Employee performance is one of those concepts that everyone talks about, yet almost no one defines well enough to provide the clarity needed for performance measurement and tracking. I have been helping organizations with performance management for 20 years, and I have encountered this firsthand. Ask 10 managers what it means, and you will get 10 answers that overlap but do not agree. Some will say it is how much work a person produces. Some will say it is whether they hit their targets. Some will describe attitude, personality traits, reliability, or the feeling they get when the person walks into a meeting. All of these answers capture something related to performance but not the complete picture. None of them, on their own, is what the research actually treats as performance.
The way you define employee performance determines what you measure, reward, punish, and develop. If you get the definition wrong, every system built on top of it carries the same flaw. You will end up paying bonuses for no performance improvements, penalizing people for things they cannot control, and mistaking lucky outcomes for true performance.
What follows is a careful rebuild of the concept from the research up. It pulls apart the assumptions most commonly attached to the term, shows what peer reviewed work has established about what employee performance actually is, and explains what changes when you start working from the evidence rather than from inherited intuition.
The Everyday Definition of Employee Performance, and why it fails
In most organizations, employee performance is treated as synonymous with output. Sales closed. Tickets resolved. Deals signed. Reports filed. The underlying thinking seems plausible. If the job is to do a thing, then doing a lot of that thing must be performance.
This logic falls apart the moment you look at it carefully. Two salespeople with identical skills will produce very different revenue numbers if one is assigned a wealthy territory and the other a deprived one. Two customer service agents with identical training will resolve very different numbers of tickets if one happens to work the shift when the product is breaking most often. Two analysts with identical ability will deliver very different reports depending on the quality of the data they are given. Output depends on effort, skill, luck, environment, tools, colleagues, timing, and dozens of other variables. To call output performance is to hold people accountable for all of them.
Decades of credible evidence supported this conclusion. John Campbell, whose foundational work on the structure of job performance remains the reference point for the field, argued that performance must be defined as what the person does, not what results from it. Behavior is under the individual's control. Outcomes are not. If the concept of employee performance is going to be useful for hiring, developing, and evaluating people, it has to focus on the part of the equation that the person actually owns.
The Scientific Definition of Employee Performance
In the research literature, employee performance refers to the behaviors that employees carry out at work that are relevant to the goals of the organization and are under the individual's control. Three things in that sentence do the work.
First, performance is behavior. Not output. Not results. Not scores on a scorecard. The thing being described is what the person does. A detailed synthesis of the construct puts it plainly: performance consists of behaviors that are observable, even when those behaviors are cognitive. Deciding how to handle a customer complaint is performance. Writing a line of code is performance. Choosing to help a colleague who is struggling is performance. Whether the customer ends up satisfied, the code ends up shipping, or the colleague ends up rescued depends on many other factors.
Second, performance is goal relevant. Not every behavior counts. Rearranging a desk, chatting about the weekend, or sitting in a meeting without contributing is not performance. The behavior has to bear on what the organization is trying to achieve. This rules out activity as a proxy. Being busy is not the same as performing. Looking productive is not the same as being productive.
Third, performance is under the individual's control. This is the constraint that protects people from being held accountable for things they did not cause. A salesperson who follows every best practice, works hard, and still misses target because the market collapsed has not performed badly. A customer service agent who resolves every ticket correctly but whose department loses customers because the product was faulty has not performed badly. Judgments of performance have to track what the person actually did, not what happened around them.
Put these three pieces together, and a working definition emerges. Employee performance is the set of observable, goal relevant behaviors that a person carries out at work and that are under their control. Everything else is either output, circumstance, or luck. This distinction changes everything about how you approach performance.
The First Kind of Employee Performance: Task Performance
Once you accept that performance is behavior, the next question is what kinds of behavior count. The research has converged on a layered answer. The most basic layer is task performance, which refers to the behaviors directly involved in doing the technical core of the job. For a nurse, this is clinical care. For a software developer, this is writing and reviewing code. For a teacher, this is planning and delivering lessons. Task performance is what the job description is written around.
Campbell's original model identified eight components that span most jobs: job specific task proficiency, non job specific task proficiency, written and oral communication, demonstrating effort, maintaining personal discipline, facilitating peer and team performance, supervision, and management. The more recent synthesis of this model retains the eight dimensions and concludes that aggregating them above that level loses information. Treating task performance as one thing, in other words, hides how people differ in the behaviors that actually matter.
This has a practical consequence that most organizations ignore. Two employees can be equally strong in overall task performance but very different in the specific behaviors that underlie it. One may be technically brilliant but poor at communication. The other may be technically adequate but exceptional at coaching peers. The single rating they receive on a performance review compresses those differences into a number that no longer tells anyone anything useful. Task performance is multidimensional. Measuring it as if it were a single thing is the first and most common mistake organizations make.
