Most organisations measure employee engagement. Far fewer do it well. And almost none of them know what the research says about how strong the link between engagement and performance actually is.
In 2021, the Chartered Institute of Personnel and Development (CIPD), working jointly with the Centre for Evidence Based Management (CEBMa), published the most rigorous review of employee engagement research to date. They screened over 600 published studies and selected only those that met strict quality criteria. Their findings challenge some of the most confident claims made about employee engagement.
This article explains what they found, in plain language, for HR managers and business leaders who want to make better decisions about how they measure and manage engagement.
For context on what engagement means and why it matters, see our article on what employee engagement is.
Why Nobody Agrees on What Employee Engagement Is
Here is the first problem with measuring employee engagement: nobody agrees on what it is.
The CIPD review found more than 50 different definitions in the research literature. The UK government commissioned MacLeod Review found the same thing. Ask ten HR managers to define engagement and you will likely get ten different answers involving words like commitment, motivation, discretionary effort, satisfaction, pride, and belonging.
This is not a minor inconvenience. It is a measurement problem. If your survey is measuring one thing and your board thinks it is measuring something else, your data is telling two different stories simultaneously.
The CIPD and CEBMa review identified four main categories of engagement definition used in research.
Personal role engagement — rooted in the original work of Professor William Kahn, published in the Academy of Management Journal in 1990. Kahn described how people invest their physical, cognitive, and emotional selves in their work when the conditions allow them to. When conditions are right, they show up fully. When they are not, they withdraw.
Work engagement — the most widely used scientific definition, developed by researchers at Utrecht University. It describes a positive, persistent state of mind made up of vigour (energy and mental resilience), dedication (enthusiasm and purpose), and absorption (being fully focused). This is what the Utrecht Work Engagement Scale measures.
Composite engagement — a blend of attitudes and behaviours, closer to what many practitioners understand engagement to mean. The CIPD review is direct about this: most surveys using this definition have not been properly tested to confirm they measure what they claim to measure. Only a small minority of studies using composite measures have been published in peer reviewed journals.
Engagement as management practice — what the organisation does to engage employees rather than how employees feel. Useful for designing interventions, but it is a management activity, not a psychological state you can measure.
The practical lesson: before commissioning your next engagement survey, ask your vendor which of these their instrument is actually based on. If they cannot answer clearly, that is an important signal about the quality of the data you are about to collect.
Which Measurement Tools Does the Evidence Support?
The CIPD and CEBMa review found that the studies providing the strongest evidence of a link between engagement and performance were those using the Utrecht Work Engagement Scale (UWES). This is the most extensively tested engagement instrument in academic research and it is freely available for non commercial use.
The UWES was developed by Professors Schaufeli and Bakker. A large validation study in Educational and Psychological Measurement in 2006 tested it with 14,521 employees across ten countries. It confirmed the three dimensions — vigour, dedication, and absorption — hold up across national and occupational contexts.
A 2019 validation study in the European Journal of Psychological Assessment tested a three item short version of the UWES with 77,465 participants across five countries. The three item version produced results within 0.02 correlation points of the full nine item version. For organisations that measure frequently, this short version is a scientifically sound option.
What the Three Dimensions Tell You Separately
The UWES produces three separate scores, not just one overall number, and each points to a different problem.
Low vigour scores signal an energy problem. Employees are running on empty. They may be completing tasks, but they lack the resilience to absorb extra demands or push through difficulty.
Low dedication scores signal a meaning problem. People have lost the sense of why the work matters. They are completing tasks without any sense of purpose or pride.
Low absorption scores signal a focus problem. Employees are physically present but mentally elsewhere. This often points to job design issues, poor task fit, or persistent interruptions.
An overall engagement score collapses all three into one number and destroys that diagnostic information. Reporting each dimension separately tells managers what to actually do.
What About the Gallup Q12?
