TL;DR
- The trap: the most expensive mistake is starting the exit conversation before the local process is mapped, because the first meeting can foreclose options you didn't know you had.
- When doing nothing is right: when the only thing wrong is that the person is in a country you didn't expect, and the rest of the operation there's small, the local entity or partner already runs the employment, and the role genuinely can be backfilled from the home country on a long horizon.
- What has to be true: you can identify a single legal employer of the worker, a recognised process exists in that jurisdiction, and the decision can wait long enough to follow it.
- How the options split: run the exit through the local entity, run it through the employer of record that employs the person, transfer first and exit second, or in narrow cases end the assignment without ending the employment.
- Decision rule: the question isn't "do I want to let this person go", the question is "who is the legal employer right now, and what does that employer's jurisdiction require of them".
- Outcome to expect: a slower calendar than any at-will leader is comfortable with, a heavier documentation load, and a defensible position if challenged, or a fast calendar and a much larger problem later.
The room where it goes wrong
A people director opens a personnel file on a Tuesday morning. The employee in question is a senior account lead in a country her team has hired into for the first time in eighteen months. The role is a genuine hire, not a contractor misclassification, the contract is signed, the payroll is live, and the head of the region has already told her that the relationship is no longer working and he wants it ended this quarter. She has done this conversation a hundred times at home. She has a draft separation letter on her screen. But the country on the file isn't one of the ones her team usually operates in, and the entity structure isn't the one she is used to, and the head of region's "this quarter" carries a different weight in a jurisdiction where consultation is a legal stage rather than a courtesy call. She has until the end of the week to decide whether to send that letter.
So here's the part no one in her office will say out loud. The thing she is most afraid of is that the letter is wrong, or the timing is wrong, or the reason cited in it doesn't exist in the local law. The thing she should be more afraid of is that the letter is fine and the conversation that delivers it forecloses a process she didn't know she had. In most places, the first meeting is a moment of record. What is said in it, what is offered, and what the employee says back can narrow the available exits from several to one. She isn't choosing how to end the relationship. She is choosing which of several possible endings becomes the only one left, and she is doing it in the first ten minutes of a conversation she has not yet prepared for. The real issue isn't whether she is allowed to let this person go. The real issue is what she has to do, on the record, before she lets this person go, and whether she has done it.
When you genuinely do not need to act
This piece is written for readers who do need to act. But the credibility of everything below depends on saying clearly that there are four honest stages of exposure, and the first one is the absence of one. If you recognise your situation in stage one, stop reading. The work is already done.
Best tools for Global HR
Stage one: your current setup is genuinely fine. You operate in the country through a properly capitalised local subsidiary, that subsidiary is the legal employer of every worker on its headcount, payroll is run in-country, statutory filings are up to date, and the team in-country is small enough that no collective body is triggered by the change you're about to make. In this configuration the exit runs the way an exit runs in any mature employer: through HR, through the local entity, with local counsel reviewing the documents that get signed. You don't need a new vendor. You don't need a new process. You need your existing process followed, with a local lawyer reading the termination notice and the settlement language before it's sent. Many companies who think they have a cross-border problem are actually in this stage and have talked themselves into the next one.
Stage two: there's friction but no real risk. The legal employer is a third party, most often an employer of record, and the third party has been the legal employer from the start. The friction here's speed and process shape. The EOR will run the exit on its own cycle, through its own templates, with its own internal review. Your influence on the reason cited, the documentation, and the calendar is real but indirect. You provide the business case. The EOR owns the legal case. The risk in stage two isn't breach. The risk is timeline slip and a settlement shaped by the EOR's standard rather than your preferences, which is usually fine and occasionally isn't.
Stage three: there's real risk and you have to slow down. The worker is on your home-country payroll, has been for years, and has been working from the second country with a verbal or written understanding that this is a long-term remote arrangement. At this point the worker may, in many jurisdictions, have acquired local employment rights by operation of law regardless of what the contract says, and the home-country contract may be the wrong document for the exit you want to run. The OECD's recent guidance on when a remote home office can create a permanent establishment makes this more than a labour-law question. It's also now a corporate-tax question, and the answer in either direction needs to be confirmed locally before the first meeting. Any action in stage three without local counsel is gambling.
