Room to learn and grow, together with people’s own confidence and resilience, sits at or near the top when researchers rank the drivers of employee engagement. A broad meta-analysis of work engagement, published by Greta Mazzetti and colleagues in 2023, found development resources and personal resources more closely linked to engagement than support from co-workers or control over the task.¹ Demands split in two: pressure that stretches people goes with higher engagement, while hassles and conflicting instructions go with lower engagement.²

Almost all of this evidence is correlational, so a ranking tells you where to look first, not which change will work. Read as a practical order, it says: check whether people are learning, strip out the hassles that waste their effort, then measure again several months later. Engagement here means the energy, dedication and absorption people bring to their work, as explained in what employee engagement means and how it is measured.

Why lists of engagement drivers rarely agree

Lists of engagement drivers disagree largely because they rest on different evidence: correlations inside one survey, pooled correlations across many studies, studies that follow people over time, and a few controlled trials. Eean Crawford and colleagues noted in 2010 that the typical engagement study is cross-sectionalcross-sectional: Describes a study that measures everyone in its sample at a single point in time. Because the possible cause and the outcome are recorded together, it can show that two things occur together but not which of them came first.Full entry in the glossary, so even pooled results do not establish cause.²

One broad map is a 2017 systematic review by Catherine Bailey and colleagues, which the CIPD, the UK’s HR professional body, summarized in 2021. It sorted 155 studies of what comes before engagement into five areas: people’s psychological states, job design, leadership and management, organizational and team factors, and organizational interventions.³ Every area had some support, which is why almost any list of drivers can find a study to cite.

Survey driver lists add a problem of their own. When a report calls a question a driver because its answers move with an overall engagement score from the same questionnaire, and both questions ask about nearly the same thing, part of the link is built in. For that reason Crawford and colleagues left Gallup’s measure out of their meta-analysis: its items describe working conditions such as supervisor feedback and co-worker support, so linking job resources to it risked measuring overlap between predictor and outcome.²

An illustrative case: a pulse survey finds that “I have the tools I need” moves closely with “I would recommend this company.” That correlation alone cannot show that new laptops would raise engagement.

The lesson for anyone handed a driver report is three questions. What was correlated with what? Were the same people followed over time? Did anyone change the driver and measure the result?

A ranking of correlations tells an organization where to look, not what will work.

The drivers of employee engagement, ranked in a broad recent meta-analysis

Mazzetti and colleagues’ 2023 meta-analysis pooled studies that all measured engagement with the same scale, the Utrecht Work Engagement Scale. Every kind of resource was linked to higher engagement, but development resources and personal resources were linked most strongly, and social and job resources least.¹

The study

Moderate evidence

Learning and resilience on top: 113 worker samples on one engagement scale

The authors grouped resources into five kinds plus leadership, following the categories of the Energy Compass, a paid survey tool that grew out of co-author Wilmar Schaufeli’s research. Personal resources such as optimism, self-efficacy and resilience (a correlation of .48) and development resources such as learning opportunities and feedback (.45) were linked to engagement more strongly than social resources (.36) and job resources such as job control and task variety (.37). Learning opportunities alone reached .51. Support from co-workers had the weakest link of any single resource (.27). The authors note that correlations cannot show the direction of cause, and the personal-resource results come from far fewer workers than the others.¹

These links range from moderate to fairly strong, and the gap between the top and bottom groups is real but modest: every kind of resource matters somewhat, and some matter more. In practice, that describes a pattern across many people, not a rule for any one person.

The authors suggest the order reflects closeness. Optimism or self-efficacy sits inside the person, close to how energetic and dedicated they feel; support from colleagues or control over a schedule sits further away, in the job around them. Learning opportunities fit the JD-R model, the job demands-resources model widely used in engagement research, which describes resources as the parts of a job that help people cope and also stimulate their learning and development.¹

The middle of the ranking is less settled. Organizational resources and engaging leadership also showed some of the highest average links, but with wide uncertainty: leadership rested on only five samples.¹ And task variety, middling here, was among the strongest links in Crawford’s earlier meta-analysis, so the order within the middle depends on which studies are pooled.²

An illustrative case: two analysts do the same job for the same pay. One has a stretch assignment and a monthly conversation about what she is learning; the other has a friendly team but no new skills in a year. The research links the first situation more closely with energy and dedication.

A standing one-on-one meeting is a natural place to ask that question every month and follow up on the answer.

Further reading

• 

  The Progress Principle: Using Small Wins to Ignite Joy, Engagement, and Creativity at Work

  by Teresa Amabile and Steven Kramer

  From about 12,000 work diary entries, the authors show how progress on meaningful work, support and everyday setbacks shape motivation at work.

As an Amazon Associate WiserHours earns from qualifying purchases.

Pressure can raise engagement; hassles lower it

Demands at work do not all push engagement the same way. In a 2010 meta-analysis of 64 samples by Crawford, Jeffery LePine and Bruce Louis Rich, demands people tend to see as challenges, such as workload, time urgency and responsibility, were linked to higher engagement. Demands they tend to see as hindrances, such as administrative hassles, organizational politics, role conflict and role ambiguity, were linked to lower engagement, while resources were linked to higher engagement throughout. The authors’ explanation is appraisal: challenges trigger positive emotions and active problem-solving, while hindrances trigger negative emotions and more passive coping.²

Picture two kinds of pressure in one week. A product launch with a hard date asks a lot, but the effort goes toward a goal people can see. An expense claim that needs three signatures and comes back twice asks for effort that achieves nothing.

