TL;DR
- Core decision: Treat these as two distinct processes with two different conversation partners, not two variants of the same template.
- When doing nothing is right: If you can't act on what a stay interview would surface, the cheaper move is to wait until you can.
- What has to be true: Someone with budget and authority has to own what comes out of the room, or the conversation creates an obligation you can't meet.
- How the options split: The lever is whether you're extracting from a leaver or asking an active employee to help you see what they can't yet name.
- Decision rule: Run a stay interview only where you would act on the answer inside a quarter; run an exit interview wherever the role of the conversation is to learn, not to fix.
- Outcome to expect: A small reduction in duplicated paperwork, and a clearer sense of which signal belongs in which bucket.
The Moment the Templates Start to Look the Same
A people operations lead sits down on a Tuesday with two interview templates side by side. One is the exit script the company has used for six years. The other is a stay interview template she downloaded last week, after a director asked why the team keeps losing mid-performers with two years of tenure. The questions read almost identically: what would make you stay, what gets in your way, how is your relationship with your manager. She starts to merge them. Then she stops, because the person answering the first one has resigned, and the person answering the second one has not.
That gap is the whole article. An exit interview asks someone with nothing left to lose about a decision that's already made. A stay interview asks someone with everything still at stake about a decision they haven't made and may not have consciously considered. The templates can be similar, but the act of asking isn't, because what each person is risking by answering honestly is different, and what you may credibly promise in the room is different, and what you're allowed to do with the answer afterwards is different. The real issue isn't which questions to ask. The real issue is which conversation you're actually holding, and whether you've the authority to keep the promise that conversation implies.
When the Current Setup Is Genuinely Fine
Some companies should change nothing this year, and saying so out loud is part of why the rest of this piece is worth trusting.
Best tools for Employee Retention
The exit interview is producing signal you trust. A small company with a stable leadership team runs a thirty-minute exit conversation with a senior HR generalist. Notes are written, patterns are discussed in the monthly people review, and two changes in the last eighteen months can be traced back to what leavers said. The conversation isn't compulsory, the questions aren't threatening, and leavers treat it as a chance to be heard rather than a formality. There's no stay interview programme. Attrition isn't a problem. Under these conditions, introducing a parallel stay conversation adds process without adding learning. The right move is to leave the exit interview alone, formalise the pattern review so it survives the next HR generalist, and revisit only if attrition starts to move.
A stay interview would be a courtesy with no follow-through. A mid-sized employer is six months into a restructuring. Reporting lines are unsettled, some teams have been told their work will change, and leadership is honest that not every question can be answered yet. Running a stay interview programme in this window would be a way of looking like you're listening while the people in the room know you can't act. The honest move is to wait. Tell the team why. Name the constraint. Say when the conversation can happen for real. The cost of the pause is lower than the cost of asking people to surface concerns you'll then set aside because the business can't move on them yet.
You have one team leader who is the problem, and the data would be obvious. A manager is losing people. Exit interviews name them, in roughly the same words, often within weeks. Everyone in HR knows. A stay interview programme would, in effect, ask the rest of the team to name what the exits have already named. The right move is a manager conversation, a performance plan, possibly a transition. The stay interview is the wrong tool because the diagnostic is already done.
You're a small team and the conversation already happens. In a fifteen-person company, the founder has lunch with most people most weeks. The informal version of a stay interview is already running. Formalising it adds structure, but structure without a new action is overhead. The honest move is to keep the informal channel and to reserve a formal stay interview process for the day the team crosses a size where the founder can no longer hear everyone.
The Questions This Reader Asks at 11pm
Are we about to run the same conversation twice? Probably not in the questions, often in the spirit. The questions look similar because both are trying to learn what the employee thinks. The spirit is opposite because one person has resigned and one has not. The duplication risk is in the analysis afterwards, where the same theme can land in two spreadsheets and get reviewed twice. The cost of getting this wrong is that you under-act on a signal because two different processes both felt like they were doing something.
