The first exit interview I ran took twenty minutes and produced nothing.
She had been with us two years, she was good, and she was going to a competitor. I asked why she was leaving. She said it was a great opportunity and the timing was right. I asked whether there was anything we could have done. She said no, honestly, she had loved working here. I asked for feedback on my management. She said she had learned a lot from me.
Six weeks later a colleague told me, in passing and without meaning to land a blow, that she had been trying to move off a specific project for eight months and had given up asking.
Nothing in that exit interview was a lie. It was just the polite version, delivered by someone with no remaining reason to give me the other one. And the information I needed had been available, to me, for eight months.
That is the honest starting point. The exit interview is a compromised instrument, and whether you get anything out of it depends on something settled long before the resignation: whether trust was already there. If your team does not tell you difficult things while they still work for you, the exit conversation is not where that begins. Understanding exactly how the instrument is compromised is what lets you extract anything from it at all.
Here is how to run one anyway, and what to use instead.
Three reasons exit interviews under-deliver
The timing is wrong by design. By the time this conversation happens, the decision is made, the offer is signed, and the start date is set. You are asking for diagnostic information at the exact moment it can no longer be acted on for this person. Everything they say is now archaeology.
The incentive points the wrong way. A departing employee has one live interest in the room: leaving on good terms. They may want a reference. They may work with you again in a small industry. They will almost certainly be asked about you by someone. Candor costs them something and buys them nothing, which is not a character flaw, it is arithmetic.
The interviewer is often part of the answer. Research on manager quality keeps finding that the direct manager is one of the largest factors in whether someone stays. Which means in a meaningful share of departures, the person asking “why are you leaving?” is a significant part of the reason, and both people in the room know it. Nobody says that out loud in a twenty minute meeting.
None of this makes the exercise worthless. Harvard Business Review’s review of the practice, Making Exit Interviews Count by Everett Spain and Boris Groysberg, argues the opposite: done deliberately, exit interviews surface problems leadership cannot otherwise see and turn departing people into long-term advocates. Their central caveat is the one most companies ignore, and it is about what happens to the answers rather than what happens in the room. More on that below.
Four decisions to make before you ask anything
Who runs it. If you are the direct manager, you will get a softer version than someone one step removed. If you can have HR or your own manager run it and share themes rather than transcripts, do that. If you are a small company and it has to be you, say so explicitly at the start: “I know I am the wrong person to ask this, because I am part of what you are evaluating. I would still rather hear it than not.”
That sentence does more work than any question on the list below.
When you run it. Not the last day. On the last day people are collecting laptops, saying goodbyes, and thinking about Monday. Run it roughly a week before the end, when the goodwill is intact and the pressure is off.
What you can promise. Decide in advance, and say it out loud: what you will do with the answers, who sees them, and whether anything is attributable. Then keep to it exactly. If you promise themes-only and then quote them to a colleague, you have taught everyone still working for you what your promises are worth. This is the single fastest way to make every future exit interview useless.
What format. A conversation gets you nuance and loses you honesty. A written form gets you the reverse. If the departure was difficult, a form is often the better instrument, because it removes the social pressure to reassure you in real time.
The 12 questions
Grouped by what they are actually trying to surface. You will not need all twelve in every conversation.
The decision itself
1. When did you first start thinking about leaving?
The most useful question on the list, and almost nobody asks it. The answer is a date, and that date tells you how long the problem was live before you noticed. If the answer is eleven months, you have a detection problem, not a retention problem.
2. What was actually happening around that time?
The trigger is rarely the reason people give. “A recruiter called” is what happened. What was happening around it is why the call worked.
3. Was there a point where something specific would have changed your mind?
Ask for the specific thing and the specific moment. “More money” is a deflection. “If the project reassignment had happened in March when I asked” is information you can act on for somebody else.
The work
4. What part of the job drained you most, and was that visible to me?
The second half is the important half. Plenty of jobs contain draining work. The question is whether the manager knew.
5. What did you spend time on that you thought was pointless?
Departing people will answer this honestly more often than any other question here, because it criticizes a process rather than a person. It is frequently the highest-yield question in the whole set.
6. What would you protect if you were me? What should not change after you go?
Balances the conversation, and it surfaces things you did not know were load-bearing. Sometimes you find out that an informal thing one person quietly maintained is about to disappear with them.
Us, as seen from inside
7. What do we do badly that we seem not to know about?
The phrasing matters. “Seem not to know about” gives permission to say something you might find obvious or embarrassing, without implying you are stupid for missing it.
