TL;DR
- The design job: a small set of identical questions that produce comparable data, plus deliberate protection of the space where the form stops and the interviewer's judgment begins.
- When not acting is fine: a single site or team of under fifty leavers a year, no reporting need, and a manager who is already trusted to run the conversation.
- What must be true: someone owns the template, that owner can remove questions, and the document is reviewed at a fixed cadence even when nothing is wrong.
- How options split: by how much of the conversation is scripted, who writes the form, and whether the artefact lives in the leaver's record or only in a roll-up.
- Decision rule: standardise only what you intend to compare. Anything that exists to surface something you didn't expect must remain open.
- Outcome to expect: fewer comparable numbers in year one, better ones in year two, and a process a manager can't quietly bend to the answer they want to hear.
The Form That Ate the Conversation
A compensation lead in a 400-person SaaS company sits down on a Tuesday morning to redesign the exit interview template she inherited. The previous version has 31 questions. Two of them are about parking. One asks the leaver to rank nine factors of dissatisfaction on a five-point scale. Another has a free-text box underneath it labelled "Comments (optional)". The HR director wants the roll-up back in Power BI. The line managers want a script they can follow without thinking. And the compensation lead, who has run more of these conversations than either of them, knows that the most important thing an exit interview ever produced in her last role was a single sentence the form didn't ask for, about a promotion that had been promised in March and never came back in writing.
She opens a blank document and types the word "Standardisation" at the top. Underneath, she writes "What do we actually need the same across every leaver?" Then she stops, because the question she is really trying to answer isn't what to standardise. It's what to leave open. The form she is building will outlive her. Every fixed question she adds today is a decision about which signals her successor is allowed to ignore tomorrow. That's the real design job, and it's harder than it looks.
But the temptation, when you inherit a bloated template, is to start cutting. That instinct is usually right, and it's also where most redesigns go wrong. Cutting feels productive. Cutting is also the easy half. The hard half is deciding what to protect: the shape of the conversation, the room for the sentence nobody saw coming, and the governance that prevents the next person from quietly filling the gaps you left with their own list of questions.
Best tools for Employee Retention
When You Do Not Need to Act Yet
Most HR leads who read an article like this one don't need to redesign their template this quarter. They need to know whether they're in the group that does. The four stages below are written for that. If the first two describe your organisation, this article is useful background. If the third or fourth does, it's urgent.
Your current setup is genuinely fine. You're a single business unit or a single site with a handful of leavers a month. Picture the marketing team of a regional retailer: one HR lead, twelve people who've worked together for years, and a leaver every other month who walks out with a handshake and a coffee. Your HR lead has run most of the conversations personally for years and remembers what people said. The "template" is a one-pager with three questions and a blank page. Nothing needs to change. The moment you replace a working conversation with a form, you make the conversation worse and the data flatter, and you do it for an audience that, right now, is just you. Leave it alone until you've a second site, a second HR generalist, or a leadership team that has started asking for the numbers and you've discovered you can't produce them.
You're starting to feel friction. Picture an accountancy firm that opened a second office eighteen months ago. You now have two managers, or two sites, or one HR generalist who has changed in the last year, and the answers you get back aren't comparable. One manager's exit notes read like a resignation letter. Another's read like a Post-it. You don't yet disagree about what to do, but you're losing the thread between conversations. This is the stage where a light template earns its keep. A short fixed spine, four or five questions identical across the company, will restore comparability without turning the conversation into a survey. The cost of not acting here is that the next time leadership asks, you'll have a spreadsheet of impressions instead of evidence, and the impression that wins will be the one that was written down first.
There's now real risk. Picture a logistics company whose board has asked for the third time why the night-shift operation keeps losing drivers. Leadership has started asking why voluntary turnover in one team is consistently higher than in another, and the answers you can produce are anecdotes. You have a leaver who has asked, in writing, for a copy of what was said about them in their exit conversation. You have a manager who has quietly added their own questions to the form for a departing report they didn't like. The risk is no longer theoretical. It's the moment a defensible template, with a clear owner and a review cadence, becomes a piece of evidence that you take the process seriously. Without it, the same board will keep asking the same question, and the next answer you give will be the one you can't back up.
