We are here to help: 
In 2025, 20% of employees worldwide were engaged at work and 16% were actively disengaged, according to Gallup’s global workplace measurement, with the remaining 64% sitting in the middle as not engaged (global, data year 2025). That single line does two jobs at once. It tells you the majority of the working world is emotionally uncommitted to the job it does every day, and it tells you that the number has been roughly this bad for years. The word “crisis” gets attached to it constantly. Whether that word is earned, and what it should change in your organisation, is a harder question than most articles admit.
This page is written for people who have to answer to a board or an executive committee about a falling engagement score and need to know what is actually true. It is not a list of things to try on Monday. It is an attempt to answer four questions honestly: is the employee engagement crisis real, how big is it, what does it cost, and what does a serious organisational response look like. Every figure below carries its source, its date and the region it covers, because a claim about the workforce is only as good as the survey behind it.
The Employee Engagement Crisis, Defined
The employee engagement crisis is the sustained condition in which the majority of a workforce is emotionally uncommitted to its work, engagement scores stall or decline across successive measurement cycles, and the loss shows up in productivity, turnover and discretionary effort rather than in obvious performance failures. It is a chronic structural problem, not a sudden event, and it persists across industries, regions and economic conditions.
Hold onto the word “chronic”. Most of the confusion in this subject comes from people treating a long, slow, structural decline as if it were a shock that arrived last quarter.
What “Employee Engagement Crisis” Actually Means
The term is used constantly and defined almost nowhere, which is why arguments about it go nowhere. Four related terms sit underneath it, and they are not interchangeable.
Employee engagement. The degree to which an employee is psychologically committed to their work and their organisation, expressed through involvement, enthusiasm and discretionary effort. It is a state of mind, not a satisfaction score, and not a measure of how much someone likes their employer.
Disengagement rate. The share of a measured population that is not engaged. The formula is straightforward: divide the number of employees classified as not engaged or actively disengaged by the total number of employees surveyed, then multiply by 100. If 1,000 people respond and 640 are classified as not engaged and 160 as actively disengaged, the disengagement rate is 80%. The number is only comparable to another number produced by the same instrument, on the same scale, with the same classification thresholds.
Actively disengaged. A distinct category in the standard engagement taxonomy, and the one most often confused with simple indifference. Actively disengaged employees are not merely uninvolved. They are unhappy at work in a way that leaks outward: they voice resentment, undermine the work of engaged colleagues, and act out the frustration their engaged peers are trying to work around. Someone who is not engaged has checked out quietly. Someone who is actively disengaged is doing something about it.
Engagement benchmark. A reference value that lets you judge whether your own score is good or bad, drawn from a defined population such as an industry, a country, a company size band or a global average. Benchmarks are region-bound and industry-bound. A North American manufacturing benchmark tells a European professional-services firm almost nothing, and comparing across instruments is worse than not benchmarking at all.
Engagement decay. The gradual erosion of engagement in a population over successive measurement cycles, as distinct from a single low reading. Decay is the pattern that makes the word “crisis” defensible: a point-in-time low can be a bad quarter, while a multi-year slope is a structural condition. If you only ever look at this year’s number, you cannot tell the difference between the two.
That distinction matters for how you respond. A point-in-time low invites a campaign. Engagement decay invites a change to how the organisation is built.
Is the Engagement Crisis Real? What the Data Shows
Short answer: the label holds, but not for the reason most people assume. It is not that engagement collapsed recently. It is that it has never been high, and the recent direction is down.
Global. Gallup’s global measurement put employee engagement at 20% in 2025, down from 21% in 2024 and 23% in both 2022 and 2023 (global, published April 2026). The same series shows 16% actively disengaged globally in 2025. Read the long series and the picture is worse than the recent dip suggests: global engagement was 12% in 2009 and climbed slowly to 23% by 2022. Even the peak of that climb left more than three-quarters of the world’s employees uncommitted.
United States. Gallup’s US measurement found 31% of employees engaged and 18% actively disengaged in the first half of 2026, drawn from 43,262 responses from US adults employed full or part time, surveyed in February and May 2026 (United States, published July 2026). US engagement reached a recent high of 36% in 2020 before declining to 31% in 2024, where it has stayed.
