Why Employees Leave: What 2.8 Million Responses Reveal
Ask most executives why employees leave, and the answer defaults to compensation almost automatically. It’s the easiest explanation, the one that requires the least uncomfortable self-examination, and, according to a growing body of research, largely the wrong one.
Across responses collected through the Most Loved Workplace® certification program, spanning millions of employees at certified organizations, the pattern that emerges consistently is this: pay is rarely the primary driver of departure. Belonging, trust in leadership, and recognition are.
The Scale of the Data
Most conversations about why employees leave rely on a single company’s exit interviews, or a survey of a few hundred people at a conference. That’s not nothing, but it’s not enough to separate a real pattern from noise specific to one organization’s circumstances.
The Most Loved Workplace® certification program draws on research validated across 2.8 million employees at more than 1,800 organizations, combining structured quantitative scoring with open-ended qualitative feedback. At that volume, patterns that hold up aren’t explained away by any single company’s unique circumstances. They’re patterns that hold across industries, company sizes, and geographies.
That scale matters because it lines up with, and reinforces, what some of the most rigorous independent research on this topic has already found.
The Independent Research Says the Same Thing
McKinsey’s widely cited Great Attrition research asked departing employees directly why they left, and separately asked employers why they thought people were leaving. The gap between the two answers was stark. Employers assumed compensation and job market opportunity. Employees gave a different answer entirely: 54% said they didn’t feel valued by their organization, 52% said they didn’t feel valued by their manager, and 51% cited a lack of belonging at work. All three of the top reasons are relational, not transactional, and none of them get solved by a bigger raise.
Gallup’s research points to the same underlying driver from a different angle. Gallup has found that managers account for at least 70% of the variance in team engagement, and that it takes more than a 20% pay raise to lure an engaged employee away from a manager they trust, while a disengaged employee can be poached for next to nothing. The manager relationship isn’t one factor among many. It’s close to the whole story.
McKinsey’s research has also found that roughly 41% of employees cite limited advancement opportunity as a primary reason for leaving, a figure that lines up closely with what MLW’s own certified-employer data shows: growth and career pathway clarity are consistently among the most frequently cited gaps in open-ended employee feedback, even at otherwise highly rated organizations.
What "Belonging" Actually Means in the Data
Belonging gets used loosely enough in HR conversations that it’s worth being specific. In employee open-ended responses, belonging rarely shows up as an abstract philosophical concept. It shows up as concrete, specific language: being included in decisions that affect their work, having their input actually incorporated rather than just solicited, and seeing people like themselves represented at senior levels of the organization.
McKinsey’s research found this pattern has a demographic dimension worth naming directly: employees who identified as non-white or multiracial were more likely than their white counterparts to cite lack of belonging as their reason for leaving. That’s not a footnote. It means belonging as a retention driver isn’t evenly distributed across an organization, and a company that measures overall belonging scores without segmenting by demographic group can miss exactly the population most at risk of leaving for this specific reason.
What "Trust in Leadership" Actually Means in the Data
Trust in leadership, similarly, isn’t a single monolithic score. Across certified employer data, it consistently breaks into two distinct layers that don’t always move together: trust in an employee’s direct manager, and trust in executive leadership further removed from day-to-day work.
The gap between those two layers is itself a meaningful signal. It’s common, particularly at fast-growing or recently restructured organizations, for manager-level trust to be strong while executive-level trust lags behind, simply because employees have far more direct exposure to their manager’s behavior than to distant executive decision-making. That gap isn’t necessarily a crisis. Left unaddressed for years, though, it’s exactly the kind of erosion that shows up later as unexplained attrition among otherwise satisfied teams, because a strong direct-manager relationship can only compensate for distant, distrusted leadership for so long.
What "Recognition" Actually Means in the Data
Recognition, in employee language, is rarely about the existence of a formal recognition program. Plenty of organizations with award ceremonies and “employee of the month” plaques still score poorly on recognition in employee feedback, because what employees are actually describing when they cite recognition as a reason for staying or leaving is whether their specific, individual contribution gets seen by the people whose opinion matters to them, not whether a generic program technically exists.