The CIPD review addresses the Gallup Q12 directly and its conclusions are worth knowing. The review found that the Q12 correlates at r = 0.91 with a simple job satisfaction measure. In plain terms, the Q12 and a straightforward satisfaction survey are measuring almost the same thing.
For this reason, Bailey and colleagues excluded Q12 studies from their systematic review of 214 engagement studies, describing the scale as a composite measure lacking construct validity. The CIPD review agrees: the Q12 is widely used in organisations but does not meet the scientific standards required to claim it specifically measures engagement rather than a broader mix of workplace attitudes.
This does not mean Q12 data is worthless to organisations that use it. It does mean the data should be interpreted as a broad indicator of workplace conditions and job satisfaction rather than as a precise measure of engagement.
Related: Employee Engagement: Everything You Need to Know
The Honest Finding: Engagement and Performance Are Related, But the Effect Is Small
This is where the CIPD and CEBMa review delivers findings that most engagement vendors do not tell you.
The review found consistent evidence that engagement is positively associated with performance. Across multiple systematic reviews and longitudinal studies, higher engagement goes with better performance. That part is clear. But the honest answer to how strong that relationship is requires looking at the effect sizes, not just the statistical significance.
What the Systematic Reviews Found
Bailey, Madden, Alfes, and Fletcher published the most comprehensive synthesis of engagement research in the International Journal of Management Reviews in 2017. They reviewed 214 studies covering outcomes of engagement. Of the 42 studies examining performance, all found a positive association between engagement and performance at the individual level. But the review did not quantify one overall effect size, because the studies varied so much in design and measurement.
Christian, Garza, and Slaughter conducted a meta analysis of over 200 studies in Personnel Psychology in 2011. They found engagement correlated with task performance at r = 0.43 and with extra role behaviour at r = 0.35. These are meaningful associations. However, the CIPD review notes that this study draws primarily on cross sectional data, meaning both engagement and performance were measured at the same point in time. That shows a relationship. It does not prove which one causes the other.
Motyka reviewed 71 studies in the International Journal of Management and Economics in 2018. Of those studies, 48 found a statistically significant association between engagement and task performance. However, a statistically significant finding does not automatically mean a large or practically important effect. With a large enough sample, even a very small relationship will appear statistically significant.
What the Longitudinal Studies Found
The most important part of the CIPD and CEBMa review is its assessment of the longitudinal studies, the ones that measured engagement at one point in time and performance later. This is the design that can tell you whether engagement actually predicts performance, rather than just being associated with it.
The review found 23 relevant longitudinal studies. The conclusion the CIPD draws is direct: throughout these studies, reported effect sizes are consistently small. In Cohen's standard framework for judging effect sizes, a small effect means the relationship exists and is real, but engagement alone explains only a modest portion of the variation in performance. Many other factors — skills, job design, resources, leadership, and culture, also drive performance and likely explain more of the variation.
One study found engagement at time one predicted performance at time two with a small effect. Another found the relationship, but also found that good performance at time one predicted higher engagement at time two, suggesting the relationship may actually run in both directions. A third found that success predicted subsequent engagement more strongly than engagement predicted subsequent success.
The CIPD review states this clearly: the evidence that engagement predicts performance is weaker than the evidence that engagement is associated with performance. Most cross sectional studies show a relationship. The longitudinal evidence is more modest.
What This Means in Plain Language
Imagine ranking 100 employees by their engagement score and separately ranking them by their performance rating. You would see a tendency for higher ranked employees on engagement to also rank higher on performance. But the overlap is not large. An employee with a high engagement score is more likely to be a strong performer, but engagement alone does not determine performance.
The CIPD review's honest conclusion is that engagement is worth measuring and managing, but it is not a performance lever that works in isolation. It works as part of a broader system of good management practice.
Related: How To Design An Employee Engagement Survey
Three Reasons Your Engagement Data May Be Less Reliable Than You Think
Even when the relationship between engagement and performance is real, measurement problems can make it harder to see and easier to misread. The CIPD and CEBMa review identifies three issues that affect most organisational engagement surveys.