Managers specifically. The same global series shows manager engagement falling from 27% in 2024 to 22% in 2025, having been 31% in 2022. Managers are the layer most organisations rely on to carry engagement work, and that layer is disengaging faster than the people it manages.
That reading is echoed in a global workforce study drawing on nearly 12,000 executives, HR leaders, investors and employees surveyed between September and October 2025, which reported that only 44% of employees say they are thriving at work, down sharply from 66% in 2024 and lower than during the pandemic (global, published February 2026).
Two honest caveats belong here, and most pages skip them.
First, these instruments do not measure the same thing. “Engaged”, “thriving” and “satisfied” are different constructs with different question sets and different thresholds. A 20% engagement figure and a 44% thriving figure are not contradictory; they are answers to different questions. Do not stack them into a single headline.
Second, falling employee engagement is measured on self-report surveys with all the response bias that implies. The direction of travel across multiple independent instruments is consistent, which is why the crisis framing survives scrutiny. Any single percentage point of movement does not.
What Two Real Programs Show, and What They Do Not
Industry statistics tell you the shape of the problem. They do not tell you what happens when a specific organisation does something structural about it.
Method and limitations. These are outcomes reported from two named customer programs running on MentorCity, not a survey and not a controlled study. There is no control group, no random assignment, and an obvious selection effect: organisations that build a structured program are already organisations willing to act. Two programs is not a dataset. Treat what follows as existence proof that the mechanics work, not as a population estimate.
CPLED, a legal-education nonprofit. After moving its bar-readiness program onto MentorCity, administrative time per intake fell from roughly 700 hours to 50 hours, a reduction of about 92%, and the no-show rate on scheduled sessions fell from 12% to 1.46%. In October 2025 alone the program scheduled 1,941 video calls. The no-show figure is the one that speaks to engagement: attendance is behaviour, not sentiment, and it moved by about 87%.
AO Foundation, a global healthcare organisation. In the AO Access mentorship pilot, 81% of participants maintained their mentoring relationships through completion, 85% rated platform usability good or very good, and 81% reported professional growth. Relationship completion is a demanding measure. Most voluntary workplace programs lose people quietly in month three.
Neither number proves that mentoring cures employee disengagement across a whole organisation. What they do show is that participation and follow-through are movable with structure, and that the movement is measurable in behaviour rather than in sentiment scores. A first-party dataset across a wider set of programs would settle far more than two cases can, and that is a worthwhile next step rather than something to assert now. In the meantime you can look at programs that produced measurable change and judge the mechanics for yourself.
What Employee Disengagement Costs, and How That Figure Is Built
Before any number: almost every published cost-of-disengagement figure is modelled, not measured. Nobody sends an invoice for disengagement. The figures come from a chain of assumptions, and the honest thing is to show the chain.
Cost of disengagement. An estimated financial loss attributed to the gap between the output of an engaged workforce and the output of the workforce an organisation actually has. It is produced by multiplying a population by a disengagement share by an assumed per-person productivity loss, then converting that loss into currency using salary, revenue per employee or gross domestic product as the denominator.
There are two common ways to build it.
Top-down. Take total economic output, estimate the share of that output lost to low engagement using productivity research, and report the result as a currency figure. Gallup’s global estimate takes this route and puts the cost of low engagement at approximately $10 trillion in lost employee productivity, or 9% of global gross domestic product, for the most recent year in its 2026 report (global, published April 2026). The number is enormous and the method is transparent, but it rests entirely on the productivity-loss coefficient chosen. Change that assumption and the headline moves by trillions.
Bottom-up. Take your own headcount, apply your own measured disengagement rate, apply an assumed productivity loss per disengaged person, and multiply by fully loaded salary cost. For 1,000 employees at an average loaded cost of $80,000, a disengagement rate of 60%, and an assumed productivity loss of 15% among that group, you get 1,000 x 0.60 x 0.15 x $80,000, or $7.2 million a year. Every one of those four inputs is an estimate, and the productivity-loss assumption is the one nobody can verify from the outside.
The useful move is not to quote a global figure to your executive committee. It is to build the bottom-up version with your own headcount, your own salary bands and your own measured disengagement rate, then state your assumptions on the same slide as the number. A defensible internal estimate with visible assumptions carries more weight in a budget conversation than a trillion-dollar figure from a survey your CFO did not commission.