The Second Kind of Employee Performance: Contextual Performance
The second layer of employee performance is contextual performance, also known as organizational citizenship behavior. It covers behaviors that support the social and psychological environment in which the technical work gets done. Helping colleagues. Volunteering for tasks that are not in the job description. Defending the organization when it is criticized. Persisting with enthusiasm when the work is hard. Complying with rules and procedures even when no one is checking.
The meta analytic evidence on feedback and organizational citizenship behavior confirms that contextual performance has its own antecedents, its own dynamics, and its own relationship with organizational outcomes. It is not a softer version of task performance. It is a different thing.
The distinction matters because the two kinds of performance often require different abilities and respond to different management levers. Task performance is predicted most strongly by cognitive ability and job knowledge. Contextual performance is predicted more strongly by personality, particularly conscientiousness, agreeableness, and the parts of the person that govern willingness to cooperate. A hiring process that tests only for task performance will bring in people who can do the job but may contribute nothing to the team. A performance review that measures only task performance will reward technical stars who slowly erode the environment that everyone else depends on.
There is also an uncomfortable finding in this area. Contextual behaviors are voluntary. They are extra. They are not written into the employment contract. This means organizations that rely heavily on them without recognizing or rewarding them are, in effect, extracting free labor. When employees stop contributing these behaviors, the work does not fall below the contractual minimum, but the team starts to degrade. Managers who notice the degradation but cannot point to a specific failure often misdiagnose the problem.
The Third Kind of Employee Performance: Adaptive Performance
For most of the last century, models of employee performance assumed the work was stable. You learned the job, you did the job, and the job stayed roughly the same. That assumption no longer holds in most industries. Technology shifts, business models change, customer expectations move, and the work itself evolves. A definition of employee performance that cannot account for change is incomplete.
The research response to this came from Pulakos and colleagues, whose taxonomy of adaptive performance identified eight distinct behaviors that capture how people handle change at work. These include handling emergencies and crises, managing work stress, creating and solving problems, dealing with uncertain and unpredictable situations, learning new tasks and technologies, demonstrating interpersonal adaptability, demonstrating cultural adaptability, and demonstrating physical adaptability. A sample of more than 1,600 respondents across 24 jobs confirmed that these dimensions hold up as distinct aspects of performance.
Adaptive performance is the behavior that lets an employee stay effective when the conditions around them shift. Two people can be equally strong on task and contextual performance under stable conditions, then diverge sharply when something breaks. One panics, freezes, or clings to the old way of working. The other absorbs the change, figures out what the new conditions require, and keeps delivering. The second person is higher on adaptive performance.
Follow up validation research of this taxonomy, involving more than 700 military personnel, found that cognitive ability, emotional stability, and achievement motivation all predict adaptive performance, but that past experience with adaptive situations adds information beyond those traditional predictors. In other words, some of what makes people adaptable is trait like, but some of it is built through exposure. This matters for how organizations develop people. You cannot train adaptability in a classroom. You can expose people to conditions that require them to exercise it, and then help them reflect on what they did.
The Fourth Kind of Employee Performance: The Counterproductive Side
A complete definition of employee performance has to account for behaviors that actively harm the organization. These are not absences of performance. They are performance in the negative direction. The research literature refers to them as counterproductive work behaviors.
Counterproductive work behaviors include theft, sabotage, deliberate withdrawal of effort, interpersonal aggression, gossip that damages morale, and rule breaking that harms colleagues or the organization. A study of job stressors and counterproductive behavior found that perceived injustice, organizational constraints, and interpersonal conflict all trigger these behaviors, with negative emotion acting as the mechanism that connects stressor to behavior. When people feel mistreated at work, a measurable proportion retaliate through behavior that the organization never fully sees until the cumulative damage shows up.
This dimension is worth naming explicitly because most performance management systems ignore it. A person can be strong on task performance, polite in meetings, and still be quietly engaged in behaviors that cost the organization. A meta analytic review covering dozens of primary studies found that counterproductive work behaviors respond most strongly to how people are treated by supervisors, coworkers, and the organization itself. The lever is relational, not technical. Fix the conditions that produce the frustration, injustice, or conflict, and the counterproductive behavior tends to drop. Try to discipline it out of existence without changing the conditions, and it migrates underground.
Why Employee Performance Is Not the Same as Results
One of the most important distinctions in the research literature is between performance and effectiveness. Performance is behavior. Effectiveness is what happens as a result. Productivity is what the organization gets for the cost of that effectiveness.
