Do Exit Interviews Work? What the Evidence Says

Evidence grade: Limited. Some evidence exists, but the designs are weak, the populations narrow, or the outcome measured is not turnover. This grade describes the published evidence. It is not a prediction about your workforce and not a result RetainScore has produced.

Two claims travel together, and only one of them is defensible

The first claim is that exit interviews collect useful information. The second is that collecting it reduces turnover. The second claim is the one that justifies the budget and the calendar time, and it is the one with almost nothing behind it. We looked for a controlled study showing that running exit interviews lowers turnover and did not find one.

What the research actually documents

Robyn Johns of the University of Technology Sydney examined the exit interview process at a professional publishing organization of roughly 700 full-time equivalent staff, observing interviews as they happened and interviewing both departing employees and the HR personnel running the process. Her conclusion was that "despite the accolades awarded to exit interviews they may not be as effective in practice as we have been lead to believe."

Two failure modes came out of that work. They are independent, which means fixing one leaves the other in place.

Departing employees manage the conversation

Every departing employee in the study protected the relationship rather than reporting accurately. Johns records that "all of the employees interviewed essentially followed the general principle of making sure that they didn't burn any bridges behind them." The stated reasons were consistent: the need for a future reference, the possibility of returning to the employer later, and concern about consequences for the colleagues staying behind.

This is worse than noise. Noise averages out. A bias that runs consistently away from managers and working conditions, and toward reasons that are safe to say out loud, does not average out. It is a plausible reason why pay shows up so often as the stated cause of a departure and so rarely survives closer examination.

Nobody acts on what gets collected

The second failure is administrative rather than psychological. Johns found that "actual changes in organisational policies and/or work procedures as a direct result of the information obtained, was negligible," and attributed that to a lack of capability in analyzing the data and presenting it to senior decision makers. Interviews were conducted, filed, and reduced to a monthly count of primary and secondary reasons. Nothing downstream moved.

What would have to be true for this to work

• The answer has to be honest. That argues for separating the conversation from the reference relationship: a third party, a delay of some weeks after the departure, or genuine anonymity.
• Someone has to own the analysis. A tally of stated reasons is not analysis. Reading those reasons against tenure, team, and manager is.
• The finding has to reach a decision. An exit interview program with no route to someone who can change something is a filing exercise with a calendar invite.

The honest boundary

The Johns study is a qualitative case study and the author is explicit about its limits: it "only examined ten employees and two human resource personnel at one organisation, so the conclusions should be limited" to that population. Ten people at one employer cannot settle a question this size. What a study like that can do is document a mechanism, and Johns places the mechanism she documents alongside earlier work on response distortion in exit interviews.

Read the grade correctly. Limited evidence is not evidence of no effect. It means the case has not been made, and that anyone telling you exit interviews reduce turnover is claiming more than the literature supports.

If the exit interview is your only instrument

The deeper problem is timing. An exit interview asks the question after the decision is final, of a person with every incentive to be gracious. If you want to know why people leave while you can still do something about it, the measurement has to happen earlier and the answers have to be safe to give. That is a diagnosis problem, and diagnosis is the part most retention advice skips.

Work through the retention diagnostic to see which factors the research finds predictive, or read the companion grade on whether stay interviews work.