Exit Interviews: What to Ask and What to Do With the Answers

The exit interview is one of the most widespread practices in HR and one of the least productive.

Everett Spain and Boris Groysberg published in Harvard Business Review a review of exit interview programs at 210 organizations across 33 industries. The finding was that very little positive organizational change came out of that feedback. Their work also reports that roughly three in four companies run exit interviews, and that fewer than a third of executives could name a specific action taken because of one.

So the data gets collected and does not get used. There are design reasons behind that.

Why the answer you get is soft

The person leaving has three incentives pushing toward the safe answer. They need the reference. They do not want a fight in their last week. And they suspect whatever they say will reach their boss.

With those three incentives, the most likely answer is “an opportunity I couldn’t pass up.” It is polite, partly true, and useless.

Three format decisions change the quality of what you hear.

Who asks. Never the direct manager, and preferably nobody involved in the exit. Someone two levels up, or from another function, gets different material.

When. Not the last day. The person is already mentally gone and the conversation is a formality. The second-to-last week works better, and a follow-up three months after departure often produces the most honest material of the whole process.

How the use is declared. Say up front that answers are aggregated and reported without names, then hold to it. One leak destroys the entire program permanently.

The eight questions that produce information

The generic list — “what did you like and what didn’t you?” — produces generic answers. These point at concrete decisions.

1. When did you start looking? The date matters more than the reason. It is almost always months before the resignation, and that is the moment where something could still have been done.

2. What happened around then? The trigger event. A promotion that did not come, a change of manager, a project that got cancelled.

3. At that point, what would have made you stay? The answer is rarely just money. And when it is, money is usually a signal for something else.

4. Which part of the job was different from what you were told during hiring? This question evaluates your hiring, not your retention. Repeated gaps mean the posted profile does not describe the actual role.

5. What does your new employer do that we don’t? It is the most concrete question on the list and the one most often skipped.

6. Did you tell anyone here, and what happened? It separates a communication problem from a listening problem. If they said it and nothing happened, the defect is not in the channels.

7. Would you recommend this place to someone you respect? What kind of person does it work well for? The second half is what gives you something usable.

8. What should change before the next person takes your seat? It turns a complaint into an operable recommendation.

What not to ask

One-to-ten satisfaction scales. They produce an average you cannot act on.

“Are you leaving over pay?” It is a closed question that offers a comfortable exit, and nearly everyone takes it.

Anything about other people by name. It turns the conversation into a trial and blocks the rest of the answers.

What to do with what you hear

This is the stretch where the program collapses, according to Spain and Groysberg’s own work. Three conditions hold it up.

An owner. A named person who reads every interview in the quarter and produces a synthesis. Without an owner, the notes pile up in a folder.

Aggregation, not anecdote. One exit is noise. Four exits from the same team in six months, with the same cause, is data. How to split the number is in how to calculate employee turnover.

A cross with regretted turnover. Exits you wanted to prevent weigh differently from the ones you did not. Treating both the same dilutes the signal.

And a minimum commitment back: a quarterly synthesis to the leadership team, with what was heard and what was decided. Even when the decision is to do nothing.

The structural limit

The exit interview arrives late by definition and surveys a biased sample, which is the people who already left. You will never learn from it what the person still inside, who just started looking, is thinking.

That is why the highest-yield complement is the stay interview, run with the people you want to keep and before anything happens. Four questions are enough. What would make you leave. What keeps you here. Which part of your job would you do more of if you could. What frustrates you most in a given week.

Same information, two years earlier and still in time.

And the cut almost nobody makes

If most of your exits happen before the one-year mark, the exit interview is the wrong instrument. You are looking at an entrance problem with a retention tool.

There the lever is measuring fit against the role and against the culture before the offer, not asking better questions on the way out. That distinction is developed in culture and retention.

If you fix one thing in your current program, change who runs the conversation. It is free, and it moves the quality of the answers more than anything else.