This is a meaningful distinction for any leader tempted to treat recognition as a solved problem because a program is already in place. The data doesn’t measure whether a recognition program exists. It measures whether employees feel recognized, which is a considerably higher bar.
Why Pay Still Matters, Just Not the Way Most Leaders Think
None of this means compensation is irrelevant. Pay operates as what’s sometimes called a hygiene factor: falling meaningfully below market rate is a genuine, legitimate reason to leave, and no amount of belonging or recognition fully compensates for being underpaid relative to comparable roles.
What the data across both MLW’s certified-employer responses and the independent research consistently shows is that once compensation clears a reasonable, competitive bar, additional pay increases produce diminishing returns on retention, while the relational factors, belonging, trust, recognition, keep predicting departure regardless of how competitive pay already is. Raising pay for an employee who doesn’t feel valued by their manager buys time. It doesn’t fix the underlying reason they were already looking.
How This Varies by Career Stage
The relative weight of belonging, trust, and recognition isn’t identical across every point in someone’s career, and treating retention strategy as one-size-fits-all across tenure and career stage misses a meaningful part of the pattern.
Early-career employees, particularly those within their first two years at an organization, tend to weight belonging and mentorship access especially heavily, since they’re still forming their professional identity and looking for signals about whether they fit and are valued, independent of how well their specific technical contribution is being recognized yet. Mid-career employees, several years into their field but not yet in senior leadership, tend to weight growth and advancement clarity most heavily, consistent with the 41% advancement-related figure cited above, since they’re actively evaluating whether staying versus moving elsewhere offers a better path forward. Senior, more tenured employees tend to weight trust in executive-level leadership and alignment with organizational direction most heavily, having already established belonging and secured advancement, with departure risk increasingly tied to whether they still believe in where leadership is taking the organization.
This matters practically because a single, undifferentiated retention strategy, one recognition program, one mentorship initiative, applied uniformly across an entire workforce, will underperform relative to a strategy that accounts for which factor matters most at which career stage. An organization losing early-career talent and an organization losing senior talent are frequently experiencing two genuinely different problems, even if both show up as the same aggregate turnover number.
The Cost of Misreading This Data
Getting this analysis wrong isn’t a neutral mistake, it’s an expensive one. Gallup estimates that replacing a single employee costs between 50% and 200% of their annual salary, with leadership and technical specialist roles running toward the higher end of that range. When an organization responds to attrition primarily with compensation adjustments, addressing a driver that the data consistently shows isn’t the primary cause, it pays twice: once for the retention initiative that doesn’t actually solve the underlying problem, and again for the replacement costs of the employees who leave anyway because the real issue, belonging, trust, or recognition, was never addressed.
There’s a secondary cost that’s harder to quantify but arguably more damaging: McKinsey’s research on the gap between what employers assume and what employees actually report found that responding to attrition with compensation alone sends an unintended signal to the employees who remain, that the organization views the relationship as fundamentally transactional. That signal tends to accelerate exactly the belonging and trust erosion that was driving departures in the first place, turning a retention strategy meant to solve the problem into something that quietly reinforces it.
Why Exit Interviews Alone Undercount These Factors
Most organizations already collect exit interview data, and many leaders assume that data would surface belonging, trust, and recognition issues if they were genuinely driving attrition. In practice, exit interviews systematically undercount relational reasons for leaving, for a specific, well-documented reason: a departing employee has little incentive to give their soon-to-be-former employer honest, potentially uncomfortable feedback about a manager or leadership team they may still need a reference from, or may encounter again professionally.
The result is that exit interviews disproportionately surface safer, more socially acceptable reasons for leaving, a better offer, career growth, relocation, even when the actual underlying driver was a strained manager relationship or a persistent feeling of not belonging. This isn’t dishonesty on the departing employee’s part so much as reasonable self-protection, but it means organizations relying primarily on exit interview data to understand attrition are working from a systematically skewed sample.