Who actually has the authority to act on a stay interview? The person who runs it. If the answer is "not me, and probably not anyone in this meeting", the conversation is theatre. The cost of getting this wrong is that you create an expectation you then break, and the next stay interview is the one where the employee stops being candid.
Can the line manager run a stay interview honestly? Sometimes. It depends on whether the line manager is the right person to hear the answer, which depends on whether the line manager is the problem. Where they're not the problem, the manager is the right person because they can act. Where they're part of the answer, the conversation has to be run by someone the employee can speak past the manager to. The cost of getting this wrong is that the most important signal in the room never gets said.
What can I promise the person in the chair? In an exit interview, the honest promise is that you'll listen, that nothing they say will change the outcome that's already set, and that patterns are taken seriously. In a stay interview, the honest promise is that what they raise will be looked at, that some of it can be acted on, and that some of it can't, and that you'll say which. The cost of getting this wrong is a broken promise. Broken promises in either conversation are remembered for years.
Do I have to keep the notes, and can the employee see them? The notes are personal data about a living individual, and in most places the individual has a right to see them on request. Treat the record accordingly. The cost of getting this wrong is a data subject access request that surfaces a working note someone wrote in a hurry and that you didn't mean to keep.
Three Honest Categories the Approaches Split Into
One process, two scripts, owned by HR. Many companies run a single interview programme where HR holds the script, holds the notes, and feeds the patterns up. The exit interview is older and more mature. The stay interview is bolted on, often after a director sees the term in an article. It works when HR has the standing to receive both signals, when the line is clean about which process is which, and when there's a quarterly pattern review that uses both. It fails when the two get merged into a single template because the questions are similar, and when the same person runs both without a clear separation of the promise being made. The risk is duplication at the data layer, where the same theme shows up twice and gets counted twice.
Split ownership, with line managers holding stay and HR holding exit. A second shape is to give the stay interview to the line manager, with HR providing training, question prompts, and a quarterly synthesis. The exit interview stays with HR because of the legal-adjacent nature of the conversation. It works when line managers are coached, when the synthesis is real, and when HR doesn't use the stay interview data to grade the manager. It fails when the line manager is the source of the friction and the employee can't say so in front of them. It also fails when HR doesn't have the resources to coach managers, and the stay interview becomes a box-tick that the manager runs once a year without preparing. The real example is a company where every people manager was given a half-day of training and a one-page prompt sheet, and the HR business partner attended every third conversation for calibration.
A shared light system, with no formal cadence in either direction. A third shape is to drop the formal templates and run both conversations as the moment demands: a stay conversation when something shifts, an exit conversation when someone leaves. It works in small teams where the relationships are already strong, where the founder or HR lead knows the temperature of the floor, and where there's no need to demonstrate process to a board. It fails as the company grows, because the informal version depends on the people in the room knowing each other well, and a thirty-person organisation can't rely on every manager being good at the conversation. The real example is a fifteen-person company where the founder schedules a walk with each person once a quarter, takes notes in a notebook, and reviews the notes when someone resigns.
Five Diagnostic Questions the Reader Can Self-Assess Against
Do I have authority to act on what I hear in a stay interview? Answer by drawing the last three stay interviews you've run, or by writing down what you would do if a senior engineer told you their real problem was a tool the team doesn't have. If you can't name the action, you don't have the authority.
Is my exit interview actually being read? Pull the notes from the last year. Count how many themes reappear. Check whether any of those themes led to a documented change. If nothing changed, the exit interview is a record, not a signal, and the right move is to fix the loop before adding a second process.
Is the line manager the right person to hold the stay interview for this team? For each team, ask whether the manager is the one the employee would speak to first if something was wrong. If yes, the manager can hold the conversation with HR support. If no, the conversation needs a third party in the room, and the manager needs to be told separately.
