Replacing an employee, the popular line goes, costs between half and twice their annual salary. Gallup repeated that range in 2019 without naming where it came from.¹ The published case studies behind the cost of employee turnover tell a less dramatic, more useful story. In a 2012 review of published case studies for the Center for American Progress, Heather Boushey and Sarah Jane Glynn found that replacing a worker typically cost about a fifth of a year’s salary, with executives far above that.² Your own figure depends on the job, and five steps get you there:

1. Count the departures you will replace
2. Add up the direct costs of leaving and hiring
3. Price the empty seat and the learning curve
4. Run a low, middle and high case
5. Check your number against published estimates

In the illustrative worked example further down, a support role paying 50,000 US dollars a year costs roughly a fifth to two-fifths of salary to replace, and most of that is lost work rather than recruiting.

This page sticks to the money. For the bigger picture of engagement itself, and why researchers and Gallup define it differently, see what employee engagement means.

Where the “half to twice a salary” figure comes from

The half-to-twice range has a long trail and little visible data. Gallup’s 2019 article by Shane McFeely and Ben Wigert, which is company research and was not peer-reviewed, states it, calls it conservative and cites nothing for it; Gallup also sells the manager training the article recommends.¹ A 2008 guide published by the SHRM Foundation, the nonprofit affiliate of the US Society for Human Resource Management, gives a similar range of 90 to 200 percent and credits it to a textbook by Wayne Cascio and a 2001 article by Terence Mitchell and colleagues.³

can range from one-half to two times the employee’s annual salary

Neither source lets a reader see which jobs, firms or cost items produced the range. That matters because turnover costs are not a law of nature: they are an accounting choice about what to count. Put the value of a salesperson’s lost clients into the sum and the cost can pass a year’s pay; count only the job ad and the interviews and it may be a few weeks’ pay.

Pricing turnover is also an old habit. Peter Hom and colleagues, looking back over 100 years of turnover studies in 2017, trace it to 1917, when consultants were already writing up turnover costs and ways to reduce them.⁴ A century later, a round and alarming number still travels further than a worksheet.

Myth

Replacing any employee costs between half and twice their salary.

Fact

That range comes without a visible data source. Published case studies find most jobs cost far less to replace, and a few senior roles cost far more.

The practical lesson: when a slide or a vendor quotes a turnover cost, ask which job it describes and which costs it counts. If nobody can show the list, treat the figure as a slogan.

Case studies put the typical cost nearer a fifth of salary

Pooled case studies land well below the popular range for most jobs. Boushey and Glynn gathered published papers that costed turnover with a detailed method, turned each estimate into a share of annual pay, and reported the median, the middle value, noting that a few very expensive senior roles skew the data upward.²

The study

Limited evidence

Boushey and Glynn (2012), Center for American Progress

Excluding executives and physicians, the median cost of replacing a worker across 27 case studies was 21 percent of annual salary. For jobs paying 30,000 US dollars or less a year it was 16 percent, while very highly paid and senior roles reached up to 213 percent. Estimates ran from 5.8 to 213 percent, and 17 of the case studies for jobs paying up to 75,000 dollars fell between 10 and 30 percent.²

Two caveats point in opposite directions. Only two of the review’s eleven papers tried to count indirect costs such as lost productivity, so the median mostly covers direct costs, and most salaries were assigned from national occupation averages rather than taken from the firms. The review was also written by a think tank arguing for workplace flexibility and earned sick days, so the authors had a reason to show that turnover is costly, not cheap.²

An independent employer survey points lower still. Arindrajit Dube, Eric Freeman and Michael Reich of the University of California, Berkeley, analyzed a survey that asked California workplaces in 2003 and 2008 what it cost to replace a worker, including separation, hiring and on-the-job learning. The mean answer was about a tenth of the average annual wage per hire, but the median workplace reported only 2 to 4 percent; a long tail of costly cases pulled the mean up.⁵

For a manager, the lesson is to start from the job, not from a headline. A cashier and a senior engineer do not share a turnover cost, and an average across both tells you little about either.

Why much of the cost of employee turnover is hidden

In the most itemized of these studies, the costs of separating, recruiting, selecting and training added up to slightly less than the cost of lost work. That larger part is work that does not get done: the weeks a seat is empty, the hours colleagues spend covering it, and the months a new hire needs before working at full speed.⁶

J. Bruce Tracey and Timothy Hinkin of Cornell’s hotel school built a costing tool around five categories: pre-departure, recruitment, selection, orientation and training, and lost productivity. In a convenience sample of 12 hotels, lost productivity was on average the largest share, about half of the total. The report was published by an industry-supported research center, and the sample is small.⁶

1. Costs you can see: separation paperwork, job ads and agency fees, interviews, orientation and formal training
2. Costs that stay out of sight: the empty seat, colleagues covering the work, the new hire’s learning curve, lost customers and know-how

The textbook model is similar. Wayne Cascio and John Boudreau’s book Investing in People, published in an alliance with the Society for Human Resource Management, costs turnover as separation, replacement and training, then adds the performance difference between a leaver and a replacement, and lost productivity and business.⁷

Picture a support agent who resigns. The job ad costs a few hundred dollars. The real bill is quieter: a queue that grows for five weeks, two colleagues on overtime, and a new hire who needs help with every unusual ticket until spring. None of that arrives as an invoice, which is why a budget that only tracks recruiting spend tends to underestimate turnover. The lesson: count time as well as invoices, meaning the empty seat, the colleagues covering it and the new hire’s learning curve.