This is a large part of why aggregate, anonymized, independently administered assessment produces a meaningfully different, and more accurate, picture than exit interviews alone. Current employees, surveyed anonymously and independently rather than on their way out the door, have less reason to soften their answers, which is exactly why the pattern across millions of responses collected this way diverges so clearly from what most executives assume based on their own organization’s exit interview archive.
What This Means for Retention Strategy
The practical implication is a reordering of where retention investment goes, not an argument to ignore compensation entirely.
Audit belonging by demographic segment, not just in aggregate. Given the documented gap in how belonging as a leaving-reason varies by employee background, an organization-wide belonging score can hide exactly the population at highest risk.
Separate manager-trust data from executive-trust data. Because these two layers frequently diverge, especially during periods of growth or change, treating “leadership trust” as one undifferentiated number obscures where the actual erosion is happening.
Measure recognition as felt experience, not program existence. Whether a recognition program exists is an input metric. Whether employees report feeling specifically, individually recognized is the outcome metric that actually predicts retention, and it’s worth measuring directly rather than assuming the former guarantees the latter.
Treat compensation as a floor, not a lever. Get pay to a genuinely competitive level, then redirect further retention investment toward the relational factors the data says are doing the actual predictive work.
Where Most Organizations Get the Diagnosis Wrong
The most common mistake isn’t ignoring this data entirely, most people leaders have encountered some version of “it’s not about the money” by now. The more common mistake is accepting the general principle while still designing retention strategy as if compensation were the primary lever, because compensation is the easiest variable to change quickly and the easiest one to justify to a board or finance team.
A pay adjustment is a single decision, executed in one budget cycle, with a clear before-and-after number to point to. Fixing a manager-trust gap, or a belonging problem concentrated in a specific demographic group, or a recognition culture that technically has a program but doesn’t produce the felt experience employees describe, requires sustained behavioral change that’s harder to schedule, harder to budget for cleanly, and harder to claim credit for in a single quarter. That asymmetry in difficulty, not a genuine disagreement with the data, is why compensation-first retention strategy persists even at organizations whose own exit data says something different.
The second common mistake is treating this as a one-time diagnostic rather than an ongoing measurement. Belonging, trust, and recognition all shift over time, in response to leadership changes, restructuring, growth, and external market conditions. An organization that measured these factors two years ago and hasn’t reassessed since is operating on a diagnosis that may no longer reflect current reality, particularly if anything meaningful has changed organizationally in the interim.
Where This Data Comes From
The scale and consistency of this pattern is only meaningful because of how it was collected: independently, through direct employee assessment rather than employer-authored surveys, across Most Loved Workplace® certified organizations. That independence is what makes the aggregate pattern trustworthy. A single company’s HR-administered exit survey reflects one organization’s blind spots. Millions of responses collected the same, independently administered way across a wide range of organizations reflect something closer to a genuine, generalizable pattern in why people actually leave. Best Practice Institute’s research team reviews the employee evidence and makes the certification decision, and use of the earned badge requires an active paid license.
For organizations building their own retention strategy, the starting point isn’t guessing which of these factors matters most internally. It’s getting an independent read on where your own organization actually stands across belonging, trust, and recognition specifically, rather than assuming compensation data alone tells the whole story.
Frequently Asked Questions (FAQ)
Is pay the main reason employees leave their jobs?
How much does a manager relationship actually affect retention?
Does belonging affect all employees equally?
If pay isn't the top driver, does compensation matter at all?
Louis Carter is the founder and CEO of Best Practice Institute, Most Loved Workplace, and Results-Based Culture. Author of In Great Company, Change Champions Field Guide, and Best Practices in Talent Management, as well as a series of Leadership Development books. He is a trusted strategic advisor and coach to CEOs, CHROs, and leaders of mid-sized to F500 companies – enabling change and steering employer brand development together with highly effective teams, leaders, and organizations as a whole.
0 Comments