There is a catch. Challenge demands and hindrances alike went with more burnout, and resources overall still showed a stronger link with engagement than challenge demands did.² Stretching work is not free, which is why the demand side belongs next to what causes burnout.

1. Job resources: such as autonomy, feedback, support and chances to develop; linked to higher engagement
2. Challenge demands: such as workload, time urgency and responsibility; linked to higher engagement, less strongly than resources, and also to more burnout
3. Hindrance demands: such as administrative hassles, organizational politics and role conflict; linked to lower engagement

The practical reading: hindrances are the first candidates to remove, because they are linked to lower engagement and more burnout at once. Cutting all demands is a different matter; Mazzetti and colleagues note, citing earlier work, that reducing job demands is not the most effective way to raise engagement, because a suitable level of demand makes a job feel stimulating.¹

Over time, the way work is organized mattered most

Research that tracks the same employees over time points toward the organization of work. In a 2020 meta-analysis of 55 longitudinal studieslongitudinal study: A study that measures the same people more than once over a period of time, so it can see how they change and which thing came first. Following people over time still does not show that the earlier thing caused the later one.Full entry in the glossary, Tino Lesener and colleagues found that job resources at the group, leader and organizational level all predicted later engagement, but organizational-level resources, reflecting how work is organized, designed and managed, contributed much more strongly. The same review found engagement and job resources highly stable over time. Only the review’s abstract was available for this article.⁴

That stability fits the CIPD’s reading that engagement is likely to change over the medium to long term, and that a quick fix is unlikely to work.³

Longitudinal designs show which came first, not proof of cause. Still, they shift attention from single relationships to the system: clear roles, workable processes and decisions made at the right level.

An illustrative case: a team with a warm, well-liked lead still loses energy when priorities change every week and half its work waits on other departments. A friendlier lead cannot fix that; a clearer process might.

So before launching another engagement initiative, list the structural hindrances a team meets every week, such as unclear decision rights, handoffs that stall or approvals that add nothing. Then fix one at the level where it is caused, even if that level sits above the team.

What Gallup’s 70 percent manager estimate does and does not show

Gallup, which runs engagement surveys and advises companies on management, produced the most quoted figure about managers. In 2015, Randall Beck and Jim Harter wrote that managers account for at least 70 percent of the variance in engagement scores across business units, a Gallup estimate from its State of the American Manager report. Their argument is that great managers create the right environment for engagement. The article does not describe the method behind the estimate, and as company research it has not been peer-reviewed.⁵

A 2022 Gallup article, updated in 2024, states the finding more strongly, calling it probably the company’s most profound and saying the variance is “determined solely by the manager”.⁶ A share of variance describes how much of the difference in scores between teams goes with the manager; it does not say how much engagement would change if one manager were replaced. Gallup’s measure also asks about conditions such as supervisor feedback, which managers shape directly.²

Independent evidence gives managers a real but less dominant place. In Lesener’s longitudinal review, leader-level resources contributed less than organizational ones.⁴ And when Caroline Knight and colleagues pooled controlled interventions in 2017, leadership training was one of four program types, none clearly more effective than the others.⁷

An illustrative case: a new manager inherits a team with low scores. Being more encouraging helps little if role conflict and constant approvals remain; clarifying who decides what, and clearing one approval step, changes the conditions the survey asks about.

The practical reading for managers: put your effort into the resources you control, such as feedback, development, clear roles and removing hassles, rather than relying on personality alone.

What to change first, and how to tell if it worked

Start with the team’s own two lists, the learning people value and the hassles that waste their effort, then change one thing together and measure again months later. A ranking of correlations tells an organization where to look, not what will work. In one major review of controlled engagement programs, Knight and colleagues’ 2017 meta-analysis, average gains were small.⁷

That review, described in more detail in the engagement explainer linked above, rated every included study at high risk of biasrisk of bias: A rating of how much room a study's design, conduct and reporting leave for its result to be systematically wrong. It judges the method, not the motives of the researchers, and a high rating means the result may be off, not that it is.Full entry in the glossary. One of its subgroup results is worth using with care: programs run in groups did better than others, though the difference became borderline when a category holding a single study was removed. The authors suggest group sessions give people a chance to talk with colleagues and voice an opinion.⁷ That is a modest argument for fixing drivers with the people affected, not for them.

The table gathers what the reviews in this article found, driver by driver.

Evidence on rewards is thin: in Crawford’s meta-analysis, rewards and recognition had the weakest and least consistent link of nine resources studied.²

Put the whole team’s answers side by side and the two lists appear. An illustrative case: in a team of six, four people name the same sign-off step on client proposals, and two say nothing has taught them anything new this quarter. The sign-off step is the first hindrance to remove, and those two people are the first to get a stretch task with regular feedback.

WiserHours

From driver ranking to a first change

• Read the survey items behind each "driver" before trusting the ranking
• Ask the team which hassles waste their effort most often
• Remove one hindrance before adding any new program
• Give each person one stretch task and regular feedback on it
• Decide changes with the team, in a group session if possible
• Measure again after several months with the same questions

Find this checklist online:
https://wiserhours.com/employee-engagement/employee-engagement-drivers/#checklist-employee-engagement-drivers-action

The bottom line

Learning, feedback that helps people improve, their own confidence and resilience, and a working system with few hassles make up the shortest honest list of engagement drivers. Those links are mostly moderate and correlational, and the evidence on managers is less settled than the popular reading of Gallup’s 70 percent figure suggests. Start by removing one hindrance and building one learning opportunity with the team, then measure again months later.