What happens to the notes, and who owns them? Write down the file location, the access list, and the retention rule. If the answer is unclear, the notes are at risk of being a data subject access request waiting to happen.
Can I keep the two conversations distinct in the meeting itself? Read the opening sentence of each script out loud. If they sound the same, a leaver and a stayer sitting in the same chair would be told the same thing. That's the duplication you're trying to remove, and it has to be fixed in the first ten seconds of the conversation.
Where the Two Diverge, Point by Point
Who Should Conduct It
In an exit interview, the safest choice is someone outside the leaver's reporting line, often an HR generalist, a people partner, or an external interviewer for senior roles. The leaver has nothing to gain by telling the truth to their manager about the manager, and the manager has a personal stake in not hearing it. In a stay interview, the line manager is often the right choice because the manager is the one who can act on what they hear, and because skipping the manager turns the conversation into a complaint channel. The weakness of the line manager running it's real: if the manager is part of the problem, the employee won't say so in front of them. The mitigation isn't to remove the manager from the conversation but to add a second channel the employee can use, and to make sure the manager knows the channel exists.
What Can Honestly Be Promised in the Room
In an exit interview, the only honest promise is that the leaver will be heard, that what they say will be taken into account, and that the outcome of their departure is already set. You can't promise to fix things for them. You can promise to take patterns seriously. In a stay interview, the promise goes further: that the question will be looked at, that some of it can be acted on, that some can't, and that you'll say which. The reason this is harder is that you're asking a current employee to spend emotional capital on a conversation that you may not be able to deliver on. The weakness is that the promise is harder to keep. A stay interview that produces no change has damaged trust in a way an exit interview never can, because the employee is still in the building and still watching.
The Reliability of What You Hear
Exit interviews are candid in a way stay interviews often aren't, because the leaver has less to lose by being honest. But exit interviews also distort: the leaver is rewriting their own story as they leave, attributing decisions to factors that flatter them, and sometimes punishing a manager they didn't like. The signal is real but it's shaped by the moment. Stay interviews are more fragile for the opposite reason: the stayer has everything to lose, and the answer is filtered through what the employee thinks is safe to say. The weakness of relying on either is that you're reading a single moment. Triangulate with other signals, including performance data, manager one-to-ones, and the shape of attrition over time.
What You May Do with the Answer
Exit interview data can be aggregated, used to inform retention strategy, and shared in pattern form with leadership. It should not be used to grade the line manager by name, and it should not be used to challenge a leaver after they have left. Stay interview data is more constrained: it refers to a current employee, the employee has a right to know what is recorded, and the data can't be used in a way the employee wouldn't recognise. The weakness in both is that aggregation hides the individual, and the individual is the one who took the risk of speaking. Design the reporting so the individual can see what was kept and what was dropped.
The Cadence
Exit interviews happen when someone leaves, which is event-driven and lumpy. Stay interviews are usually run on a cycle, often quarterly or half-yearly, and the cycle has to fit the rhythm of the team. The weakness of a fixed cycle is that the conversation becomes routine, and the employee stops preparing for it. The weakness of no cycle is that the conversation only happens when something is already wrong. A useful shape is a light cycle with an opt-in path for the employee to ask for the conversation off-cycle.
How the Two Data Sets Should and Should Not Be Combined
Both can feed a single pattern review, but the two signals aren't the same and should not be averaged into one score. A theme that shows up in both is high signal. A theme that shows up only in exits may be the moment of leaving, not the cause. A theme that shows up only in stays may be the thing people have learned to live with. The weakness of combining them is that the report can read as if the company has a single retention number, and the two distinct conversations get compressed into a single line in a slide. Resist that. Keep them separate, and let the pattern review name where they converge and where they don't.
