What the Exit Interview Tells You (If You Run It Right)

"We need to talk" is the scariest four-word sentence a manager hears, and it almost never comes before good news. Someone you rely on has a foot out the door, and no amount of "but why" is putting it back. In this industry two or three years in a seat is normal and a single year isn't rare, so when resignation letters start stacking up, that's not bad luck. That's data, and the exit interview is how you read it.

It's also the cheapest honest feedback you will ever get, and why it's honest is the whole game. The person is candid for one reason: they no longer need anything from you. The power that normally sits between a manager and a report, over their next review, their comp, their projects, has evaporated. They've signed somewhere else. There's nothing left to protect and nothing to angle for, and into that vacuum comes the truth you could never extract while you held the leverage. Most companies I worked for didn't run these, or didn't care. I ran them anyway, because the feedback was worth more than the org's indifference to it.

Run it like the gift it is

A few hard-won rules.

Timing: not the last day. By then the person has mentally moved out and repainted the walls, and just wants their laptop collected and their badge killed. Do it a week or two after the announcement, once the emotional weather has cleared but they're still around and engaged. That's the window where you get reflection rather than tears or a shrug.

Don't run it yourself if you're the subject. If a developer is leaving a team I manage, I'm exactly the wrong person to extract honest feedback about that team, because the leverage hasn't fully evaporated: I'm still a reference, still a node in a small industry's network, still someone they might want something from in three years. So they'll be polite to my face on reflex. A peer or my own manager gets the real answers. HR gets them if the problem might be me.

Shut up and don't fix anything. The instinct is to defend, to explain, to solve the thing they're raising. Kill it. Every minute you spend justifying the past is a minute you're not learning, and it tells them their feedback is being argued with, so they stop giving it. You are not there to win. You are there to listen and write it down.

And it's rarely the money, even when they tell you it's the money. I've had people paid under market stay for years, and others leave for reasons no raise would touch, like a review they never forgave. Compensation is just the most socially acceptable thing to write on the form. The real reason under the stated one is usually running out of anywhere to grow.

The most useful question I steal from Everett Spain and Boris Groysberg: ask them to finish the sentence "I don't know why this company doesn't just ______." People are startlingly candid filling that blank, and they hand you the obvious fix you've gone face-blind to.

The thing I actually learned from one

The exit interview that changed how I run teams wasn't about pay or growth. It was a quiet developer, the only fully-remote person on a co-located team, and when I asked what one thing he'd change, the whole story came out. He'd spent a year as a ghost haunting his own project. The team did its real thinking in hallway huddles and at adjacent desks, decisions half-made before they reached a channel he could see, and the written updates meant to include him were an afterthought when anyone remembered. He wasn't bitter. That was the part that landed hardest. He'd simply concluded, reasonably, that he was peripheral, started behaving like it, and was leaving for somewhere he'd be in the room. None of it had ever surfaced while he worked there, because the cost of raising it felt higher than the cost of quietly leaving. His exit interview became a blueprint: default-written decisions, remote-first meeting hygiene, a rule that if one person's dialed in, everyone's dialed in. I learned more about how my team actually operated from one departing person with nothing to lose than from a year of standups with everyone who stayed.

The conversation you should have a year earlier

Which points at the real lesson, the one Spain and Groysberg put plainly: the exit interview should be the last conversation, not the first. If the only time you ask what would make someone stay is on their way out, you've run the diagnostic after the patient already left the building. So I do the reverse, on a schedule, and it's harder for the exact reason the exit interview is easy: the power is still in the room. You have to manufacture the safety a departure hands you automatically, by asking early, asking often, and proving over and over that the honest answer won't be used against them. Get that right and the best exit interview becomes the one you never have to book, because you asked the question a year early and then, crucially, did something about the answer.