
15.9 Keep People Close to the Consequences
Make accountability explicit and keep people connected to the consequences of their work, especially in dispersed teams. This follows from the evidence that size and dispersion made loafing easier to justify (Alnuaimi et al., 2010) and that moral disengagement predicts misconduct and weaker contribution at work (Ogunfowora et al., 2022). It is a principle derived from the mechanism rather than a tested fix, because I have found no trial of an intervention against moral disengagement in knowledge-work teams.
15.10 Make It Safe to Say You Are Stuck
Make it safe for people to raise problems, admit blockers and ask for help, so that withdrawal becomes visible and can be addressed. Psychological safety predicts learning behaviour in teams (Edmondson, 1999), and a meta-analytic review linked it with information sharing, learning and performance across many studies (Frazier, Fainshmidt, Klinger, Pezeshkan & Vracheva, 2017). Whether it reduces loafing directly has not been tested, but it is the condition under which a team can talk about loafing at all.
15.11 What Not to Do
Avoid fixed numerical thresholds presented as evidence-based, such as no team above eight or no more than 35 per cent of time in meetings, blanket mandates such as cancelling all meetings for a month, and personality ideal profiles applied mechanically to hiring or team allocation. The evidence supports principles and trade-offs, not formulas, and a formula that looks precise is more likely to have been invented than measured.
Do not count on culture to do the work either. Group productivity loss disappeared among participants from a collectivist background working towards group goals in one experiment (Erez & Somech, 1996), but across 21,435 teams, national collectivism did not change the link between team size and performance (Bernerth et al., 2023). Group goals and shared identity reduce loafing wherever a team sits, and no culture is immune.
Chapter 16: Back to the Programme
16.1 Three More People
Go back to the steering group in Chapter 1. The programme director is about to ask for three more people, and almost everything in this article suggests that will slow the programme down rather than speed it up. It adds forty-five potential pairings to the ninety-one the team already has, pulls the two central people further into briefing and onboarding, and adds three more names to outputs that nobody can trace to anyone.
A better first move is to split the programme into smaller teams with outputs each of them owns, take the architect and the commercial lead out of the routing for routine decisions, protect blocks of focused time, and look hard at the mix in each workstream, starting with whether conscientiousness is even. None of that requires anyone to try harder.
16.2 What the Rope Was Telling Us
Ringelmann’s men were not weak, and most of them were not lazy. They were in a situation that made it hard to pull in time with each other, and easy to let the person in front take the strain without anyone knowing.
Knowledge work multiplies both problems. The rope is invisible, and so is each person’s pull on it; work passes back and forth rather than adding up; the number of potential connections rises much faster than headcount; and the load concentrates on a few willing people until they break. The people on the rope aren't interchangeable either, and how their personalities combine decides whether the team absorbs that pressure or amplifies it.
The rope did not go slack because people stopped caring. It went slack because no one could feel their own pull.
16.3 Questions for Monday Morning
- Could this team be split into smaller teams with clearer, self-contained goals?
- Can every person point to an outcome that is visibly theirs?
- Does the work genuinely need the interdependence we have built into it, and do the rewards match it?
- Who are our 3 to 5 per cent, and what are we doing to protect them?
- Is conscientiousness even across the team, and has the team drifted into a mix that suits itself rather than the work?
- Would someone in this team feel safe saying they are struggling, or that a colleague is not pulling their weight?
