A pub had its best summer in a decade. Takings were up thirty percent, the garden was full every evening, and the manager got a bonus. Was it a good manager? Possibly. It was certainly a good summer. Nothing about the manager, the staff or the beer had changed. The weather had. As the saying goes, a rising tide lifts all boats.
I tell that story to leadership teams because it exposes how loosely we use the word "performance". Then I run a short exercise. I ask each person round the table to write down their own definition of performance, privately, before anyone speaks. I have yet to have everyone write the same thing down or mean the same thing by commonly used terms. Every answer is defensible. Not one of them is complete. And until that moment, everyone in the room had assumed they meant the same thing.
That matters more than it might seem. Performance is the concept we hang everything on: who we hire, who we promote, who we pay, what we restructure, which strategy we back. If the concept is fuzzy, everything built on top of it inherits the fuzz.
So I did something I should have done years ago. I went back through six decades of peer-reviewed research in organisational psychology, management and strategy and asked a simple question: what does the evidence actually say performance is?
Here is what I found. Five things, and one uncomfortable conclusion.
1. Performance is behaviour not outcome
Ask a manager whether Priya is performing well and they'll reach for a result. She hit her number. She delivered the project. Her team's turnover is down.
John Campbell, whose work has shaped this field more than anyone's, argued that this is the wrong place to start. Performance, he said, is behaviour: the things a person does that matter to the organisation. Not the outcome. Priya's number depends on the market, on the leads she was passed, on whether a colleague in operations delivered on time. Her actions are the only part she controls, and therefore the only part she can fairly be held to account for.
Once you look at behaviour rather than results, it turns out not to be one thing but four. A person can be strong on one and weak on another:
- Task performance: doing the core work of the role well
- Contextual performance: covering for a colleague, taking on the job nobody owns, following the process when nobody would notice a shortcut
- Adaptive performance: adjusting when the ground shifts, and starting the shift when you see it coming
- Counterproductive performance: behaviour that damages the organisation, from padding expenses to sabotage
Most performance systems measure task outcomes. That is doubly narrow: it sits on the wrong side of Campbell's line, and it captures one dimension out of four. Such systems routinely miss the contextual and adaptive behaviour that collective results depend on, and they rarely notice counterproductive behaviour until it has become a crisis.
2. The variability of performance
We talk about people as if they have a fixed performance level. Sarah is a strong performer, Tom is average, and a good appraisal will confirm it. The research says otherwise. Track the same person over a year and their performance moves around a great deal. On some days Sarah is exceptional and on others she is ordinary, and the gap between her good weeks and her bad weeks can be as large as the gap between her and Tom. A difficult conversation at 9am degrades the analysis produced at 10am. Mood, energy, distraction and the immediate situation all pull at how much attention a person can bring to the task in front of them.
Two consequences follow. An annual rating is an attempt to summarise twelve months of ups and downs in a single number, and it does so badly, because raters remember what happened recently and what stood out. And the question "how good is this person?" is less useful than it sounds. A better one is: "under what conditions does this person perform at their best, and how often do we create those conditions?"
3. A team is more than the individual performances
Organisations build teams because the work is too big for one person. Then, when they want to know how the team is doing, they add up the individuals. Five strong performers, two average, one weak: the team must be above average. It is the logic of a spreadsheet column, and the research says it is wrong.
What sits between a team's inputs and its results is a set of shared states that no individual owns. Does the team have a common picture of the task and who does what? Does it believe it can succeed together? Do people trust each other enough to say what they think? Amy Edmondson's work on psychological safety showed that the teams who perform best are the ones where people can admit mistakes and raise concerns, regardless of how skilled the individuals are. Anita Woolley and colleagues asked whether a team has an intelligence of its own, found that it does, and found that it barely correlates with the average or highest IQ in the room. It correlates with whether members take roughly equal turns speaking, and how well they read each other.
Put bluntly: you can hire the ten smartest people in the industry and build a poor team.