Stage four: the edge case that's its own animal. The worker is on a posted-worker or detached assignment, the original posting has a defined end date, and the head of region wants to "end it early". Posted-worker arrangements carry their own regime, and "early" is a word that has to be checked against the assignment's actual terms, the host country's posted-worker rules, and the home-country law that the assignment was built on. This is the stage where assumptions are most often fatal, and where a five-minute conversation with local counsel before anything is written prevents a six-month problem afterwards.
The five questions you ask yourself at 11pm
Who is the legal employer of this person, right now, on the paperwork and in the eyes of the local authorities. If you can answer this in one sentence with confidence, the question is settled. If you can't, nothing else below can be settled either, because every step of the exit flows from who the legal employer is. Get this wrong and the worker is owed protections by an entity that didn't employ them, and the entity you thought employed them has a process it has to follow that you've not started. Confirm it on paper, in the local language, against the most recent filing.
Does the jurisdiction require a reason the law recognises, and have I documented the business case to that standard. In much of the world dismissal needs a reason that the local law treats as a valid ground, and the standard for what counts as valid is set by statute and case law in that country, not by your home practice. Performance is a valid ground in many places and not in others. Restructure is valid in some and not in others. The reason you would lead with at home may not be the reason that survives review locally. Document the case to the local standard, not to the home standard, before the meeting.
Is there a collective body I have to consult before I can act. In countries with works councils or similar representative structures, the consultation is a stage, not a courtesy. It has a defined scope, a defined sequence, and a defined consequence for skipping it. Skipping it doesn't speed the exit up. It invalidates the exit you ran. Find out whether the role, the headcount in the country, or the type of change triggers a body, and treat the consultation as a non-skippable part of the process, because it's.
What does the timeline actually look like, and can the business tolerate it. Many local processes run longer than a US-style at-will exit, sometimes considerably longer, and the delay isn't slack. It's the legally required cooling-off, consultation, and notice period. If the business can't tolerate the calendar that the local process imposes, the answer isn't to compress the process. The answer is to take the action that doesn't run through the process, which usually means transferring the employee to a different legal entity, ending the assignment, or having the EOR run the exit on its own clock, depending on who the legal employer is.
If this is challenged in twelve months, what does the file look like. A defensible exit file has the reason, the process, the consultation evidence, the notices, the communications, and the signed settlement, in the local language, in the right order, with the dates correct. An indefensible file has the reason, an offer, a signature, and a gap. Picture the file a hostile local counsel would want to see, and build that one.
Three honest categories the approaches split into
Run the exit through the local entity. This is the right route when the local entity is the legal employer, the entity is properly set up, the in-country team is past any thresholds that trigger collective consultation, and the reason for the exit maps to a ground the local law recognises. The work is local: local counsel drafts or reviews the notice, the local HR function runs the process, the entity signs the documents. The advantage is that the entity that employs the person is the entity that ends the employment, which is the structure every local court expects to see. The route fails when the local entity isn't properly capitalised, has not been running its own payroll, or has been treating the in-country team as a soft extension of the home operation. In that case the local entity exists on paper and the courts will treat the home operation as the actual employer, and the home operation is now in a foreign labour court with no process of its own.
Run the exit through the employer of record that already employs the person. The EOR is the legal employer. The EOR runs the local payroll and the employee-level obligations. That means the exit runs through the EOR and on the EOR's process, not yours, and on a calendar the EOR sets, not one you set. The EOR has its own templates, its own standard settlement language, and its own internal review. You provide the business case and the timing pressure. The EOR provides the legal case and the document. The route fails when the EOR is being used as a workaround for a permanent establishment exposure that exists regardless of who runs the payroll, when the EOR's standard settlement isn't strong enough for the actual dispute risk in the file, or when the business needs the exit run on a specific reason that the EOR's process won't let them cite. The EOR is the right legal employer for the exit, but the EOR is also a third party with its own interests in the settlement size and its own risk tolerance that may not match yours.