The edge case. You operate in a sector or a jurisdiction where the line between an exit conversation and a grievance record is thin, and a former employee has indicated they may need to rely on what was said at the point of departure. Picture a regulated services firm where the leaver has flagged, before leaving, that they intend to bring a claim. At this point, what you need isn't a better template. It's local advice on what the conversation is allowed to be, who is allowed to be in the room, and what must be written down versus what should not be recorded at all. Confirm the legal shape before you change the form. A template designed for honest feedback and a record designed to be evidence are not the same artefact, and trying to build one that does both jobs is how defensible processes become indefensible ones.
Five Questions This Reader Asks at 11pm
How long should the template actually be? Short enough that a tired manager can hold all of it in their head. That usually means a single page. If you can't fit it on a page, the second page is the part you should cut. The reasoning is that comparable data comes from a small set of identical questions, and the marginal question past the tenth one starts costing you signal from the first nine, because the interviewer's attention is finite and the leaver's patience is shorter than the form designer assumes. If you get the length wrong, the form stops being a prompt and becomes a script, and the conversation starts to feel like a survey, and the leaver gives the answers they think will end the meeting fastest.
Should the questions be rating scales or open text? Both, and never on the same question. A scale gives you comparability; open text gives you the thing the scale can't see. A scale on its own produces numbers that look authoritative and mean nothing. Open text on its own produces paragraphs you can't roll up. Use each where it earns its place, and never use a scale to make a question feel easier to answer. The reasoning is that scales measure the gap between what the leaver is willing to score and what they're willing to say, and open text measures what they would say if you gave them the room. If you get the mix wrong, you either trust a number that doesn't reflect the situation, or you drown in a corpus nobody has time to read.
Who owns the template? Someone with the authority to remove a question, not just to add one. That's usually a head of HR or a people operations lead, not a committee. The single biggest failure mode in template governance is a form everyone can amend and nobody is responsible for. The reasoning is that adding a question is a one-off cost borne by the next interviewer, while removing a question is a small confrontation with whoever added it, and organisations avoid small confrontations reliably. If you get this wrong, the form grows on every cycle and shrinks on none, and within a year you have a thirty-one-question document that asks about parking.
What do we do when a manager wants to add their own questions? You give them a space below the fixed spine, marked clearly as optional, and you make sure the answers to their questions never enter the same roll-up as the fixed ones. If their questions are good, they migrate to the fixed spine at the next review. If they're not, they die in the optional box without contaminating your comparability. The reasoning is that a manager with skin in the conversation often knows something the central template doesn't, and forcing them back into a fixed form is how you lose the next useful piece of evidence. If you get this wrong, the manager adds the question to the spine anyway, the roll-up shifts, and your comparability for that quarter is broken in a way nobody can spot until much later.
How do we stop the form from growing every quarter? You schedule a review. Not a renewal, a review. The review asks one question of every question on the form: what decision did this question change in the last twelve months? Any question whose answer is "none" gets cut, even if it's the one the HR director added last spring. The reasoning is that templates decay by accretion and only the discipline of removal keeps them honest. If you get this wrong, the form becomes a graveyard of well-intentioned questions, and the next redesign will be twice as long and twice as painful as it needed to be.
Three Honest Categories the Approaches Split Into
The fully scripted interview. Every word the interviewer says is on the form, in order, with prompts underneath. Imagine a call-centre operation running fifty exits a quarter, where the interviewer is a different person every time and the leaver may later need to demonstrate that the conversation was conducted in a particular way. The strength is consistency. Every leaver hears the same opening, the same transitions, the same close. The manager doesn't have to think, which is also the weakness. A scripted interviewer stops listening at the point the leaver says something the script doesn't have a box for. In practice, the form becomes a ceiling rather than a floor. This approach is right where the interviewer pool is large and untrained, where the conversation may later be reviewed, and where the only thing the organisation can guarantee is that the words used were the words approved. It fails the moment the conversation needs to go somewhere the script doesn't, and the leaver notices that the interviewer has stopped being a person and become a reader.