People Tend to Agree With Positive Statements
Most engagement surveys are written with positively worded statements — I feel energised by my work, I am proud to work here. People tend to agree with positive statements regardless of how they truly feel, simply because agreeing requires less effort and feels less socially awkward.
The result is that survey scores for positive items are systematically inflated. The more items phrased in the same positive direction, the more inflated the overall score will be. Relative differences between teams still carry information, but the absolute score should not be read at face value.
Employees May Not Trust Anonymity
Social desirability affects engagement scores, especially where anonymity is not credible. In a team of six people, collecting job level, tenure, and department data alongside survey responses makes it easy to identify who said what. Under those conditions, honest answers to negative questions become professionally risky.
The fix is straightforward: do not collect demographic information you do not genuinely need for analysis. In small teams, consider whether reporting any breakdown is appropriate. Set a minimum group size below which individual breakdowns are not reported.
Measuring Everything in One Survey Inflates Relationships
Most engagement surveys ask employees about their engagement and also about the factors believed to drive it — supervisor support, role clarity, development opportunities — all in the same questionnaire at the same time.
Researchers call this common method variance. When the same person responds to both cause and effect questions in a single sitting, the correlations between them are inflated by the shared context alone, not because of any real relationship. A review by Podsakoff and colleagues in the Journal of Applied Psychology found this can artificially inflate correlations by 0.20 to 0.30. For a relationship that is already small, an inflation of that size makes it appear much stronger than it is.
A practical response: measure engagement and its drivers in separate surveys, or at different times, to get cleaner data.
What the Evidence Says Drives Engagement and Performance Together
If the direct relationship between engagement and performance is modest, the more useful question for practitioners is: what conditions produce both engaged employees and strong performance simultaneously?
The CIPD review points to five categories of factors that the research consistently links to higher engagement.
Individual psychological states — self confidence, resilience, and a belief that one has the resources needed to perform. These personal characteristics shape whether a person is likely to be engaged regardless of what the organisation does.
Job design — autonomy, task variety, feedback, and skill use. Research within the Job Demands and Resources framework, developed by Schaufeli and Bakker and cited extensively in the CIPD review, shows that engagement rises when job resources are present. A longitudinal study in the Journal of Organizational Behavior found engagement is driven almost entirely by resources, not by reducing demands.
Leadership — the quality of the relationship with the direct manager is one of the strongest and most consistently replicated predictors of engagement. Supportive, fair, and clear management raises engagement. Poor management depletes it faster than almost anything else.
Organisational and team factors — psychological safety, trust, fairness of processes, and team climate all predict engagement consistently across the research base.
Organisational interventions — training, development opportunities, and recognition programmes show positive associations with engagement when they are well designed and genuinely valued by employees.
The practical implication: measuring engagement without also measuring these drivers gives you a temperature reading with no diagnosis. You know something is wrong but not where or why.
Engagement Predicts More Than Just Performance
One reason the CIPD still recommends measuring and managing engagement, despite the modest performance effect sizes, is that engagement predicts a range of other outcomes where the evidence is stronger.
The CIPD review and the Bailey et al synthesis of 214 studies both found consistent evidence that higher engagement is associated with lower turnover intentions. People who are engaged are less likely to be looking to leave. This has direct cost implications: replacing an employee typically costs between half and twice their annual salary, depending on the role.
Engagement is also positively associated with wellbeing and negatively associated with burnout. As Schaufeli and Bakker showed in their Journal of Organizational Behavior study, engagement and burnout are driven by different conditions and share only 10 to 25 percent of their variance. They are related but distinct states. Reducing burnout does not automatically raise engagement.
Engaged employees show higher rates of extra role behaviour, doing things that are not in their job description but that benefit colleagues, customers, and the organisation. They are more likely to share knowledge, help new colleagues, take initiative, and suggest improvements.