How to estimate the cost of employee turnover, step by step

A usable estimate prices one typical departure for one job, multiplies it by the number of departures you will replace, and reports a range rather than a single number. The steps below follow the cost categories used by Cascio and Boudreau and by Tracey and Hinkin.⁷⁶

1. Count the departures you will replace

Start from your own records: how many people left this job in the last year, and how many of those roles you filled again. Keep resignations, dismissals and layoffs apart: a laid-off worker may not be replaced at all, and a role that is cut creates no hiring or training cost.²

National figures give context. In the US, an average of about 2 percent of workers quit each month in 2025, in the Bureau of Labor Statistics JOLTS survey, from under 1 percent in government to about 4 percent in accommodation and food services.⁸ The monthly quits rate was still about 2 percent in July 2026.⁹ For a team of 50, the national average works out to about one resignation a month; the useful comparison is your own rate against your industry’s.

2. Add up the direct costs of leaving and hiring

List separation costs (exit paperwork and final pay), recruiting (job ads, agency fees), selection (interview hours, checks) and onboarding and training (trainer time, materials).⁶ Price staff time at hourly pay plus benefits. This part is easy to find and usually the smaller one. If hiring itself is slow or costly, look next at how your hiring process is designed.

3. Price the empty seat and the learning curve

For the vacancy, estimate how many working days the seat stays empty and what covering it costs: overtime, temporary staff, or work that simply waits. For the learning curve, estimate how many months a new hire takes to reach full output and roughly what share of full output they manage meanwhile. Tracey and Hinkin list both, along with the lower output of someone working out their notice.⁶ Experienced supervisors can usually give a first guess; write it down as an assumption.

4. Run a low, middle and high case

Multiply the cost of one departure by the number you expect to replace in a year. Then repeat the sum with optimistic and pessimistic assumptions, because the vacancy and learning-curve guesses move the answer most. The illustrative example below uses invented figures for a US support team, not data from a study.

The example values lost output at the employee’s pay, a cautious convention; for a role that brings in revenue, the value of the lost work can be much higher.

5. Check your number against published estimates

If your middle estimate for a frontline job comes out at twice a salary, recheck the learning-curve assumption before you present it. If it comes out at only a few percent of salary, check whether you left out the vacancy or the ramp-up. The published case studies are old and mostly American, so use them as a sense check, not a target.²

WiserHours

What a credible turnover estimate shows

• Name the job the estimate covers
• Count only departures you will actually replace
• Price staff time at pay plus benefits
• Include vacancy days and the learning curve
• Write down every assumption
• Show a low, middle and high figure
• Compare with published ranges for similar jobs

Find this checklist online:
https://wiserhours.com/employee-engagement/cost-of-employee-turnover/#checklist-cost-of-employee-turnover-estimate

Not every departure is a loss

A turnover cost is only half of a decision, because some departures help. The SHRM Foundation guide separates dysfunctional turnover, the loss of people who are valuable and hard to replace, from functional turnover, such as the exit of a poor performer, which does not hurt the organization.³

Research on whole organizations also suggests that the kind of exit matters. A 2013 meta-analysismeta-analysis: A study that combines the results of earlier studies on the same question into one overall estimate. Pooling makes the estimate more precise, but it cannot repair the studies it pools: a meta-analysis of surveys is still survey evidence.Full entry in the glossary by Tae-Youn Park and Jason Shaw found that higher turnover rates were linked to lower organizational performance, a modest correlationcorrelation: A measure of how closely two things move together, running from minus one, where one rises as the other falls, through zero, meaning no link, to plus one. It says how strong the relationship is, not what causes it, and it is not a percentage.Full entry in the glossary that was clearer for voluntary quits and layoffs than for dismissals and other involuntary exits, which were barely related. The link also varied with how high turnover already was.¹⁰ These are correlations, so they show that high turnover and weaker results tend to go together, not that one causes the other.

So, before you spend to cut turnover, ask who is leaving. Losing a strong performer in a hard-to-fill role costs more than the average; losing someone who was struggling may cost little or nothing. The goal is to find the departures that hurt, and a regular one-on-one with each person is one place to hear about them early.

The bottom line

The cost of losing an employee is real, but “half to twice a salary” is a slogan, not a finding. For most jobs the published case studies point lower, and the biggest items are the empty seat and the learning curve rather than the recruiting bill. Estimate your own figure for one job at a time, show a range with your assumptions, and put it next to the question of who is leaving.