The semi-structured template. A small fixed spine of identical questions, usually four to six, plus a defined list of optional prompts the interviewer can use or set aside. The spine gives you comparability. The prompts give you the room to follow what the leaver actually said. Imagine a mid-sized software company with one HR business partner who runs most exits, and managers who sit in occasionally and have their own questions about the team. This is the shape most mature programmes settle into after one or two redesigns. It's right when you've an HR function that runs most exits and managers who occasionally sit in. It fails when the optional prompts quietly harden into a second script, which they will, unless someone reviews the notes. The signal you're capturing is the consistent part of the conversation. The signal you lose is the part the prompts never anticipated, and over time the prompts become a self-fulfilling list of the things the team expects to hear.
The free-form conversation with a post-hoc summary. No template, no script, just a trained interviewer and a blank page, with the summary written within forty-eight hours. Imagine a research function where the same senior HR partner has run every exit for a decade, and the leadership team has learned to read that person's summaries the way they'd read any other trusted analyst. The strength is that you can hear the thing nobody thought to ask. The weakness is that you can't compare across leavers, you can't defend the conversation to a reviewer, and you're entirely dependent on the skill of the interviewer. This approach is right in small, expert teams where the interviewer is the same person for every exit and the output is read by someone who trusts them. It fails everywhere else, because the moment the interviewer changes, the output changes with them, and the programme is no longer a programme but a person.
Five Diagnostic Questions You Can Answer About Your Own Organisation
Who actually reads the output of your exit interviews today, and what decision have they made from it in the last twelve months? Sit down and write the answer. If the answer is "no one" or "no decision", the template is performing a ritual, not a function. Either change the audience or change the form. You can't fix this with better questions. The way to answer it for your own organisation is to pick a name and a meeting: find the person who should be reading the roll-up, ask them what they did with the last one, and write down what they say. If the answer is "nothing", the form is the wrong thing to be designing.
Of the questions currently on your template, how many have changed a decision in the last year? Walk the list. Any question whose answer didn't change anything is a candidate for removal. The temptation will be to keep them because they might be useful someday. Resist it. A question that has not earned its place in a year won't earn its place in the next one. To answer it honestly, you need a log, however rough, of decisions that referenced exit data, and a willingness to count the questions that did not feature. If you don't have the log, that's already the answer to a sixth question you didn't ask.
Can a manager, without permission, add their own questions to the form? If yes, your comparability is already compromised. The answer should be no, with a clearly marked optional box below the fixed spine for questions that belong to the team rather than the company. Anything added above the spine without approval should be a governance issue, not a flexibility feature. To answer it for your own organisation, ask three managers in three different teams what they did the last time they wanted to ask a leaver something the form didn't cover. If two of them wrote it in the margin, you already have your answer.
When did you last remove a question from the template, and who authorised it? If you can't answer, the template is growing. Find out who owns it and give them the job of removing one question every review cycle. The honest general point is that a template grows because adding a question is easy and removing one requires someone to say it was not useful, so the governance question is who is allowed to remove a question, not who is allowed to add one. To answer it for your own organisation, look at the version history of the form. If the history only ever shows additions, or if there is no version history at all, the form is unowned.
If a former employee asked for a copy of what was recorded about them, would you be comfortable handing it over? In most organisations, exit notes are personal data about a living individual, and the leaver has a right to ask what was written. If the answer makes you uncomfortable, the issue isn't the template. It's what gets recorded, by whom, and in what form. To answer it for your own organisation, take the last exit note you produced, read it as if you were the leaver, and ask yourself whether you would be willing to hand the page over without redaction. If the answer is no, the note is doing two jobs, and one of them is making you exposed.
Six Template Decisions, Reviewed
Fixed Rating Scales
A scale is a question with a fixed set of numbered answers, usually three to seven, where the leaver picks one. "On a scale of one to five, how would you rate your relationship with your manager?" is the prototype.