Transfer first, then exit. This is the route for the case where the employee is on the home-country payroll, has been working in-country for long enough that local rights may have attached, and the home-country process wouldn't survive local review. Transfer the legal employment to a local entity or to an EOR first, on terms the local law accepts, run a short transition under the new employer, and only then run the exit through the new legal employer under the local process. The route is slow and expensive on paper. It's faster and cheaper than the alternative, which is an exit that the local court sets aside. The route fails when the transfer itself is a sham, when the local entity isn't ready to be the legal employer, or when the business needs the exit done inside a window the transfer can't fit inside. The transfer-first route isn't a trick. It's the honest fix for a structure that was never going to survive an exit.
Five diagnostic questions you can actually answer
Is the legal employer in this file the entity I think it's. Open the most recent payslip. Open the most recent local filing. If the employer named there's not the entity you would name, the exit you're about to run is built on the wrong employer, and the rest of the answers below are unreliable. If you can't read the local filing, get someone who can.
Can the reason I want to cite, in plain language, be matched to a ground the local law recognises. Write the reason down. Hand it to local counsel. Ask whether the ground exists, whether the documentation supports it, and whether the timeline you've in mind is consistent with how a local court would expect the process to unfold. If local counsel can't say yes to all three, the reason is wrong or the documentation is, and one of them has to change.
Is there a works council, a staff representative, or a collective body I have to consult. Find out whether the headcount, the role, or the type of change triggers a body in the country. If it does, the consultation is a stage with its own minimum timeline. The exit doesn't start before the consultation ends. The consultation doesn't end before the body has had the time the law gives it.
What is the realistic calendar from decision to final signature, on a clean process. Ask local counsel for the calendar on a clean process, with the consultation built in, the notice built in, and the settlement window built in. That calendar is the answer. Not the calendar the business would like. Not the calendar the EOR quotes you on the phone before it has seen the file. The calendar that survives a hostile reading in twelve months.
If the worker challenges the exit in the local court, what is the document I would put in front of the judge first. That document is the one the entire file has to be built around. If you don't have a clear answer to that question, the file isn't yet organised around the right thing, and the exit is premature.
Six Things That Do Not Travel
At-will employment
At-will employment is the assumption that either party can end the relationship at any time, for any reason that's not a protected category, with notice set by contract. It's the default in some US states and it doesn't exist as a concept in most of the world. In many jurisdictions dismissal needs a reason the law recognises, a process that can be evidenced, and notice that can't be bought out as casually as you expect. The instinct to treat the next conversation as the last conversation is the instinct that has to be parked at the door, because the next conversation is in many places the opening of a process rather than the closing of one. Where it falls short as a guide: a US-based HR leader who has run hundreds of exits on at-will thinking will read "recognised reason" and assume the bar is lower than it's, because their bar has been on the floor for their entire career. The local bar is set by statute and by the case law of hostile counsel, and it's the hostile counsel's reading that matters.
Notice you can simply pay in lieu of
Pay in lieu of notice, or PILON, is the practice of paying the worker their notice period as a lump sum and ending the employment on the same day. In some jurisdictions PILON is a recognised mechanism and the contract can call for it. In others it's restricted, the contract clause is unenforceable, and the worker is still an employee until the notice period has actually run, with all the rights and costs that carries. The home instinct is that a PILON clause in a contract settles the question. The local answer is that the clause has to survive local-law challenge, and the test for that's set locally. Where it falls short: a senior leader reads a PILON clause in a contract, assumes the question is closed, and writes the same-day termination letter. Two months later the worker is still legally employed, the company is on the hook for the intervening pay and benefits, and the contract clause is the document the local court is about to read.
Performance management as an informal process
Performance management in many home jurisdictions is a documented but conversational practice: a series of one-to-ones, a written summary, a final conversation, and an exit. In many other jurisdictions that informal practice isn't the evidentiary record the court expects to see. Performance is a valid ground for dismissal in many places, but only when the file can show that the underperformance was identified, that the employee was told what was required to improve, that they were given a real chance to improve, and that the assessment of failure was made against the standard that was set, on the timeline that was set. The home instinct is to treat the performance file as a record of the relationship. The local expectation is that the performance file is the only thing standing between the dismissal and a finding of unfair dismissal. Where it falls short: the HR leader arrives at the exit with six months of meeting notes and no formal improvement plan, and the local court reads the file as a person who had no clear chance to improve, because in the language the court is reading, the chance was never formalised.