This distinction is important to every practitioner who wants to make a difference. Two employees can exhibit identical performance and produce wildly different effectiveness because one is working in better conditions. A manager who interprets the effectiveness gap as a performance gap is about to take a step backward. They will reward the lucky and punish the unlucky, learn nothing about what either person actually did, and corrode trust on both sides. Research on organizational performance as a construct has been warning for decades that most organizational narratives about who performed well and who performed badly are reconstructed after the fact to fit the outcomes, which is a form of storytelling rather than measurement.
This is not to say outcomes do not matter. They matter enormously. But they matter to the organization as a whole, not as a direct measurement of any individual. When assessing employee performance, the behavior within the employee’s control is the important part. The outcome, measured without controlling for everything else that produced it, cannot constitute credible performance.
The Difference Between What Someone Can Do and What They Usually Do
Another layer the research has uncovered is the gap between maximum and typical performance. Maximum performance is what a person can do when they are focused, motivated, and fully engaged. Typical performance is what they do day in and day out under normal conditions. Studies consistently find that typical performance hovers at roughly 70% of maximum capability across most jobs.
This matters because most hiring tools measure maximum performance, while most management challenges concern typical performance. A candidate performs well in a structured interview because they are focused and motivated for the duration of the interview. Whether they will do the same thing every Tuesday afternoon in year three of the job is a different question. A job knowledge test measures what they know. Whether they will apply that knowledge consistently when they are tired, distracted, or frustrated is not what the test tells you.
Organizations that ignore this gap end up surprised by people who looked strong during hiring but underperformed on the job. The employee did not lie. They reverted to their typical level, which is usually lower than the level they display when they know they are being evaluated. The management challenge is not to turn every day into a maximum performance day, which is not sustainable, but to raise typical performance closer to the maximum through conditions, feedback, and meaningful goals.
What Actually Determines Employee Performance
Having defined what employee performance is, the research also identifies what produces it. Campbell's theory points to three direct determinants: declarative knowledge, which is knowing what to do; procedural knowledge and skill, which is knowing how to do it; and motivation, which governs whether the person chooses to expend effort, how much, and for how long. Every other factor that affects performance, including ability, personality, training, experience, and work conditions, operates through these three.
A meta analysis of human resource management practices confirms that employees' experience of the practices that surround them, including training, development, feedback, and support, translates into performance through both resources (skills, empowerment, supportive relationships) and attitudes (job satisfaction, commitment). The practices do not affect performance directly. They affect the determinants that affect performance.
This is useful because it explains why so many performance interventions fail. Sending someone on a training course when the problem is motivation will not fix the performance. Giving someone more autonomy when the problem is lack of skill will make things worse. Changing the incentive scheme when the problem is that people do not know how to do the work is a waste of resources. The determinants have to be diagnosed accurately before the intervention is chosen. Most organizations do the opposite: they pick an intervention first and hope it matches the cause.
Why Most Measurements of Employee Performance Are Wrong
If employee performance is defined as behavior, then measuring it requires observing behavior. Almost no measurement system in common use does this. Most systems measure output because output is easier to count. Most of the rest rely on ratings from supervisors, who often have limited visibility into the actual behavior and whose judgments are systematically affected by biases.
The meta analytic evidence on performance measurement shows how wide this gap has become. A synthesis of studies that included both objective performance records and subjective supervisor ratings for the same employees found that the two correlated only moderately, and not strongly enough to be treated as measures of the same thing. Organizations that rely on one assume it substitutes for the other. The evidence says it does not. Output records capture what happened. Supervisor ratings capture someone's judgment of what happened. Neither measures behavior directly, which, if examined carefully, the definition of performance would require.
This is one reason performance management systems often feel unfair to employees. The employees are being evaluated on a number that doesn't reflect what they did. They know this. They find it demotivating. And the organization, looking at the same number, keeps wondering why the system does not produce the behaviors it was designed to produce. The problem is that the definition of performance used in the system does not match the definition of performance that would actually be useful.
A Working Definition of Employee Performance, Rebuilt from the Evidence
Taking all of this together, here is what the research supports as a definition of employee performance worth using in practice. Employee performance is the set of behaviors that an employee carries out at work that are relevant to the goals of the organization and under the individual's control. It has four layers. Task performance, which is the technical core of the job. Contextual performance, which is the voluntary behavior that supports the social environment in which the work happens. Adaptive performance, which is how the person handles change. And counterproductive performance, which is the behavior that actively harms the organization or its people. Together these four capture what the research treats as the full picture.
This definition has implications you will recognize. It rules out treating output as performance. It rules out treating effectiveness as performance. It rules out treating a single scorecard number as a meaningful measure of anything. It requires distinguishing between what a person can do and what they typically do. It requires measuring behavior, not just results. It requires recognizing that contextual and adaptive behaviors matter even when no one is being paid to produce them. It requires acknowledging that harmful behavior is also performance, and that it responds to the same kinds of conditions that produce the positive kinds.
