Reduce employee turnover

Most retention programs are built against a single figure that describes nobody in the company. The annual rate goes up, a budget gets approved, the rate comes down two points the following year, and the same people leave.

That is not a failure of effort. It is a failure of instrument. A company-wide turnover rate is an average, and an average is a device for making a distribution disappear. You cannot aim at what it has hidden.

Here is the version of this work that moves the number, and it starts before any of the tactics.

Why the aggregate is the wrong instrument

Two departments can post identical attrition and have nothing in common underneath. One is losing people it cannot replace. The other is losing people it was going to manage out anyway. Same figure, opposite problems. Run one program across both and it is wrong in one of them by construction.

The aggregate also moves for reasons that have nothing to do with retention work. It falls when the labor market tightens and people stop leaving anywhere. It falls when a hiring freeze changes the denominator. Right now it is falling for the second reason almost everywhere: the US quits rate has sat at or near 1.9 percent for long enough that flat is the honest description, and the Bureau of Labor Statistics called every component of its July 2026 release little changed (Bureau of Labor Statistics, Job Openings and Labor Turnover Survey, July 2026).

So a retention initiative launched into a low-quits year will show an improvement it did not cause. The instrument cannot distinguish between a program that worked and a market that cooled.

Every company knows how many people left. Almost none know which ones.

The disaggregation, and the order to do it in

The cut that resolves most of this takes an analyst an afternoon against data already in the HRIS. The order matters, because the first cut usually ends the exercise.

Cut one: by manager. Not by department, not by function. By the individual the leaver reported to. This is where the variance lives, and it is normally the largest single source of spread in the dataset. Research on culture-related resignations puts poor management at the root of 58 percent of them, which is consistent with what the cut tends to show.

Cut two: by tenure band at exit. Departures inside twelve months are a hiring and onboarding problem. Departures at two to four years are usually a progression problem. Departures at eight-plus years are frequently not a problem at all. Three different causes wearing one number.

Cut three: by performance band at last review. This is the cut that decides whether the company has an expensive problem or an ordinary one. Losing the bottom quartile at 20 percent is a functioning system. Losing the top quartile at 20 percent is an eight-month-old leak nobody has costed.

Cut four: the spread. Take the best-retained group and the worst-retained group by manager and subtract. That number is the differential, and it is the one worth managing, because it is the only one that describes something inside the company's control.

Which departures are addressable, and which are not

This is the part most retention programs skip, and skipping it is why budgets get spent flat across a population where only a fraction of the departures were ever movable.

Not addressable. A spouse's relocation. A career change out of the sector. Retirement. A better offer at a level the company does not have. A third to a half of departures in a professional population fall here, and a retention program aimed at them buys nothing.

Addressable, and usually unpriced. Departures that cluster under one manager. Departures at the same tenure point year after year. Departures preceded by a stalled progression conversation. Departures where the person raised something, was told they were being too sensitive about it, and went quiet nine months before they resigned.

That last group is the expensive one. It is also the one an exit interview will never surface, because by the time the form arrives the person has decided that honesty costs them a reference. The polite fictions on those forms — seeking new opportunities, better compensation elsewhere, time for a change — are what the system records instead.

The arithmetic, with the assumptions on the outside

Take a thousand professional employees and a ten-point spread between the best-retained group and the worst. That spread produces roughly twenty-five departures a year above the baseline the best group is already achieving.

Replacement cost is where the assumption sits, so here it is in the open. The Society for Human Resource Management's published bands run 30 to 50 percent of salary for entry-level roles, 100 to 150 percent for professional roles, and up to 200 percent for specialized ones. Gallup gives one-half to two times salary and calls that conservative. Around $220,000 per professional departure sits inside those bands rather than above them, and it is the figure the calculator uses. It is an assumption, not a measurement. Anyone running this should substitute their own loaded cost and watch what happens.

Twenty-five departures at that figure is about $5.5 million a year, and about $16.5 million across three. That is the spread, priced. It is not the cost of turnover. It is the cost of the part of turnover that one group of managers is producing and another is not, in the same company, under the same policies.

Two things about that number. It is a difference rather than a total, which means it survives the objection that some turnover is healthy. And because it is a difference, it has a defensible target: the best-retained group is already hitting it.