The confidentiality of a settlement
A settlement, in many home jurisdictions, is a private document between two parties, and the duty of confidence sits on both. In other jurisdictions the settlement isn't the end of the process in the way the home instinct expects. Tax authorities, social security authorities, and labour inspectorates can have rights of access to settlement documents that override the duty of confidence in the contract. A settlement that names a figure which the tax authority later reads as under-reported creates a problem the settlement was supposed to prevent. The home instinct is that the signed settlement closes the file. The local answer is that the signed settlement is one of several files the authorities can open, and the language in it has to anticipate that. Where it falls short: a settlement is signed with a confidentiality clause the worker has agreed to, and the company is surprised six months later when the tax authority asks for the document and the worker hands it over without breach, because the local law required the worker to hand it over.
Unilateral changes to terms during exit
In some home jurisdictions, an employer can adjust terms during a notice period, reduce responsibilities, change reporting lines, or remove access, as part of garden leave or a managed exit. In other jurisdictions, any change to terms during the notice period is itself a unilateral variation, can be challenged as a constructive dismissal, and can produce a finding against the employer even where the original exit was sound. The home instinct is to clean the role down during the notice period. The local answer is that the terms on day one of the notice period are the terms that have to be honoured through to the end, and changes to those terms have to be agreed, not imposed. Where it falls short: the worker is put on garden leave with a reduced compensation package that the contract didn't provide for, and the local court reads the variation as the real exit and the original exit as the pretext.
The assumption that a probation period behaves the same way
A probation period at home is usually a window of reduced notice and reduced protection, on the assumption that either side can walk away inside the window. In other jurisdictions, probation doesn't reduce protections in the way the home instinct expects, or the length of the probation is capped, or the documentation required during probation is heavier than after it, or the probation clause in the contract is unenforceable as drafted. The home instinct is that the new hire is on probation and the exit can run on a short clock. The local answer is that the probation clause has to be read against the local statute, the case law, and the contract language, and the short clock is the first thing to be checked. Where it falls short: a hire is made, the new joiner doesn't work out, and the company tries to exit on the probation clause only to find that the clause is unenforceable, the statutory protections applied from day one, and the company has been running the wrong process for the entire probation.
The Decision Table
| Situation | Scale | Setup | Primary Pain | Recommended Starting Point |
|---|---|---|---|---|
| Small team, in-country entity is real, role is not collective-trigger | Single country, small headcount | Local subsidiary as legal employer | Documentation standard, not legal exposure | Local counsel review of notice and settlement; run through local entity |
| Small team, EOR is the legal employer from day one | Single country, small headcount | EOR as legal employer | Calendar and EOR's standard settlement | Brief the EOR on the business case; let the EOR own the process on its clock |
| Long-tenured remote worker on home-country payroll | One or two workers, one country | Home entity as named employer, worker physically in-country | Local rights may have attached; tax exposure on the home office | Local counsel before any conversation; consider transfer-first route |
| Posted-worker or detached assignment, early end requested | Single worker, defined assignment | Home entity remains legal employer | Posted-worker regime and assignment terms | Counsel review of the assignment terms and the host-country rules before the request is granted |
| Larger in-country team, headcount change may trigger a body | Multiple workers, one country, mid-sized team | Local subsidiary as legal employer | Collective consultation stage | Map the consultation stage with local counsel before any individual conversation |
| Restructure of a function across multiple countries | Multi-country | Mix of local entity, EOR and home-payroll workers | Inconsistent treatment and PE exposure | Cross-country mapping with single local-counsel coordinator; sequence the exits |
| Senior individual contributor on EOR, sensitive settlement required | One worker, one country | EOR as legal employer | EOR's standard settlement may not be strong enough | Joint working with the EOR; shape the settlement before the EOR's process starts |
| EOR being used to mask a permanent establishment exposure | Multiple workers, ongoing | EOR employs but local activity is real | PE risk on top of the exit itself | Resolve the structure first, then run the exit; the exit does not fix the structure |
The Sequence, and Where It Is Different
The work that has to happen before the conversation is heavier than the work that has to happen after it. The instinct is to do the opposite: have the conversation, hand over the notice, and let the documentation catch up. In most jurisdictions outside the US, that sequence is the wrong one. The documentation has to be in place before the conversation, because what is said in the conversation can fix the reason, the offer, and the calendar in ways that the documentation afterwards can't undo.