Turnover is the one part of this that is genuinely measured rather than modelled. You know what a departure costs you in recruiting fees, vacancy time and lost institutional knowledge, because you pay it. Where employee disengagement precedes resignation, the turnover line is the closest thing to a hard cost the whole subject offers.
Why Employee Engagement Is Falling
The causes worth discussing at an executive level are structural, not situational. Individual workplace irritants matter to the people who live with them, and they are covered properly elsewhere in this cluster. What follows is the macro layer.
The manager layer is thinner and more loaded. Flatter structures and successive cost programs widened spans of control, so more people report to each manager and each manager has less time per person. That is the same layer whose own engagement fell from 27% to 22% globally between 2024 and 2025. An organisation that routes all relationship work through managers has routed it through its most depleted resource.
Distributed work removed the informal relationship layer. Hybrid and remote models did not damage engagement directly. They removed the incidental contact that used to build belonging without anyone planning it. Relationships that once formed by proximity now form only if something creates them on purpose, and most organisations never replaced the mechanism.
Uncertainty about the shape of the work itself. In the global study cited above, employee concern about job loss due to artificial intelligence rose from 28% in 2024 to 40% in 2026, and 53% said they were worried about lacking future-ready skills. People who cannot picture their role in three years invest less in it today. That is a rational response, not a morale problem.
Measurement without consequence. Many organisations have run engagement surveys for a decade and changed very little as a result. Each cycle that produces a report and no visible change teaches employees that answering honestly accomplishes nothing, which depresses both the score and the response rate. The survey becomes part of the engagement crisis it was meant to detect.
None of these are fixed by a campaign, a benefits refresh or a communications push. They are all questions about how the organisation is put together.
Quiet Quitting Is a Symptom, Not the Condition
Quiet quitting. The practice of doing only what a job formally requires and no more, withdrawing discretionary effort while continuing to meet the stated requirements of the role. It is a behavioural expression of employee disengagement, not a synonym for it.
The distinction is practical. Disengagement is the underlying psychological state. Quiet quitting is one visible way that state shows up, and it maps closely to the “not engaged” category rather than to active disengagement. It is not the only expression, and it is not even the most expensive one.
An employee can be deeply disengaged and still overwork. Presenteeism, compulsive availability and volume without judgement are all common in disengaged high performers, particularly where visible effort is how people protect themselves during uncertainty. If you go looking for the engagement crisis by counting people who leave at five, you will miss the disengaged employees who never stop working and are producing nothing that matters.
Two implications follow. Measuring engagement by observable effort is unreliable. And a policy response aimed at quiet quitting, such as tightening attendance or activity monitoring, treats the symptom while making the underlying condition worse.
How Engagement Is Trending
The trend line is more useful than any single reading, and it is the reason engagement decay is the right frame.
Global engagement rose slowly from 12% in 2009 to 23% by 2022, then fell to 21% in 2024 and 20% in 2025. That is a long climb followed by a short, consistent decline. In the United States, engagement peaked at 36% in 2020 and settled at 31%, where it has remained through the first half of 2026. In both series the recent movement is small in absolute terms and consistent in direction.
Benchmarks are worth using and easy to misuse. They are bound to a region, an industry, a company size and a specific instrument. A global average that includes economies with very different labour markets will not tell a mid-sized Canadian employer whether its own score is a problem. The only benchmark that reliably informs a decision is your own organisation measured the same way over time, with an external comparison used for context rather than as a target.
For most organisations, the honest position is this: your absolute score matters less than your slope, your response rate and whether the drivers behind the score are ones you have any structural ability to change.
What a Serious Organisational Response Looks Like
Serious responses to falling employee engagement have a common shape. They change something about how the organisation works rather than adding an activity on top of it. Four structural moves, described at the level a leadership team decides on rather than as a task list.
Measure less often and act more visibly. The response to engagement decay is not a higher survey frequency. It is a shorter distance between measurement and visible change. Pick fewer questions, commit publicly to acting on a small number of findings, and report back on what changed. A survey that produces one visible change earns more honest answers next cycle than a survey that produces a fifty-page report.