It earns a place because it's the only way to get a number you can put on a chart and compare across leavers. Without at least a few scales, your roll-up is paragraphs and your leadership team can't read it. The chart is what makes the question defensible at the leadership table.
Where it falls short is that the leaver picks a number that fits the question you asked, not the situation they lived. A manager who was technically excellent and personally toxic scores well on "competence" and badly on "trust", and the average looks acceptable. Scales also flatten group effects. A whole department can be quietly miserable and score its manager a four, because four is what you score a competent manager when you don't want to put down a five. The number on the chart will tell the leadership team the wrong thing with the appearance of precision, and the precision is the part that makes the mistake expensive.
Fixed Reason Taxonomy
A taxonomy is a closed list of reasons for leaving that the leaver picks from, like "career progression", "manager", "compensation", "workload", "personal". The list is the template.
It earns a place because it produces a chart your leadership team will actually read. Without it, you've free-text reasons that nobody can roll up. With it, you've a pie that updates every quarter, and the pie is the artefact that survives the meeting.
Where it falls short is that the categories drift away from how people actually leave. Five years in, "manager" may have eaten everything, because every leaver who could have said "strategy" or "product" says "manager" instead, and your chart is no longer telling you anything. The honest general point is that a fixed taxonomy is right until it's wrong, and the moment it's wrong is the moment a new reason appears that nobody put on the list. Review it on a fixed cadence and mean it. The risk is that the chart will keep telling a confident story about a category that no longer reflects how the building works.
Mandatory Open Questions
A mandatory open question is one that every interviewer must ask and that the leaver is invited to answer in their own words. "What would have changed your mind about leaving?" is the prototype.
It earns a place because it's the only question type that can surface the thing the form didn't anticipate. Every comparable number on your template comes from a question you knew to ask. Every useful insight, in most programmes, comes from one of these. Without the open question, the form is a mirror of the questions you had the nerve to write down last year, which is not the same as the questions that needed asking.
Where it falls short is that the answers are heavy, slow to read, and inconsistent in quality. Some leavers write paragraphs. Others write nothing. The interviewer has to be good enough to handle silence without filling it. Mandatory open questions are the place where a template either works or doesn't, which is also why they're the first thing cut when a form is under pressure to be shorter. The cost of getting them wrong is that the form's spine is exposed: scales and taxonomy with no test of whether the spine is pointing at anything real.
Optional Open Questions
These sit below the fixed spine. They're prompts the interviewer can use or ignore, drawn from a list someone has approved. "Was there anything about your role you would change for the person who replaces you?" is a typical example.
They earn a place because they let a manager dig into the team-specific question without contaminating the comparability of the spine. A sales lead can ask about quota design. An engineering lead can ask about on-call. Neither question enters the company-wide roll-up. The optional box is the place where a template stays alive to the parts of the business the spine can't see.
Where they fall short is that "optional" quietly hardens into "standard", and within two review cycles every interviewer is using the same four prompts in the same order. The optional box has to be reviewed and trimmed with the same discipline as the fixed spine, or it stops being optional in practice. When it does, the spine and the optional list become a single bloated form with a polite label on the second half, and the roll-up starts to drift.
Interviewer Prompts Rather Than Scripted Wording
A prompt is a sentence the interviewer can paraphrase. A script is the exact wording. The difference looks small. The consequence isn't.
Prompts earn a place because they let the interviewer match the leaver. A leaver who is emotional gets a softer prompt. A leaver who is matter-of-fact gets a tighter one. The conversation stays human, and the signal in the answers goes up. A prompt also gives the interviewer room to follow a thread the script didn't anticipate, which is the moment a template either pays for itself or doesn't.
Where they fall short is that two interviewers will read the same prompt in two different ways, and the answers you get are no longer perfectly comparable. The trade-off is between the comparability of the wording and the signal in the answers. Most programmes choose signal. Make sure you're choosing it on purpose. The risk is that "prompt" becomes a euphemism for "the interviewer asked whatever they wanted", and the roll-up is no longer comparable to anything, including itself across quarters.