When you add up lower turnover, better wellbeing, and more discretionary effort, the case for measuring and managing engagement remains strong even if the direct performance effect is smaller than often claimed.
Related: Employee Engagement Metrics You Should Be Tracking
How Often Should You Measure Engagement?
The research on this question reveals a genuine tension.
A seven year longitudinal study found work engagement to be a highly stable state, more stable than job satisfaction or burnout. Macey and Schneider, in their analysis published in Industrial and Organizational Psychology, concluded that engagement is expected to stay relatively constant as long as the organisational conditions remain constant. On this view, annual surveys may be adequate for tracking broad trends.
On the other side, diary studies that measured employees multiple times across a working week found that 80 percent of the variation in engagement scores was within the same person across different days, not between different people. Research by Sonnentag published in the Journal of Applied Psychology found that recovery on the previous evening predicted engagement the very next morning.
The most useful interpretation for practitioners is that engagement has both a slow moving, stable component and a more responsive day to day component. Annual surveys capture the stable part. They completely miss the daily and weekly fluctuations that respond to immediate conditions, workload spikes, how a team meeting went, whether a manager acknowledged someone's effort this week.
Monthly pulse surveys using the three item UWES offer a practical balance: a validated instrument, low survey fatigue, and a signal frequency that lets you detect changes within a quarter.
What Good Engagement Measurement Looks Like in Practice
Drawing the CIPD and CEBMa evidence together, here is what well designed engagement measurement looks like.
Use a Validated Tool
The UWES nine item or three item version are the instruments with the strongest evidence base in scientific research. If your organisation uses a proprietary survey, ask the vendor for published peer reviewed validation studies. Not a marketing claim — actual studies in academic journals showing the instrument measures a distinct and valid construct.
Do Not Use One Overall Score
Report vigour, dedication, and absorption separately. Each dimension points to a different cause and a different management response. Averaging them into a single number destroys the diagnostic value.
Measure the Drivers Alongside Engagement
At a minimum, collect data on autonomy, quality of feedback, manager support, colleague support, and development opportunities. Without this, your engagement score tells you there is a problem, but not where it is or what is causing it.
Measure More Frequently Than Once a Year
Annual surveys miss the variations that matter most for management decisions. Quarterly or monthly pulse surveys with three to five validated items generate far more actionable information for a modest increase in employee time.
Treat the Score as a Diagnostic, Not a Target
The CIPD review is clear: the relationship between engagement and performance is real but small. If managers are rewarded for raising engagement scores, they will raise engagement scores. Whether genuine engagement changes is a separate question. Use the data to understand what is happening and what to change, not as a performance metric in its own right.
Cross Check Survey Data Against Behaviour
Voluntary resignation rates, absenteeism, participation in development programmes, and internal transfer requests all provide a behavioural cross check on survey data. If engagement scores are rising but turnover is also rising, one of those signals is misleading. Investigating the gap is more useful than averaging the two.
The Bottom Line for Practitioners
The CIPD and CEBMa evidence review confirms that employee engagement is a real and measurable psychological state. It is associated with better individual performance, lower turnover, better wellbeing, and more discretionary effort. These are good reasons to measure and manage it.
But the same review shows the performance relationship, while consistent in direction, produces small effect sizes in the most rigorous longitudinal studies. Engagement is one important input into performance, not its primary driver. The organisations that get the most from engagement measurement are those that use it to understand and improve conditions, job design, leadership quality, team climate, development opportunities, rather than those that treat the engagement score as the goal.
Measuring engagement well means using a validated instrument like the UWES, reporting dimensions separately, collecting driver data at the same time, measuring more frequently than once a year, and reading the data honestly. The research does not promise that high engagement scores produce high performance automatically. It shows that the conditions that produce genuine engagement tend to produce better outcomes across a range of things that matter to both employees and organisations.
That is a strong enough case to invest in doing this properly.