The Decision Table
| Situation | Scale | Setup | Primary Pain | Recommended Starting Point |
|---|---|---|---|---|
| Stable team, low attrition, exit loop is working | Any | HR-led exit only | Risk of process for its own sake | Hold; formalise the pattern review |
| One manager losing people, pattern visible in exits | Team level | HR-led exit only | Diagnostic already done, conversation is the action | Manager conversation, not a new programme |
| Director asks for stay interviews after seeing the term elsewhere | Mid-sized | Exit process mature, no stay process yet | Authority to act not yet mapped | Map ownership first, then pilot one team |
| Restructuring in progress, answers cannot move | Mid-sized | Both processes paused | Acting later costs less than acting now | Pause both, name the constraint, set a date |
| Line managers are the right people to hear the answer | Team level | Split ownership, HRBP support | Risk of manager-as-problem going unspoken | Manager-led stay, HR-led exit, with escalation path |
| Founder-led informal conversations work today | Small | No formal process | Formalising too early breaks the signal | Keep informal, formalise when team size forces it |
| Multiple sites, exit notes inconsistent | Multi-site | HR-led exit only | Aggregation hides the site-level signal | Standardise the exit template, not the stay one |
| Stay interviews have run, no follow-through | Any | Manager-led stay, weak loop | Trust damage if continued | Pause stay, fix the loop, then resume |
Splitting the Question Set
| Question | Exit Interview | Stay Interview | Belongs in neither |
|---|---|---|---|
| What would have made you stay? | Yes, as a counterfactual | Not as posed, ask instead what would keep you here next year | – |
| What is the single thing that would make you leave? | Useful as a warning | Yes, directly | – |
| How is your relationship with your manager? | Yes, but with care on what is acted on | Yes, but only if the manager is not the listener | – |
| Do you feel fairly paid? | Often asked, often distorted by the moment | Yes, but be ready for it to become a conversation about pay | – |
| What would you change about the team? | Yes | Yes, with the warning that the team will see the notes | – |
| Are you happy you took this job? | Yes, and a useful reflection prompt | No, too far from the decision the employee is making | – |
| What do you plan to do next? | Often asked, usually not your business | No | Yes, for a current employee |
| What is your manager like under pressure? | Yes, candid signal expected | Only with a third party present | – |
| What did the recruitment process promise that was not delivered? | Yes, useful to feed back to talent | No | – |
| Are you willing to be contacted after you leave? | Yes, for alumni reference | No | – |
| What is one thing the company does that you wish it did not? | Yes | Yes | – |
| How is your workload right now? | Often the real answer, rarely the stated one | Yes, the most candid stay question | – |
When a Stay Interview Becomes a Negotiation
The moment a stay interview turns into a pay conversation, the manager needs to know what they're allowed to do next. If the manager is authorised to discuss pay in the meeting, the conversation can continue. If they're not, the right move is to thank the employee, say that pay is handled in a separate process with a separate script, and offer to escalate. What a manager should never do in a stay interview is name a number, agree to a change, or imply that the conversation is the route to a pay rise. The reason isn't that pay is a forbidden topic. The reason is that the moment the conversation is about pay, it isn't a stay interview any more, and the promise you made at the start of the meeting, that what they raise will be looked at, is now a promise about money, and the manager may not be the person who can keep it.
A second risk is the conditional. The employee says: I would stay if I could work from home two days a week. The manager agrees, and the policy says no. The conversation has created a commitment that the business can't honour, and the employee will know. The right move is to take the request away, to be honest about what can and can't move, and to come back with an answer. The wrong move is to agree in the room to keep the employee happy in the moment, and then discover the policy doesn't allow it afterwards.
| Manager's situation in the room | What they can do next | What they must not do |
|---|---|---|
| Employee names a pay figure | Listen, do not commit, escalate to HR or the comp owner | Agree to a number, even a tentative one |
| Employee names a working pattern | Listen, do not commit, take it to the policy owner | Agree off the record |
| Employee names a manager issue | Acknowledge, do not defend the manager in the room | Promise the manager will change |
| Employee names a workload issue | Commit to looking at it, with a timeline | Commit to a specific change in role or headcount |
| Employee names a career goal | Listen, take it to the career conversation framework | Promise a promotion |
| Employee asks for something you can actually give | Yes, in the room, and follow up in writing | – |
What to Put in Writing
The conversation is the part people remember. The record is the part that protects you.