Free-Form Notes With No Template at All
The interviewer writes what they want after the conversation, with no fixed questions at all.
It earns a place because it removes every ceiling. The interviewer can follow whatever the leaver actually said. Nothing on the page was decided before the conversation happened. In a small expert group, this is the form that catches the most, because the interviewer isn't being asked to follow a script that was wrong by the second quarter.
Where it falls short is that you can't compare across leavers, you can't defend the process to a reviewer, and you're betting the entire programme on the skill of one or two interviewers. This is the right answer for a small expert group. It's the wrong answer for a company that has to demonstrate consistency across fifty leavers a quarter, because the demonstration of consistency is the thing the form is supposed to produce, and no form means no demonstration.
The Decision Table
| Situation | Scale | Setup | Primary Pain | Recommended Starting Point |
|---|---|---|---|---|
| Single site, under fifty leavers a year, one HR lead runs all exits | Tiny | Conversation + one-pager | None | Leave the template alone |
| Two sites or two HR generalists, comparability starting to slip | Small | Fixed spine, optional prompts | Inconsistent answers | Semi-structured template, four to five fixed questions |
| Leadership asking for turnover charts by team | Medium | Existing form, no roll-up | Anecdotes instead of numbers | Add scales and a taxonomy, keep open questions |
| Multi-site, multi-country, leaver has asked for their record | Larger | Scripted interview, third-party summariser | Legal exposure | Prompts rather than script, clear ownership, written governance |
| Sector where exit conversation can become evidence | Specialist | Free-form notes, expert interviewer | Defensibility | Take local advice before changing anything |
| Form grown over years, nobody owns it | Any | Bloated, every manager amends it | Governance vacuum | Run a review, name an owner, remove one question per cycle |
| Manager quietly adds own questions for tricky leavers | Any | Mixed inputs, no separation | Comparability corrupted | Optional box below the spine, separate roll-up |
| HRIS or survey tool has hard-coded the template | Any | Vendor-locked form | Cannot edit without paying | Audit what is actually being asked before any platform change |
The Minimum Viable Template
The table below is the form a serious small programme can run from. Anything above this is a deliberate expansion, and the expansion has to be justified by a decision someone plans to make from the answer.
| Element | Fixed or Open | What it buys you |
|---|---|---|
| Reason for leaving (from a closed taxonomy of around eight categories) | Fixed | Comparable roll-up, a chart leadership will read |
| Overall experience rating (single five-point scale) | Fixed | One comparable number across leavers and time |
| "What would have changed your mind?" | Open | The single highest-signal question in most programmes |
| "Anything we did not ask that you want us to know?" | Open | The space where the unexpected surfaces |
| Optional team-specific prompts below the spine | Open | Team signal without contaminating comparability |
| Interviewer summary within forty-eight hours | Open, with a deadline | A defensible record of the conversation |
| A named template owner and a review cadence | Governance | The only thing that prevents the form from quietly growing |
That's the floor. Anything you add must come with the name of the decision it changes and the person who will make it.
Where Templates Go Wrong
| Failure pattern | What it looks like in practice | What fixes it |
|---|---|---|
| Template grows every quarter | Three new questions after every senior exit, two more after the last engagement survey, none ever removed | A single named owner with the explicit job of removing one question per review cycle |
| Scales nobody calibrated | The "manager" scale averages a 4.2, the "workload" scale a 2.8, and nobody can tell whether managers are better than the work or whether the scales are measuring different things | Calibrate against an external anchor or against free-text, retire scales that do not move |
| Categories that no longer match how people leave | The taxonomy still has "commute" and "office facilities", and the leavers are all leaving because of hybrid policy | Review the taxonomy against the last fifty exit reasons; cut anything that did not appear in the last year |
| Optional questions become standard | The optional box has twelve prompts everyone uses in the same order | Treat the optional list like the fixed spine: review it, prune it, own it |
| Manager amends the form | A line manager adds a question to a difficult leaver's form, gets the answer they wanted, and the roll-up shifts | Lock the spine; allow additions only in the optional box, and only outside the roll-up |
| The owner is a committee | Every change needs three sign-offs, so nothing changes, and the form rots in place | One owner, one deputy, a review on a fixed date regardless of whether anything feels wrong |
| Notes are kept indefinitely | The leaver's exit note sits in their file for years, alongside everything else, and nobody is sure what it can be used for | A retention policy for exit records, written down, that says how long notes are kept and who can read them |
What to Put in Writing
The decisions in this article aren't enforceable on their own. They become enforceable when they're written down, owned, and dated. The artefacts below are the ones a defensible exit process leaves behind. Each one has a purpose beyond the conversation itself.