Most performance management systems in use today fail at least one of these requirements, and usually most of them. This is not a moral failing of the people running those systems. It is what happens when a concept with a careful research definition gets implemented by people working from intuition and convention. The gap between the research definition and the applied version is where almost every complaint about performance management comes from.
What This Means When You Think About Your People
Next time you assess an employee, try this. Separate what they did from what happened. Ask whether the task behaviors were strong. Ask whether the contextual behaviors, the helping, the volunteering, the supporting, showed up. Ask how they handled the moments when something unexpected happened. Ask, honestly, whether any behaviors you saw actively damaged the work or the people around it. That is employee performance. The rest is context and luck.
If the picture you end up with differs from the picture your formal system produces, the problem is almost certainly not the employee. It is the system. Understanding what employee performance actually is, in the sense that the research has established, is the first step to fixing systems that measure the wrong things for the wrong reasons and then treat the output as truth.
Key Takeaways
- Employee performance is behavior, not output. Output depends on effort, skill, and dozens of environmental factors that the employee does not control. Holding people accountable for output treats luck as a key element of performance, yet labels it something else.
- The research treats employee performance as goal relevant behavior under the individual's control. This three part test rules out activity that looks productive but is not, and protects employees from being judged on circumstances.
- Employee performance has four layers worth measuring separately: task performance, contextual performance, adaptive performance, and counterproductive performance. Collapsing them into a single rating hides the information that makes the rating useful.
- Typical performance sits at around 70% of maximum performance. Hiring tools measure maximum. Most management challenges are about typical. Mistaking one for the other produces surprise hires and underwhelming outcomes.
- Performance is determined by what a person knows, what they can do, and how much effort they choose to apply. Every intervention, from training to incentives to autonomy, works by affecting one of these three. Diagnosing the right determinant before choosing the intervention is the part most organizations skip.
- Most measurement systems measure the wrong thing. They track output because it is countable and ratings because the organization needs a number. They rarely measure behavior directly, which is what the definition of performance would require.
Implications for Practice
Agree on a performance as an organization based on this research and develop your measurement system from there. Ask every manager in the organization to answer one question: What is employee performance for this role? If the answers mostly describe outputs, results, or numbers, the definitions are incorrect and unsupported by available evidence on what employee performance is. Rework them around the behaviors that actually constitute the job, and be specific enough that a new employee reading the definition would know what to actually do differently on Monday morning.
Measure performance on all four layers, not just one. Add behavioral indicators for contextual performance, such as whether the person helps colleagues, shares knowledge, and volunteers for work beyond their role. Add indicators for adaptive performance, such as how the person responds when priorities shift or new tools are introduced. Track counterproductive behavior honestly and treat it as a signal of working conditions, not just a disciplinary matter. A system that only measures task performance is telling you about a quarter of the picture and calling it the whole thing.
Separate behavior from outcome in every performance conversation. When reviewing an employee, explicitly distinguish between what they did and what happened. This sounds trivial, and it is surprisingly hard to do in practice. The habit of collapsing the two is so deep that most reviews slide into outcome based judgment within 2 or 3 sentences. Train managers to notice when they are doing this and to return to the behavior.
Stop using single number performance ratings where you can. A 3 out of 5 or a "meets expectations" label compresses task, contextual, adaptive, and counterproductive performance into a summary that no one can act on. At a minimum, rate each of the four layers separately. Better, rate specific behaviors within each layer. The more detailed the rating, the more useful the resulting conversation.
Build your hiring process to sample behavior directly. Job knowledge tests, work samples, and structured interview questions that ask for specific past behaviors all measure closer to what employee performance actually is than unstructured conversations or years of experience. Resumes describe outcomes. Behavioral evidence describes behavior. The shift from one to the other is the single biggest significant change in hiring.
Finally, when performance is disappointing, resist the urge to jump to an intervention. Diagnose first. Is the problem knowledge, skill, or motivation? Is it task behavior, contextual behavior, adaptive behavior, or counterproductive behavior? Is the system measuring what the person actually did, or what happened around them? The answers to these questions determine what will actually help. The wrong diagnosis produces the wrong intervention, and the wrong intervention makes the problem worse while burning through the goodwill required to fix it.
Related Reading on The Human Capital Hub
For a closer look at how performance management systems connect these four layers of performance to employee outcomes, see the analysis of performance management evidence. For practical guidance on conducting evaluations that stay close to behavior rather than drifting into outcome based judgment, read the guide to performance evaluation for managers. For teams building systems that track employee performance over time, a useful starting point is the overview on employee performance tracking. For specific tools commonly used to structure performance management work, see the piece on performance management tools.
