Where the 75% of turnover is preventable figure comes from

Provenance: Definition, not measurement. The figure follows from how its author labelled the categories, so it measures the label rather than the thing the word suggests. This describes how the figure is sourced, not whether it is true, and it is not a claim that anyone acted in bad faith.

You will see it as 75%, 77% or 78% depending on which year's report the writer picked up, usually credited to Work Institute, and usually as the opening line of an argument that the employer is in control: most turnover is preventable, so here is the program that prevents it.

The number is genuine. Work Institute's 2017 Retention Report says that "over 75% of the reasons employees leave could have been prevented by the employer", and the 2025 edition repeats it on fresh data: "Work Institute's latest analysis of over 120,000 exit interviews reveals 75% employee departures in 2025 were preventable." The firm's own site states it as a finding: "Research has shown that over 75% of the reasons employees leave are preventable."

What the number measures is the interesting part, and the report explains it.

How the figure is built

The 2017 report's methodology note describes the data: "Data from 242,453 exit interviews conducted between January 1, 2010 and December 31, 2016 were analyzed". Each interview's most important reason for leaving was coded into one of ten categories. Seven of the ten are then labelled more preventable: "Career Development, Job Characteristics, Well-being, Compensation & Benefits, Work Environment, Management Behavior and Work-life Balance". The other three are the less preventable set: "Less preventable reasons included the sum of interviews citing the following reason categories as most important: Involuntary, Retirement and Relocation".

So the calculation is: the share of leavers whose main stated reason falls in the first seven categories. Put the other way, it is 100% minus the share who retired, relocated or were let go.

Read the left column as an employer. A nurse who left for a 30% raise at a competitor is counted as preventable. Someone who left because the job involves night shifts is counted as preventable. Someone who left with a health condition is counted as preventable, under well-being. No one asked whether the employer could have matched the raise, removed the night shifts or cured the condition. The label is attached to the category, not to the case.

That is also why the figure barely moves from year to year. It is a property of the coding scheme. As long as retirement, relocation and dismissal account for about a quarter of stated reasons, the other three-quarters will be preventable by construction, whatever employers do.

Work Institute's own wording is more careful than the retelling. The report says more preventable and less preventable, and describes the first group as reasons related to modifiable conditions of the company, leadership, team and job. The word more is the first casualty when the figure is quoted.

What leavers say when you ask them

Gallup asked the question the figure sounds like it answers. Its 2019 article reported that "Fifty-two percent of voluntarily exiting employees say their manager or organization could have done something to prevent them from leaving their job." In 2024 it repeated the exercise on a named sample, "717 individuals who voluntarily left an employer in the past 12 months", surveyed in November 2023, and found that "42% of employees who voluntarily left their organization in the past year report that their manager or organization could have done something to prevent them from leaving their job".

That is a different construct: the leaver's own judgement, after the fact, of whether anything could have been done. It is not a measurement either, since people are not reliable narrators of counterfactuals about their own decisions, and the 2019 figure carries no stated sample. But it is at least the leaver's opinion rather than an analyst's category label, and it comes in at roughly half to two-thirds of the Work Institute number.

The same Gallup survey supplies the finding that actually deserves the attention the 75% gets: "45% of voluntary leavers report that neither a manager nor another leader proactively discussed their job satisfaction, performance or future with the organization" in their last three months. That is an observable fact about what employers did, not a judgement about what they might have done.

What the figure will and will not carry

It supports the claim that most stated reasons for leaving concern the job, the pay, the manager or the workplace rather than retirement or a move. That is true and useful, and it is roughly what the Gallup figures say too.

It will not support the claim that an employer could have kept three-quarters of its leavers, which is how it is used. Nothing in the data tests what the employer could have done. It will not support a program's projected savings, and it should not be multiplied by a turnover cost to produce a recoverable figure, which is exactly the calculation it is most often put to.

Where this lands on us

Our own diagnostic tells some respondents to expect largely preventable turnover until their biggest gaps are closed. That is a judgement about the factors the diagnostic scores, not a quotation of this figure, and it is worth being clear that it is a judgement. Our turnover cost calculator deliberately prints no recoverable share, because no published figure grounds one, and this page is part of the reason.

To be fair to Work Institute

The categories are published, the coding rule is published, the sample is large and the qualifier is in the report. Nothing was hidden. A figure that means most stated reasons concern things employers influence became most turnover is preventable in the retelling, and Work Institute's own marketing has since adopted the shorter version, which does not help.

If you want to know what share of your own turnover was preventable, the honest method is the expensive one: look at the cases, one at a time, and ask what it would have taken. The retention diagnostic is built to find the factors that make that question answerable.