8. Who on this team is better than we realize?
An unexpectedly generous question that people answer with enthusiasm, and one of the few pieces of exit information you can act on immediately. Departing colleagues have seen work you have not.
9. If a friend asked you whether to take a job here, what would you actually tell them?
The “actually” is doing the work. This gets past the official version faster than asking about culture, because most people have already had this conversation in their head.
The honest close
10. What is the story you will tell about your time here in a year?
Distance makes candor easier. People will say things about the past tense they will not say about the present one.
11. Under what circumstances would you come back?
A real question, not a courtesy. Boomerang hires are cheap and fast, and the conditions someone names are a precise description of what is broken.
12. What have I not asked that I should have?
Always last, always worth it. The remaining thing is usually the actual reason.
When you get the polite version
You will. Here is what to do.
Do not fill the silence. The single most common mistake is rescuing someone from a pause by offering them an easier answer. Ask, then wait. Four seconds feels like a minute and produces more than any follow-up question.
Trade specificity for specificity. Vague questions get vague answers. If “how was the workload?” produces “fine, busy,” try “which week in the last six months was the worst, and what happened that week?”
Name the dynamic. “I am aware you have a reference to protect and nothing to gain from being blunt with me. I am asking anyway, and nothing you say changes what I write.” Saying the awkward thing out loud sometimes dissolves it.
Accept partial answers. Someone who tells you one true thing out of twelve questions has given you more than someone who was warm for thirty minutes. If the conversation keeps coming back to the same topic while avoiding the specifics, that topic is the answer. Learning to hear what the “fine” is covering is a general skill, and it applies long before anyone resigns.
What to do with the answers
This is the part almost everyone skips, and it is where the HBR argument actually bites. A single exit interview is an anecdote. Its value is close to zero, and acting on one departing person’s strongest complaint is how teams get whiplashed by policy changes that solve nothing.
The unit of analysis is the pattern, not the person. Keep a simple private record: date, role, tenure, when they say the thinking started, the two or three themes. After four or five departures, read the file in one sitting. That is the moment the information becomes usable, and it is the only moment.
What you are looking for:
- The same theme from people who never worked together. One person complaining about a process is a preference. Three unconnected people naming it is a defect.
- The clock. If several people say the thinking started around the same event, that event is worth understanding, whether or not anyone named it as their reason.
- Departures clustering in one part of the team. Repeated exits from one area with one common manager is the loudest signal in the whole file, and the hardest to look at squarely if the manager is you.
- Reasons that keep being pay when the pay is competitive. Money is the socially safe explanation. When it appears repeatedly against market-rate salaries, it usually means something else, and the raise conversation you have been avoiding is worth reading for the version of this that is genuinely about money.
If the file starts describing the environment rather than individual grievances, compare it against the twelve signs of a toxic work environment before you conclude you have simply had bad luck with hires.
The instrument you should actually be using
Everything above is damage limitation. The real fix is unglamorous: ask most of these questions six months earlier, of people who are not leaving.
A stay interview is a twenty minute conversation, once or twice a year, separate from performance and separate from your regular one-on-ones, with roughly this shape:
- What would make you think about leaving?
- What is the most frustrating part of your week right now?
- What are you learning here? What have you stopped learning?
- If you could change one thing about how this team works, what would it be?
- What would make the next year here clearly better than the last one?
Same information. Radically better timing. And the questions carry a signal on top of their content, because asking someone what would make them leave, while they still work for you, is itself an act of the kind of candor that makes people stay.
The cost comparison is not close. Replacing someone competent runs into serious money once you count recruitment, ramp time and the work that does not happen in between, and SHRM’s work on losing key employees is blunt about how badly most organizations underestimate it. Put your own numbers into the employee turnover cost calculator and compare the result to forty minutes a year per person. If you would rather see the risk before it becomes a cost, the flight risk calculator weights the loss by how likely your strongest person actually is to leave, which is the version that argues for having the conversation this quarter.
The uncomfortable summary
Run exit interviews. Run them a week early, ideally not yourself, with a promise you keep, and read them in batches rather than one at a time.
But do not mistake them for a retention tool. By the time you are asking why someone is leaving, the only person the answers can help is the next one. The manager who learns something genuinely useful in an exit interview has usually just discovered how long they were not being told.
If you want to know which of the people still on your team is closest to that conversation, the Am I About to Lose My Best Employee? assessment walks through the early signals in about four minutes. It is a better use of twenty minutes than any exit interview you will ever run.