| Artefact | Who owns it | When it is written | What it prevents |
|---|---|---|---|
| The current template, version-numbered | Head of HR or People Ops lead | At every review cycle | "I did not know we had agreed to that" |
| The owner of the template, named | HR director | Once, then re-stated at every review | A form that drifts because nobody is responsible |
| The taxonomy of reasons, with definitions | People analytics lead | At each review, even when nothing has changed | Two interviewers coding the same reason two different ways |
| The retention policy for exit notes | Head of HR, with legal review | Once, reviewed annually | Records kept longer than they should be, or in ways that expose the organisation |
| The review cadence and the last review date | Template owner | At the close of every review | A form that grows without anyone noticing |
| The boundary between the fixed spine and optional questions | Template owner | At every change to the form | A manager quietly amending the roll-up for their own team |
| A short note on what is recorded and what is not | Interviewer, after each conversation | Within forty-eight hours | A leaver being told, later, that something was said that was not, or was written down that should not have been |
| The escalation path when something serious surfaces | HR director | Once, then refreshed after any incident | A serious disclosure sitting in a one-page summary nobody acts on |
| A record of the last review and what was removed | Template owner | At the close of every review | The form becoming a museum of questions that have never been tested |
If you can't point to a document in this table and say who holds it, the template is a habit, not a process.
Questions to Ask Before You Commit
These are the questions to put to a vendor, an adviser or your own team before you lock the decision. Each one has a short bold label and a line on what a bad answer sounds like.
Ownership. Who is the named owner of the template, and what is their authority to remove a question? A bad answer is "everyone" or "the HR team", because that's no one.
Comparability. Which questions on the form are guaranteed to be asked in the same words of every leaver, and how do you know? A bad answer is "we use the same form", because using the same form isn't the same as asking the same question.
Flexibility. Where on the form is a manager allowed to add their own question, and what happens to the answer? A bad answer is "anywhere" or "they can just write it under the existing box", because both contaminate the roll-up.
Review cadence. When was the template last reviewed, who reviewed it, and what was removed? A bad answer is silence, or "we've not had to change anything", because a form that never changes is a form nobody is reading.
Records. How long are exit notes kept, who can read them, and what happens if a leaver asks for a copy? A bad answer is "as long as we need them", because in most places that's a longer answer than the law allows.
Calibration. How do you know your scales and your taxonomy are measuring what you think they're? A bad answer is "the manager scores them", because that's the manager judging the manager.
Signal. What is the most useful thing the process surfaced in the last twelve months, and what decision did it change? A bad answer is "we've not had a clear case yet", because that's the same as saying the process has not yet earned its keep.
Exit. What is the procedure for retiring the form, and what happens to the records when it's retired? A bad answer is "we've not thought about that", because that's what the audit will be about.
The Cost of Getting This Wrong
The visible costs of a bad exit template are familiar: low completion rates, answers that read like resignation letters, a roll-up that nobody trusts. The harder costs are the ones that show up later, in rooms where the exit process isn't the subject of the meeting.
The first is comparability you can't undo. If a manager has spent two years adding their own questions to the form, the data you've is no longer theirs and no longer yours. You can't reconstruct it. Every chart you build from that data carries the contamination, and every leadership decision based on that chart carries it forward. The next redesign will take longer and cost more, because the audit needed to clean the data is the work that should have been done in governance.