| Step | What happens | Why it has to happen first | Common failure when it does not |
|---|---|---|---|
| Confirm the legal employer on paper, in the local language | Match the payslip and the local filing to the entity you believe employs the person | Every step below is owned by the legal employer; the wrong owner invalidates the rest | Exit is run by an entity that did not employ the person; worker is owed protections by the entity that did |
| Map the local process with local counsel | Identify the required reason, the consultation stage, the notice, and the calendar | The process sets the calendar, the documentation, and the reason that can be cited | Reason cited in the conversation is not one the law recognises; the conversation fixes the bad reason |
| Build the file to the local standard | Performance, conduct, or restructure evidence, in the form the local court expects | The file is what stands between the exit and a finding of unfair dismissal | Exit is set aside for lack of evidence the worker had a real chance to improve or to respond |
| Identify and consult any collective body, where required | Run the consultation the law requires, in the sequence it requires | Skipping consultation is not a speed gain; it is a process defect | Exit is invalidated regardless of the merit of the underlying reason |
| Draft the notice and settlement in the local language | Local counsel reviews; the documents are ready before the conversation | The first conversation is a moment of record; the documents have to be ready to be referenced in it | The conversation commits to terms the documents cannot match, and the gap is what the worker later challenges |
| Have the conversation with the documents present | Walk through the reason, the process, the calendar, and the settlement, in that order | The order matters: process and reason before settlement, not the other way around | The conversation is read as a settlement negotiation in which the reason was negotiable, and the court reads it the same way |
Who You Need in the Room
A cross-border exit isn't run by HR alone. It's run by HR in concert with a small set of other roles, each of which has a specific job and a specific window in which they need to be in the room. The table below is the practical annex to the sequence above.
| Role | What they need from you | When they need it | What they own |
|---|---|---|---|
| Local counsel, employment and tax | The file, the proposed reason, the calendar pressure, the in-country activity (for PE analysis) | Before the conversation, not after | The legal shape of the exit; the notice language; the settlement; the tax-and-PE view of the underlying arrangement |
| The legal employer's local HR function (or the EOR's HR team) | The business case, the timing pressure, the documentation you have | At the start, before the documents are drafted | The local process, the templates, the calendar, the consultation stage where one is required |
| Works council or employee representative, where one exists | The information the law requires you to share, in the form it requires, on the timeline it requires | Before the individual conversation, by statute in many places | The consultation; the right to be heard on the proposed change |
| Finance, for the settlement cost and any tax-withholding on the payment | The proposed settlement figure and structure | Before the settlement is offered, not after | The funding, the tax treatment, the accounting for the payment |
| Internal communications, where the exit will be visible in-country | The line that will be given to the team, the line that will be given to clients, the timing | Before the conversation, on a sequence internal comms sets | The internal narrative, calibrated to the local process so it does not contradict the legal file |
| The head of region who initiated the request | The realistic calendar, the realistic reason, the realistic settlement range | Before they commit to a date internally | The decision to proceed on the local process, on the local clock |
The cost of getting this wrong
The invoice from the bad exit isn't the invoice anyone budgets for. The settlement is visible. The legal fees are visible. The replacement cost is visible. The costs that destroy the case are the ones that never appear on an invoice and that compound over years. A finding of unfair dismissal in a local court is a public record, and a public record in a country where the company is hiring is a document every future candidate can find, every future regulator can read, and every future works council can hold up as the company's track record. A second-order cost is the EOR's willingness to work with the company afterwards, because an EOR that has watched a client run a bad exit is an EOR that prices the next engagement for the risk, and that price increase shows up in the per-head fee, not in the line item called "settlement". A third is the headcount. Companies that run one bad exit in a country find that the in-country team doesn't grow the way the plan assumed, because the local talent market has read the same public record and the candidates the company wanted have read it too. The tax exposure sits on top of all of this when the exit exposes an underlying permanent establishment that the structure had been quietly absorbing, and the OECD's recent guidance on remote home offices has made that exposure easier to identify and harder to argue with.
