Roughly three out of four resignations were preventable. That is the finding from the Work Institute’s 2025 Retention Report, built on more than 120,000 exit interviews: about 75% of the people who quit left for reasons their manager could have influenced. Not the economy, not a competitor’s signing bonus, not something structural and out of reach. Reasons like workload, recognition, growth, and whether anyone was paying attention.

That number should change how you think about turnover. If most departures are preventable, then most of them are, quietly, a management outcome. The resignation letter is the last event in a chain that started weeks or months earlier, and the person handing it to you decided some time ago. By the time you are reading it, you are not managing retention anymore. You are processing a loss that already happened.

The good news buried in that same statistic is that the levers are yours. This is a walk through what the data says actually drives people out, how to read the early signals before the notice lands, and what holds good people in place. I have spent 20 years in IT operations leadership and fractional COO work, and I have lost people I should have kept. Almost every time, the warning was there and I was too busy shipping to look at it.

The number most managers never run

Turnover feels expensive in a vague way. Put an actual figure on it and it stops being vague. SHRM estimates that replacing an employee costs between 50% and 200% of their annual salary, depending on the role and seniority. For someone earning $80,000, that is $40,000 on the low end and $160,000 on the high end, once you count recruiting, interviewing time, onboarding, and the three-to-six-month ramp before a replacement reaches full output.

That figure understates the real damage, because it only captures the person who left. It misses the load that lands on everyone who stayed, the projects that slip while a seat sits empty, and the institutional knowledge that walks out the door and does not come back. When a senior engineer left a rollout I was running years ago, the direct replacement cost was the smallest part of it. The expensive part was the six weeks the rest of the team spent rediscovering things that had lived only in his head.

Here is the number that ties it together. Gallup’s research on the State of the American Manager found that managers account for at least 70% of the variance in team engagement. Not the company brand, not the perks, not the industry. The manager. If you want to know why one team on a floor is thriving while the team next to it is bleeding people, the biggest single explanation is usually the person running each one. That is a heavy thing to sit with, and it is also the most useful fact in this whole article, because it means the outcome is largely in your hands.

The exit interview is the polished version

Ask people why they left and they will hand you something socially acceptable. “Better opportunity.” “Career growth.” “Timing was right.” Those are the answers that keep a bridge intact. They are rarely the whole truth.

When McKinsey surveyed more than 13,000 workers about why they actually quit, the top three reasons were a lack of career development and advancement (41%), inadequate total compensation (36%), and uncaring or uninspiring leaders (34%). Notice that two of the top three have nothing to do with money. People leave when they cannot see a future, and they leave when they do not believe the person above them cares. Both are things you shape every week whether you mean to or not.

The Work Institute data points the same direction. Career development has topped its list of departure reasons year after year, ahead of health, family, compensation, and everything else. Translate that out of survey language and it says something simple: the most common reason a good employee quits is that they stopped believing they were going anywhere. The frustrations of the present feel permanent when there is no visible path out of them.

So when you get “better opportunity” in an exit interview, read it as a diagnostic, not an explanation. The real question is what made the other opportunity look better, and the honest answer is usually that the person could picture a future there that they could no longer picture with you.

Reading the signals before the notice

You do not need a survey to catch this early. Disengagement shows up in behavior weeks before it shows up in a resignation, and the tells are consistent. Watch for these five:

Withdrawal from discussion. The person who used to push back, propose things, and argue a point goes quiet. They attend, they nod, they contribute the minimum. When someone who was loud goes silent, that is not peace. That is checkout.

A slide in reliable output. Someone who has always been solid starts missing dates or turning in work below their own standard. This can be burnout, disengagement, or both, and all three warrant a conversation rather than a correction.

Attendance drift. An uncharacteristic uptick in sick days, late starts, or scattered time-off requests can mean interviews are happening, or that coming in has quietly become hard.

Pulling back socially. When someone stops joining the casual conversation and starts keeping their distance from colleagues they used to click with, they are often detaching from the team before they detach from the job.

Sudden interest in the mechanics. Pointed questions about how promotions get decided, what other teams do, or how pay bands work are frequently the sound of someone benchmarking their options.

None of these confirms that a person is leaving. Each one confirms that a conversation is overdue. The manager who notices the withdrawal in week two has options. The one who notices it in the resignation meeting has a farewell card.

This is also where reading burnout signals early overlaps directly with retention, because your highest performers rarely announce that they are drowning. They are the ones you keep handing work to precisely because they never say no, and they are the ones who leave with the least warning.

What actually holds people

Retention is not a program you launch when people start leaving. It is a byproduct of how you manage on ordinary weeks. The practices below are unglamorous and they compound.

Career conversations, on a running basis

Most managers save development talk for review season. By then the person you most wanted to keep has already had three interviews elsewhere. Given that career stagnation is the single most cited reason people quit, waiting for an annual slot is the most expensive scheduling decision you make.