Stop routing all relationship work through managers. If belonging, development and day-to-day support all depend on one person whose span of control doubled, the system fails when that person is stretched, and the global manager numbers say they are. Structured mentoring, peer networks and cross-functional pairing distribute that load to people who have capacity and interest. This is the structural argument behind a human first strategy: technology should carry the administration so people can carry the relationships.
There is also a question of which organisations should be building structured programs at all, and the answer is not everyone equally. Organisation size, turnover profile, distribution across sites and the presence of a real development bottleneck all change the calculation, which is worth reading through the lens of which organizations benefit most before committing budget.
Frequently Asked Questions
What percentage of employees are disengaged?
Globally, 20% of employees were engaged in 2025, 64% were not engaged and 16% were actively disengaged, according to Gallup’s global measurement (global, data year 2025). That puts the global disengagement rate at 80% on that instrument. In the United States, Gallup found 31% engaged and 18% actively disengaged in the first half of 2026, from 43,262 responses among US adults employed full or part time (United States, published July 2026). Use the figure that matches your region, and never compare across instruments.
Is the employee engagement crisis new?
No. Global engagement was 12% in 2009 and had only reached 23% by 2022, so a large uncommitted majority has been the normal condition for as long as this has been measured at scale. What changed recently is direction: global engagement fell to 21% in 2024 and 20% in 2025, and manager engagement fell from 27% to 22% over the same period. The chronic part is decades old. The decline, and the sharper decline among managers, is recent.
Is quiet quitting the same as disengagement?
No. Quiet quitting is a visible behaviour, doing only what the role formally requires. Disengagement is the underlying psychological state. Quiet quitting is one expression of that state and maps most closely to the not-engaged category. Employees can be disengaged and still overwork, which is why counting effort is a poor way to detect employee disengagement.
How much does one disengaged employee cost?
Every published answer is modelled rather than measured, so the method matters more than the figure. A bottom-up estimate multiplies headcount by your measured disengagement rate by an assumed per-person productivity loss by fully loaded salary. Applying a 15% assumed productivity loss to an $80,000 loaded salary produces about $12,000 per disengaged employee per year, and changing that assumption to 10% or 20% moves the answer to roughly $8,000 or $16,000. Publish the range and the assumptions rather than a single figure, and remember that the only genuinely measured component is turnover cost.
Can a company fix engagement in one year?
It can move the measured number within two to three quarters, usually by acting visibly on a small number of survey findings and by improving the manager layer. Changing the underlying drivers, span of control, how relationships form in a distributed organisation, and whether people can picture their future in the role, takes considerably longer. Treat a fast score improvement as evidence that people are willing to respond honestly again, not as evidence that the engagement crisis is resolved.
Key Takeaways
The label is defensible, the framing usually is not. Global engagement at 20% in 2025 is a chronic structural condition, not a recent collapse, and it should be managed as engagement decay across cycles rather than as a bad quarter.
Every figure needs a survey, a year and a region attached. Engagement, thriving and satisfaction are different constructs measured by different instruments and cannot be stacked into one headline.
Cost figures are modelled, not measured. Build the bottom-up version with your own headcount and salary data, publish the assumptions next to the number, and treat turnover as the only genuinely measured component.
The manager layer is the structural bottleneck. Manager engagement fell from 27% to 22% globally between 2024 and 2025, so any response that adds work to managers is fighting the trend rather than the cause.
Quiet quitting is a symptom. Disengaged employees can also overwork, so measuring effort will not find them and policing effort will make the condition worse.
Visibility is not selection. An engagement program that everyone knows about and nobody chooses is not a program, which is why participation and matching should be chosen rather than assigned.
Where This Leaves You
The employee engagement crisis is real in the specific sense that matters to an executive team: a large uncommitted majority is the normal state of the working world, the direction of travel is down, and the layer most organisations rely on to fix it is disengaging faster than everyone else. What that does not justify is another campaign. It justifies looking at how relationships, development and support are structured in your organisation, and whether they depend on people who no longer have the capacity to carry them.
If your organisation is treating the employee engagement crisis as an operating problem rather than a survey score, MentorCity gives HR and executive leaders responding to falling employee engagement an all-in-one mentoring platform for employee retention and engagement programs, with self-directed, admin-managed and hybrid matching, mentoring agreements, and reporting that shows who is actually meeting rather than who was assigned.