4. Organisational performance is more than a number
Ask whether a company is performing well and the answer seems obvious: look at the numbers. But which numbers? Revenue is up but margin is down. The share price has doubled while customer complaints have tripled. The firm is profitable this year because it stopped investing in next year.
The most useful way I've found to think about this comes from Venkatraman and Ramanujam. Picture three concentric circles. At the centre, financial performance: profit, growth, returns to shareholders. Around it, operational performance: market share, quality, innovation, productivity, the things that produce the financial results. Around that, organisational effectiveness in the broadest sense: whether the firm is creating value for everyone with a stake in it.
A company can look excellent from the centre and poor from the edge. The people furthest from the work, investors, shareholders and non-executive directors, see almost entirely the inner circle, and the pressure that creates is well documented. A survey of more than 400 financial executives found that most would sacrifice long-term value to hit a quarterly earnings target. Meanwhile, a meta-analysis of 137 studies found that firm size explained around forty percent of the variation in chief executive pay and firm performance explained less than five. The link between reward and performance at the top of most organisations is far weaker than the bonus scheme implies.
5. Context is key
Which brings us back to the pub.
The situation a person, team or firm operates in is not background to be stripped away so the "real" performance can be seen. It is part of the explanation. Three mechanisms are well established.
Demands and resources. Two people of identical ability. One has heavy demands and few resources: a crushing workload, little autonomy, an absent manager. The other has heavy demands and strong resources: the same workload, but feedback, support and room to decide. One looks like a star and one looks like a problem. The difference is the job, not the person.
Emergence. A firm can hire brilliantly and have nothing to show for it, because individual talent only becomes organisational capability when people can coordinate, information flows between them, and they share an understanding of what they are collectively trying to do.
Incentives. Steven Kerr wrote a paper in 1975 called "On the folly of rewarding A, while hoping for B". Universities that want good teaching but promote on research. Sports teams that want team play but pay for individual statistics. Half a century on, the pattern is intact: we define performance narrowly because narrow things are easy to measure, attach rewards to the narrow definition, and are then surprised when people optimise for exactly what we rewarded.
The uncomfortable conclusion
Most organisations have redesigned their performance management process at least once in the last decade. New rating scales, new competency frameworks, new software, ratings abolished, ratings quietly reinstated.
If none of it has felt like a solution, it is not for lack of effort. A hundred years of research has made performance ratings more accurate, more consistent and less biased.
There is almost no evidence that better ratings lead to better organisational performance.
The instrument has improved; the patient hasn't. What does move performance is what happens on the other 364 days: whether a manager sets clear expectations, notices what people do, and says so while it can still be adjusted.
A working definition
Pulling the research together, performance is best understood as:
Goal-relevant behaviour and its consequences, at individual, team and organisational levels, that vary over time and are shaped by the context in which they occur. Every clause carries a practical implication. Behaviour and consequences: measure both, and know which you are looking at. Three levels: do not assume one adds up to the next. Vary over time: a single point estimate is a poor summary. Shaped by context: when performance disappoints, examine the situation before examining the person.
Seven questions for your next meeting
- Individually write down your definition of performance. How many definitions do you have? Which one is actually driving decisions?
- Of task, contextual, adaptive and counterproductive behaviour, which does your organisation measure, which does it reward, and which does it merely hope for?
- Which of your teams outperform the sum of their members, and which underperform it? Would your talent processes tell the difference?
- What behaviour is your incentive system actually producing, and is it the behaviour you would have chosen?
- If financial performance and stakeholder value diverged for three years, which would you notice first, and which would you act on?
- When your board and your investors ask how the organisation is performing, which circle are they asking about, and who in the room is answering for the other two?
Try the exercise
The write-it-down exercise takes twenty minutes and needs nothing more than a room, a team and some index cards. I have yet to see it fail to start an argument worth having.
If you'd like to run it with your own leadership team, or want the full version of this piece with the research references, message me directly or comment below. I'd also like to hear the definition your team lands on. In my experience, no two are the same.
Dom Irvine is the founder of Epiphanies LLP, where he and his fellow partner, Nathan Clements work with chief executives and their top teams on leadership and organisational performance.