| Artefact | Who owns it | When it is written | What it prevents |
|---|---|---|---|
| Exit interview notes, factual only | HR generalist | Within two working days of the conversation | A data subject access request surfacing opinion, hearsay, or unverified claims |
| Exit interview pattern summary, no names | HR leadership or people analytics | Monthly or quarterly | The same theme being lost in individual notes and never rising to leadership |
| Stay interview prompt sheet, current | HRBP or people ops | Reviewed each cycle, updated when questions drift | A manager running an out-of-date version without realising |
| Stay interview notes, factual only | Line manager, with HRBP access on request | Within two working days | A drift from the data subject's own recollection six months later |
| Stay interview consent note | Line manager or HRBP | At the start of the conversation | A complaint that the employee did not know what was being recorded |
| Stay interview action log, by team | Line manager, reviewed with HRBP | After each cycle | The pattern of stay conversations producing no change, and the trust damage that follows |
| Combined pattern review, exits and stays | HR leadership | Quarterly, with a fixed template | The two data sets being merged into a single misleading average |
| Process boundary document | HR leadership | Once, reviewed annually | A stay interview being run by the wrong person because the boundary is not written down |
| Retention rule for interview records | HR or compliance | Once, applied from the first record held | Records kept forever, or destroyed too early, both of which carry exposure |
| Training record for any manager running stay interviews | L&D or HRBP | After each training session | A manager running a stay interview without preparation and producing damage that could have been prevented |
Questions to Ask Before You Commit
Authority. Who in this team has the standing to act on what a stay interview surfaces, and what can they actually do? A bad answer is "we will see what comes up and then decide". A good answer names a person, a remit, and a budget.
Boundary. Where does the exit interview end and the stay interview begin, and who enforces that line in the moment? A bad answer is "we will use the same script". A good answer is a written boundary, a separate template, and a named owner of the distinction.
Loop. What is the smallest action a stay interview can produce inside a quarter, and what is the smallest action an exit interview can produce inside a month? A bad answer is "we will review the themes". A good answer names the action, the owner, and the date it will be checked.
Records. Where do the notes live, who can see them, and what is the retention rule? A bad answer is "in the manager's inbox". A good answer is a defined system, a defined access list, and a defined rule that you can show to a data subject access request handler.
Manager readiness. Which line managers are ready to run a stay interview, and which need training first? A bad answer is "they're all adults, they will figure it out". A good answer is a list, a training plan, and a date by which the unprepared managers will be ready or out of scope.
Scope. Which team is the pilot, and what does success look like? A bad answer is "we will roll it out and see". A good answer is one team, one cycle, a small number of success signals, and a go or no-go decision written in advance.
Signal integrity. How will the two data sets be reported separately and together? A bad answer is "in the same slide". A good answer is a fixed pattern review template with the two kept distinct, and a note of where they converge.
Failure path. What do you do if a stay interview produces an answer you can't act on? A bad answer is "we will handle it if it happens". A good answer is a written rule: thank the employee, take the answer away, be honest about what can and can't move, and follow up in writing.
The Cost of Getting This Wrong
The visible cost is wasted time. Two templates, two sets of notes, two pattern reviews, two meetings a year per employee, and a manager who has to prepare for both. Over a year, in a mid-sized company, that's real hours, and the hours show up on a timesheet.
The cost that doesn't show on the invoice is trust. A stay interview that surfaces a problem and produces no change is worse than no stay interview at all, because the employee took a risk and watched nothing happen. So the cost of running a stay interview programme without the authority to act isn't the cost of the programme. It's the cost of the next stay interview, which will be the one where the employee stops being candid, and the one after that, where the employee has already decided to leave and the company finds out in the exit interview instead.
