What losing an employee costs depends on the role

Replacing someone costs between half and two times their annual salary, according to Gallup, and voluntary turnover costs American employers a trillion dollars a year. Both figures appear in retention business cases everywhere.

Both come from the same page. We read it for the working behind either number and found none.

Gallup's range comes with no source and no method

The Gallup article says the cost of replacing an individual employee can range from one-half to two times the employee's annual salary. It calls that a conservative estimate.

No source is given for the range. There is no method, no sample, and no breakdown of which costs are included. Nothing explains what puts an employee at the top of the range rather than the bottom. It is presented as a finding and it is used as one.

The trillion is derived from it, loosely. The article cites a 26.3 percent annual American turnover rate from the Bureau of Labor Statistics for 2017 and works an illustrative example. A hundred-person organisation paying an average of 50,000 dollars faces between 660,000 and 2.6 million dollars a year. The step from that example to a national trillion is not shown.

The bigger problem is the width. A four-times range turns one business case into two different decisions. Half a salary and two salaries are not the same proposal. An organisation with a thousand people and 15 percent turnover is looking at somewhere between 11 and 45 million dollars, depending on which end it picks. Whichever end it picks, nobody can challenge it, because there is no derivation to challenge.

Two meta-analyses of turnover and performance differ by a factor of five

A separate question can be answered: does an organisation with higher turnover perform worse? Two meta-analyses tried to settle it, both published in 2013. They did not agree.

Two peer-reviewed syntheses of the same literature, in the same year, a factor of five apart. Park and Shaw drew on roughly five times the sample and reported the larger effect. Hancock and colleagues applied a different set of inclusion rules and found something close to zero.

Neither is obviously wrong. The disagreement is itself the finding. The relationship between how much turnover an organisation has and how well it performs lies somewhere between negligible and modest. It is not the catastrophe the trillion implies, and it is not nothing.

Both find that who leaves matters more than how many

Set the headline numbers aside and the two papers converge.

Involuntary turnover barely registers. Park and Shaw separate the types. Voluntary turnover comes in at −0.15 and reduction-in-force turnover at −0.17. Involuntary turnover sits at −0.01. Letting go of people who are not working out is, in the aggregate, unrelated to how the organisation performs.

That contrast does more work than either headline. The damage comes from people the organisation wanted to keep, and from people leaving in a restructure. A turnover rate reported as one number mixes a problem, a neutral fact and a management action.

The damage concentrates. Hancock and colleagues find the mean correlation moves with context. It is −0.08 for managerial employees against −0.03 overall, −0.07 in manufacturing and transportation, and −0.07 in midsize organisations. Where performance is measured as customer service it reaches −0.10. Where it is measured as quality and safety it reaches −0.12.

So the cost shows up first in what customers and regulators notice. For managers it is roughly three times larger than for the workforce as a whole. The case this makes is for targeted retention spend, and it lines up with what is happening to that group: manager engagement fell five points in a single year while the workforce around them moved one.

More turnover hurts more than proportionally. Park and Shaw also report that the strength of the relationship varies with the average turnover rate itself, which suggests a curve rather than a straight line. An organisation going from 8 to 12 percent is in a different position from one going from 25 to 29. An average effect across both describes neither.

The cost of a departure is calculable only role by role

The replacement cost of one departure is real. It is also specific to the role.

The components are not in dispute. Advertising and agency fees, the hours the hiring manager and the panel spend, onboarding and training, the output lost while the seat is empty, and the ramp before the replacement reaches full productivity. Each is knowable inside a single organisation from data it already holds. None is knowable in general.

The meta-analyses say where a general multiple goes wrong. Voluntary and involuntary departures are different events with different causes and, on the evidence, very different consequences, so a turnover figure that cannot be split cannot be costed. And a manager's departure is roughly three times as consequential as the average. A departure from a team whose output a customer sees directly, or where quality and safety are measured, is worse again. Costing every leaver at a flat multiple of salary buries the distinction the evidence says matters.

Timing makes it worse. Rubenstein and colleagues, pooling 117 studies, find a correlation of −0.27 between tenure and voluntary turnover. Leavers go early. The cost of a departure and the probability of one are correlated in the least convenient way: the people most likely to leave are the ones whose recruitment and ramp the organisation has most recently paid for. Where the leaving risk sits is examined separately.

Common questions

How much does it cost to replace an employee? The widely quoted range of half to two times annual salary is published by Gallup without a stated method or source. The underlying cost components, recruitment, onboarding, vacancy and ramp time, are real and calculable inside one organisation, but they vary too much by role for a general multiple to be meaningful.

Does turnover actually hurt company performance? Somewhere between negligibly and modestly. Two meta-analyses published in 2013 report corrected correlations of −0.03 and −0.15 for the relationship between turnover rates and organisational performance.

Why do the two meta-analyses disagree? Different inclusion rules and very different samples: 24,943 people across 48 samples in one, 120,066 across 110 independent correlations in the other. Neither has been shown to be wrong, and the disagreement is a fair description of how well the question is settled.

Which turnover is most damaging? Voluntary turnover and reduction-in-force turnover, at −0.15 and −0.17. Involuntary turnover sits at −0.01, essentially unrelated to performance.

Does it matter who leaves? Yes, more than how many. The relationship is roughly three times stronger for managerial employees than for the workforce overall, and strongest where performance is measured as customer service or as quality and safety.

Is the trillion-dollar figure reliable? It is derived from the unsourced replacement-cost range and a national turnover rate, and the article that publishes it does not show the aggregation. Treat it as rhetoric rather than as a measurement.

What should we use instead? Your own numbers, split by voluntary and involuntary, weighted by role. Every input is already in your recruiting and payroll systems, and a figure you can defend line by line survives a finance review that a borrowed multiple does not.

Sources

1. Gallup, This Fixable Problem Costs U.S. Businesses $1 Trillion
2. Hancock, Allen, Bosco, McDaniel & Pierce (2013), Meta-Analytic Review of Employee Turnover as a Predictor of Firm Performance, Journal of Management 39(3), 573-603
3. Park & Shaw (2013), Turnover rates and organizational performance: A meta-analysis, Journal of Applied Psychology 98(2), 268-309
4. Rubenstein, Eberly, Lee & Mitchell (2018), Surveying the forest: a meta-analysis of the antecedents of voluntary employee turnover, Personnel Psychology 71(1), 23-65