Make career development an ongoing thread instead of a yearly event. Find out where each person wants to be in two or three years, then connect the work in front of them to that direction. You cannot always promise a promotion. You can almost always help someone build a skill, widen their scope, or take on a project that matters to them, and that momentum is often what keeps them from taking the recruiter’s call.

Recognition specific enough to be believed

Generic praise moves no one. “Great job” is noise. Specific, timely acknowledgment is signal, because it proves you were actually watching. Compare “thanks for your work this week” with “the way you kept that client calm on Thursday and pulled everyone back to the fix instead of the blame, that is exactly the judgment this team needs.” The second one lands because it could only have been said by someone paying attention.

Different people want recognition in different forms. Some want the callout in the team meeting, some would rather have a quiet message. Learn which is which. The cost is close to zero and the return on retention is real.

One-on-ones that go somewhere

Weekly or biweekly one-on-ones are the highest-leverage tool you have, and most managers waste them on status updates they could have gotten by email. Use the time for the questions email cannot answer. What is getting in your way. What are you spending time on that feels pointless. What would make this job more meaningful. Then follow up on what you hear, because a one-on-one where nothing ever changes trains people to stop being honest in it.

Some organizations formalize this with stay interviews, a structured conversation about why someone stays and what would make them consider leaving, held while they are still happy rather than on their way out. Done sincerely, they surface the fixable problems while there is still time to fix them.

Fixing the small frictions before they stack

People rarely quit over one dramatic thing. They quit over the accumulation: the broken tool nobody replaced, the process that adds an hour a week for no reason, the recurring meeting that decides nothing, the cross-team snag that never gets resolved. Individually, each is survivable. Together they read as “nothing here will ever get better.”

Ask the direct question: what is one thing that slows you down that we could actually change? Then change one of them and let people see it change. Even a small fix is evidence that speaking up is worth the effort, which is exactly the belief you need to protect.

Clarity, so people stop burning energy on ambiguity

When people do not know what success looks like or how decisions get made, they spend energy navigating fog instead of doing their best work, and that drain pushes them toward places that feel more predictable. Being explicit about priorities, saying so when goals shift, and admitting when you do not know something all cost you nothing and buy back a surprising amount of goodwill. Clarity is not having every answer. It is being straight about where things stand.

Workload and the burnout you cannot see

Gallup’s State of the Global Workplace 2025 report found global employee engagement fell to 20%, its lowest level since 2020, and that low engagement now costs the world economy roughly $10 trillion a year, about 9% of GDP. Manager engagement dropped even harder, falling five points in a single year to 22%. That last figure matters for retention in a way that is easy to miss: a depleted manager cannot hold a team. If you are running on empty, the attention this whole article depends on is the first thing to go.

Overworked people do not always complain. Your best performers are often the quiet exception, carrying more because they are capable and reliable, right up until they are gone. Stop asking “are you doing okay,” which invites a reflexive yes. Ask “walk me through what is on your plate right now, and tell me honestly whether the volume is sustainable.” Protecting a person’s capacity is a retention move. So is pushing back on unrealistic demands from above when it is your team paying the cost.

Pay matters, and it is not the whole story

Compensation is real. If your people are paid meaningfully below market, no amount of good management holds them forever, and it is worth knowing where your team sits against industry benchmarks and advocating when the gap is genuine. The BLS quits rate has settled near 1.9% with about 3.1 million people voluntarily leaving jobs in a typical recent month, according to the Job Openings and Labor Turnover Survey, which means the market for your best people is always open even when the headlines are calm.

But pay alone does not retain anyone. People who are compensated well and still feel unheard, unchallenged, or invisible will leave for a lateral move somewhere they feel seen. Money removes a reason to leave; it does not create a reason to stay. The full picture is pay, growth, recognition, relationship, and meaningful work, and all five need to be in reasonable shape at once.

When someone is already halfway out

Sometimes you catch the signals late and the person has, at least tentatively, made up their mind. You are not out of options, but you have to move carefully.

Ask for a direct conversation, and make clear it is not a performance discussion. Tell them you have noticed something feels different and you want to understand it. Then listen more than you talk. Do not jump to solutions, and do not make promises you cannot keep, because an empty gesture offered after someone has raised a real concern is worse than saying nothing. If there is something genuine you can change, a different project, more flexibility, a real conversation with leadership about their path, do it, but only if you mean it.

You will not save everyone, and you should not try to. Some turnover is healthy, and some decisions are already final. But some people are still weighing it, and a straight conversation at the right moment is occasionally the thing that tips them back.

The version of this that works

None of this is complicated, which is exactly why it gets skipped. Retention is not perks, retention bonuses, or a ping-pong table. It is the daily experience of working for you: whether people feel seen, whether they are growing, whether their work matters, and whether they trust you to be straight with them. That trust is built in small consistent deposits and spent in single careless withdrawals.

Given that most departures are preventable and the manager explains the majority of the variance, the honest reframe is this: the question is not why good people leave. It is whether you were paying attention while they were still deciding. Start this week with one conversation. Ask someone on your team what would make their job better, then change one thing they tell you. That is what retention actually is, one small act of attention at a time, done long before the notice ever arrives.