The national figures exist and are too large to act on. SHRM's 2019 work put US employers at up to $223 billion over five years, and MIT Sloan found culture roughly 10.4 times more predictive of attrition than compensation (Sull, Sull and Zweig, MIT Sloan Management Review, “Toxic Culture Is Driving the Great Resignation”). The ratio is the part to carry into a budget meeting. Ratios survive scrutiny in a way totals do not.

What my own accounting says

I have run companies at near ninety-five percent voluntary retention. It took about eighteen months of management changes, system redesigns, and an accepted productivity dip while it bedded in. Not every attempt worked, and departments with selective hiring were poor models for the rest.

My own replacement-cost estimate for a highly perceptive employee runs above the published bands, at 200 to 400 percent of salary. That is not a benchmark. It is my accounting, and it includes the pipeline disruption, the client relationships, and the problems that person was quietly preventing, which never appeared anywhere because they never happened.

Run it yourself

An afternoon, on data already held.

1. Export twenty-four months of voluntary exits. Twelve is not enough to see a manager-level pattern; the cell sizes are too small.

2. Join each exit to the manager of record at the time of exit, tenure at exit, and last performance rating.

3. Rate by manager, for every manager with six or more reports across the window.

4. Rank, take the top and bottom quartile, subtract. That is the spread.

5. Multiply the excess departures by a loaded replacement cost you can defend. Show the assumption on the slide.

6. Cross-check the top-quartile-performer exits separately. If that rate is above the company rate, the spread is understating the cost.

This cut names individual managers, so decide who sees the output before it is run rather than after.

The three changes that move the addressable ones

Everything above is measurement. These are the interventions that have moved the number for me, and they are structural rather than motivational.

Make progression criteria explicit and written. Most two-to-four-year departures are people who could not find out what the next step required. An implicit standard is one that gets applied differently by every manager holding it, and the difference is invisible until it shows up as attrition.

Put a channel between the employee and the organization that does not route through their manager. A quarterly written question, read by someone outside the reporting line. One question is enough: what is the biggest obstacle to doing your best work. The manager cut above is worthless if the only way to report a manager problem is through that manager.

Put retention into promotion criteria for managers. Where it is absent, the managers producing the worst half of the spread keep getting promoted on output their own turnover is subsidising.

What to measure ninety days out

Not the rate. Ninety days is too short for the rate to say anything.

Measure the spread instead. It is the thing under the company's control and it responds faster than the aggregate does. Then measure one leading indicator: how long it takes a new hire to stop raising things. If that interval is lengthening, something upstream is working.

Two numbers instead of one, and the second is the one that survives a budget meeting. Most finance functions have neither, and running the cut once puts a company ahead of almost everyone it competes with for the same people.

What is the spread between your best-retained manager and your worst?

The calculator that runs this arithmetic on your own headcount, with your own replacement-cost assumption: Employee Turnover Cost Calculator

Peter Allen Mann is the author of The Perception Revolution (Fast Company Press, 2026) and has spent thirty-seven years in leadership across the Navy, Dell, Tech Data and two companies he founded. He writes at peterallenmann.com.

Sources

• Bureau of Labor Statistics, Job Openings and Labor Turnover Survey, July 2026 reference month, released 1 September 2026. Quits 3.1 million, quits rate 1.9 percent. bls.gov/news.release/jolts.nr0.htm
• Society for Human Resource Management, The High Cost of a Toxic Workplace Culture, 2019. Up to $223 billion over five years in US turnover costs attributed to culture.
• Sull, D., Sull, C., & Zweig, B., “Toxic Culture Is Driving the Great Resignation,” MIT Sloan Management Review, 11 January 2022. Culture 10.4 times more predictive of attrition than compensation. sloanreview.mit.edu — Toxic Culture Is Driving the Great Resignation
• Society for Human Resource Management, published replacement-cost bands: 30–50 percent of salary entry-level, 100–150 percent professional, up to 200 percent specialized. Gallup: one-half to two times salary.
• Mann, P. A. (2026). The Perception Revolution. Fast Company Press. Ch. 2 and Ch. 3.