Understanding employee engagement: what it is, what drives it and how to measure it.

Employee engagement is one of the most used and least defined terms in management. It is quoted as a percentage, a score out of ten and a colour on a dashboard, and the three rarely agree, as the 2026 national figures show. This guide sets out a working definition, the evidence on what actually drives it, and a way of measuring it that leads to action rather than to another slide.

What engagement is, and what it is not

We define engagement as the degree to which people are committed to the organisation and willing to go the extra mile for it. In our own research we have measured it as two things that can be observed: affective commitment, the emotional attachment to the organisation and its goals, following the definition set out by Mowday, Steers and Porter in 1979; and intention to leave, measured through job-searching behaviour and stated intent. The two are related but distinct. Someone can be fond of the organisation and still be looking; someone can be staying and doing the minimum.

Engagement is not satisfaction. Satisfaction is a judgement about conditions: the pay is fair, the office is fine, the hours suit me. Engagement is a disposition towards the organisation: I want it to succeed and I will put something of myself into that. A satisfied workforce can be a comfortable, coasting one. It is also not happiness, and it is not the absence of complaints. In our experience the most engaged people in an organisation are often its most demanding, because they care how it turns out.

What drives it: the fit between the person and the place

Most organisations measure one side of engagement. Recruitment measures the person, through assessment, and ignores whether the organisation will give them what they want. The engagement survey measures the organisation, through satisfaction with its conditions, and ignores that different people want different things from it. Engagement lives in the gap between the two.

We tested this in a study of 7,915 employees across nine organisations in IT, engineering, real estate, public services, charities, a call centre and leisure, most of them in managerial or professional roles. Each person ranked fourteen areas of working life by how much each mattered to their commitment, and separately rated how far the organisation delivered on each. From those two profiles we computed a fit score for every individual: how closely what they experienced matched what they cared about.

The fit score predicted engagement in every one of the nine organisations. Across the whole sample it correlated 0.27 with affective commitment and −0.35 with intention to leave, significant at the 1% level in each organisation and in total. People we classified as potential leavers differed from likely stayers on fit by about half a standard deviation. Most usefully, fit still predicted intention to leave after general satisfaction was taken out of the picture: a partial correlation of −0.18. In plain terms, two people who are equally satisfied with their conditions will differ in how likely they are to leave according to whether the conditions they are satisfied with are the ones they care about.

The practical consequence is that a survey which asks only “how good is it?” is measuring half of the thing. It can tell you which scores are low. It cannot tell you which low scores matter, and so it will consistently send effort to the wrong place.

Engagement is one driver, not the whole picture

An organisation can have engaged people and still underperform, because engagement does not tell you whether people understand the strategy, whether the processes let them do good work, or whether leaders are equipping them to succeed. We measure engagement inside a model of six drivers of organisational performance:

Measuring only engagement tells you how people feel. Measuring all six tells you what to fix, and it turns the survey from an HR metric into something a leadership team can act on. The six drivers check is a twelve-statement version you can try in a few minutes.

How to measure it so that something happens

• Ask three questions of each area, not one. How good is it? How important is it to you? What should we do about it? The first is the standard satisfaction question and is necessary but not sufficient. The second is the one most surveys omit, and the one that distinguishes a low score that is broken from a low score nobody minds. The third replaces the guessing after the survey closes with a list of actions the respondents themselves chose. The engagement priority grid shows how the first two combine into a priority order.
• Keep it short enough to finish. A survey that takes most of a lunch break will be completed by the people with desks and lunch breaks. Ours takes about eight minutes on any device. Length is a response-rate decision, and response rate is a data-quality decision.
• Reach everyone, not just the people with a laptop. Send invitations by text as well as email; put QR codes where shift workers are; run a plainer-language version for people whose jobs are not at a screen, with the same measures underneath so the results compare. In our experience the people hardest to reach are often the ones whose engagement is most at risk.
• Report by group, and respect the small numbers. An organisation-wide score hides the team that is in trouble. Cut the results by department, site, role level and length of service, and do not report any group so small that an individual could be identified. What to promise about anonymity, and how to keep it covers the rules.
• Act on the gap, and say so. The reason survey fatigue exists is that people answer and nothing visible happens. Pick the areas where importance is high and experience is low, choose one change per team from what respondents asked for, do it, and tell people it came from the survey. From survey results to action is a one-page plan for this; the implementation plan covers the whole cycle.

A note on the numbers you will be shown

Because instruments differ, a 10% engagement figure from one source and a 70% satisfaction figure from another can both be true of the same workforce. Before comparing your result with anyone else’s, ask what was asked, of whom, and on what scale. Compare like with like where you can, and say “directional” where you cannot.

This is the method behind WinningFormula: six drivers, three questions per area, delivered to every part of the workforce, reported by team, and reviewed with your HR team and managers so that the results become decisions.