So the question to put to the room isn't whether you can afford to run both conversations. It's whether you can afford to run a stay conversation that doesn't lead to an action you can name in advance. If the answer is no, the right move is to fix the loop before you start the conversation, and to be honest with the team about why you're waiting.
When You Are Ready to Go Further
If the two conversations are running but the loop is weak, the next step is usually to look at how other organisations in your sector have set the boundary, and to compare the shape of the stay interview programmes you're considering. HROpsLab publishes independent comparison work on HR operations tooling and on the practice patterns that sit underneath it. We don't sell software, we don't run payroll, and we don't provide advice. What we do is read the documentation, watch the demos, and write up what we find, so that the next conversation you've with a provider or with your own team starts from a clearer picture.
If you would like to see how we approach that work, the comparison library is open and the editorial standards are published. There's no purchase, no sign-up gate, and no follow-up call. Read what is useful, leave the rest.
Frequently Asked Questions
What is a stay interview, in one sentence?
A stay interview is a structured conversation with a current employee, run on a cycle, designed to surface what would keep them at the company and what might push them out, before the decision to leave has been made. It isn't a performance review, it isn't a compensation conversation, and it isn't a substitute for the regular one-to-one between the employee and their manager.
Who should run an exit interview?
The exit interview is usually run by someone outside the leaver's reporting line, most often an HR generalist or a people business partner, because the leaver needs to be able to speak about their manager without the manager in the room. For senior leavers, an external interviewer is often the right choice, because the closer the conversation gets to the leadership team, the harder it's for a leaver to be candid with anyone inside the company.
Who should run a stay interview?
The stay interview is usually run by the line manager, because the line manager is the person most likely to be able to act on what they hear, and because the conversation is one of the ways a manager learns what the team actually needs. Where the line manager is the source of the friction, the conversation has to be run by someone the employee can speak past the manager to, and the manager needs to be told separately.
How often should stay interviews happen?
The honest answer is that there's no single right cadence, and the cadence that fits a team is the one the team can prepare for and act on. In practice, quarterly or half-yearly is common, and the cycle has to be long enough that the conversation has something new to hear and short enough that the manager can still act on it. The cadence should also be flexible enough that an employee can ask for the conversation off-cycle if something has shifted.
Should the two conversations use the same questions?
No, not exactly. The questions can be similar in shape, but the wording has to reflect the position of the person in the chair. A leaver can be asked what would have made them stay, because the counterfactual is honest. A stayer should be asked what would keep them here next year, because the decision is forward, not backward. The temptation to merge the two is the duplication this article is trying to prevent, and the merge usually shows up first in the questions.
What do you do if the employee asks for more money in a stay interview?
The manager should listen, should not commit to a number in the room, and should take the request to the person who owns compensation for that role. The conversation is still useful, because pay is one of the things the employee is telling you, and the rest of what they say is also signal. The line to hold is that pay is handled in a separate process with a separate script, and that this conversation is about the broader picture, not the number.
Can line managers be trusted to run stay interviews honestly?
Sometimes, and the answer depends on whether the manager is the right person to hear what the employee would say, and whether the manager has been trained to hold the conversation without becoming defensive. Where the answer is yes, the manager is the right choice because they can act. Where the answer is no, the conversation has to be run by someone else, and the manager has to be developed or moved. Trusting a manager to run the conversation without checking is the most common way the programme produces more noise than signal.
How should the two data sets be combined?
They should be reported in the same pattern review, but kept distinct in the analysis. A theme that shows up in both exit and stay interviews is high signal. A theme that shows up in only one of the two is signal of a different kind, and the difference is the point. Combining them into a single retention number is the most common way the two conversations get compressed into something that looks like evidence but isn't, and the right move is to keep them separate in the data and let the pattern review name where they converge.
HROpsLab is an independent review publication for HR operations teams. We don't sell software, payroll or advisory services. We publish what we find and we say so.