The second is defensibility. In most places, exit notes are personal data about a living individual, and the leaver has a right to know what was written. If the notes were taken by an interviewer who was also the leaver's manager, who also added their own questions, who also kept the notes in a folder nobody else could see, the position is harder to defend than the conversation warranted. The exposure is not the content of the notes. It is the chain of custody, the relationship of the interviewer to the leaver, and the absence of any document that says what the process was supposed to be.
The third is the quiet loss of trust in the process itself. Once a team has decided the exit interview is a ritual, the next exit interview in that team will be performed rather than felt, and the signal in the answers drops to zero. Rebuilding that signal takes longer than building it the first time, because the leaver reads the room before they answer a single question, and a team that has already decided the form is a sham will not give the next interviewer anything honest to work with.
So the question that actually belongs on the table isn't "what should our template look like", it's "what kind of conversation do we want to be having with people at the moment they leave, and are we set up to have it". Everything else is in service of that question.
Going Further
HROpsLab is a review publication. We don't sell software, payroll, HRIS or advice. We test what is on the market, write up what we find, and leave the buying decision to you. The independent comparison work in this category is the same kind of work as this article: a careful look at what each option actually does, set against what a serious HR function needs it to do. If the questions in this piece are the ones you're sitting with, the comparison work is the natural next step. It won't tell you what to buy. It will tell you what you would be buying, in plain language, from a position that's not trying to sell you anything.
Frequently Asked Questions
How long should an exit interview template be?
A single page. If it can't fit on a page, the second page is the part to cut. Comparable data comes from a small fixed spine, and the marginal question past the tenth one starts costing you signal from the first nine. A long form also signals to the interviewer that the form matters more than the conversation, which is the inverse of what you want.
Should we use rating scales on the template?
Yes, but never on every question and never without open text alongside. A scale gives you a number you can chart and compare. An open question gives you the thing the scale can't see. Use each where it earns its place. A scale on its own produces numbers that look authoritative and mean little. Open text on its own produces paragraphs nobody can summarise.
How many open questions should we include?
Two is usually enough. One that asks the leaver to name the thing that would have changed their mind. One that invites them to say what the form didn't ask. Anything past that starts to feel like a survey, and the leaver's patience for it drops visibly in the second half of the conversation. Two strong open questions, asked well, will produce more signal than ten perfunctory ones.
Should the template vary by seniority?
No, not on the fixed spine. The spine exists to make every leaver comparable, and that comparability is most valuable across the roles where the answers are most likely to be different. Team-specific prompts can sit below the spine, and a senior leaver may warrant a longer conversation. The form itself should not.
Who should own the template?
A single named owner, usually the head of HR or a People Ops lead, with the explicit authority to remove a question. A committee can't own a template, because committees can add and can't subtract. The owner doesn't have to write every question. They do have to be the person who says no.
How often should we review the template?
At least once a year, on a fixed date, even when nothing feels wrong. A review that only happens after an incident is a review that happens too late. The review should ask one question of every question on the form: what decision did this question change in the last twelve months? Any question whose answer is "none" is a candidate for removal.
What should we do when a manager wants to add their own questions?
Give them a clearly marked optional box below the fixed spine, kept separate from the roll-up. If their question is good, it migrates to the spine at the next review. If it isn't, it dies in the optional box without contaminating comparability. The point is that manager curiosity is useful. Manager amendments to the roll-up aren't.
Should the interviewer take notes during the conversation or write them up afterwards?
Write them up afterwards, within forty-eight hours, from memory and from any short notes taken at natural pauses. Notes taken throughout change the conversation, make the leaver cautious, and produce a record that reads like a transcript of what was said rather than what was meant. The summary written after is closer to the actual signal, and it's also the easier record to defend if it's ever asked for.
HROpsLab is an independent review publication for HR operations teams. We test the tools, write up what we find, and let you decide.