So the question that reframes the decision isn't "can we afford the settlement". The question is what the file looks like in twelve months if the worker challenges the exit in the local court, in the local language, with local counsel, and what the company looks like in that country in three years if the file is the one a hostile reading produces.
When you are ready to go further
If you've read this far and the situation you're holding doesn't fit stage one above, the next step isn't a vendor conversation. It's a structured comparison of how the major employer-of-record providers and the major global payroll operators actually behave on a real exit in a real country, because the differences between them on a clean hiring engagement are small and the differences on a contested exit aren't. HROpsLab does independent comparison work across the global employment and payroll space. We are a review publication, we sell nothing, and we don't supply software, consulting or legal advice; we publish the work and you take it from there.
Frequently Asked Questions
Does at-will employment apply anywhere outside the United States?
A few jurisdictions have at-will-style frameworks in narrow forms, but the dominant pattern outside the US is that dismissal needs a reason the local law recognises and a process that can be evidenced. The closer the situation is to a US-style at-will assumption, the higher the chance that the local process has been misread. Confirm locally before acting on the at-will instinct, and confirm by reading the most recent local filing, not by analogy.
How much notice is required?
The notice period varies considerably by jurisdiction, by length of service, by contract, and by the type of role, and the figures change often enough that any number written today is suspect within months. Take the question to local counsel, give them the contract, the length of service, and the reason for the exit, and ask for the notice required on this specific file. Treat the answer as the floor and confirm the calendar with the legal employer's HR function.
Can I just pay the person off and end it?
In some jurisdictions a properly drafted settlement can bring the employment to an end on agreed terms, but the settlement has to be drafted to the local standard, has to address the local tax position on the payment, and has to be signed in a way that survives a later challenge. A handshake and a payment isn't a settlement, and the gap between the two is where the later claims live. Where the local law restricts PILON, paying the notice as a lump sum doesn't bring the employment to an end on the same day.
What does a works council actually change?
Where a works council or similar body is in scope, the consultation is a legal stage with its own minimum timeline, its own information rights, and its own consequence for being skipped. Skipping it doesn't speed the exit up. It invalidates the exit you ran, regardless of the merit of the underlying reason. Map the consultation with local counsel before any individual conversation, and run it in the sequence the law requires.
Who terminates the worker if an EOR is the legal employer?
The EOR, as the legal employer, runs the exit on the EOR's process, on the EOR's calendar, and through the EOR's templates. You provide the business case, the timing pressure, and the in-country context. The EOR provides the legal case, the documentation, and the settlement. Where the EOR's standard isn't strong enough for the dispute risk in the file, the work is to shape the settlement with the EOR before the EOR's process starts, not to override the EOR's process afterwards.
Does my performance documentation transfer?
The content transfers, but the form has to be adapted. Performance is a valid ground in many jurisdictions, but the local court expects the file to show that the worker was told what was required, was given a real chance to improve, and was assessed against the standard that was set. A home-style file of meeting notes and informal feedback is usually not enough on its own. Rebuild the file to the local standard before the conversation, and let the documentation be the thing the conversation references.
How long does the process take?
It varies by jurisdiction, by the reason, by whether a collective body is in scope, and by the settlement window. The honest answer is that on a clean process the calendar is longer than a US-style at-will exit, sometimes considerably longer, and the delay isn't slack. Ask local counsel for the calendar on a clean process, with consultation, notice, and settlement window built in, and treat that calendar as the floor. The calendar the business would like isn't in scope.
The HROpsLab promise: independent comparison, no vendors, no advice, just the work.
