Employer of Record vs Direct Entity for International Hiring

What each model does and who carries the legal risk

The choice between an employer of record and a direct legal entity comes down to two things: headcount and how committed a company is to a given market. Most companies get the sequencing wrong in one direction or the other, and both mistakes cost real money. Deel's data puts the scale of this in perspective: 73% of companies used EOR services to expand their global workforce in 2024. The point of this piece is to lay out exactly where each option stops making sense, because that threshold is where most companies fumble.

Most companies stumble into this decision by accident, not by design. Someone finds a great engineer in another country, pays them as a US 1099 contractor because that's the fastest paperwork to process, and assumes it satisfies that country's employment rules. It almost never does. Misclassification exposure, back taxes, and benefits liability follow, often silently, until an audit or a disgruntled worker surfaces the problem years later. EOR and entity solve the same compliance problem through opposite structures, and knowing when each one wins is what keeps a scaling company from over-committing capital in one market or under-building the compliance work it owes in another.

An employer of record is a third party that becomes the legal employer for a worker in a given country. It runs payroll, tax withholding, benefits administration, and the compliance filings tied to employing someone locally, while the client company keeps control over the project, the manager, and the performance review. A direct entity reverses that structure. The company registers its own subsidiary or branch and becomes the local legal employer itself, taking on every obligation that comes with it: tax registration, payroll setup, labor law compliance, benefits design, IP assignment language, all of it.

What matters most is who signs the employment contract and who answers for it when something goes sideways. With an EOR, that's the EOR. With a direct entity, that's the company, with nowhere to point when a labor board comes calling.

An EOR is not a PEO, and mixing the two up causes real damage. A PEO splits responsibility between the client and the PEO through co-employment, and it requires the client to already have a local legal entity in place. An EOR requires no entity. It acts as the sole legal employer, full stop. That is why it works as an entry point into a country where the company has no legal presence yet.

That structure also solves a risk most companies don't think about until it becomes one: permanent establishment. Employees who habitually sign contracts on a company's behalf, or who are authorized to bind it in a given country, can trigger PE status, meaning tax authorities treat the company as having a taxable presence there even without an office or entity. An EOR sidesteps this by design. It employs the worker through its own registered entity, and the client's relationship to the EOR is a services agreement, not an employment one, which does not create PE under standard OECD treaty analysis. No cost comparison changes that math. It's a structural fact, built into how the model works, not a feature anyone is selling.

Speed to hire: why EOR wins in the first phase of any market entry

Onboarding through an EOR runs on a short, predictable clock. It involves an intake form, a compliant employment contract drafted for that jurisdiction, a signature, and local paperwork clearing, typically 5 to 14 days from signed contract to first day on the job.

Incorporating an entity in that same country runs on a different clock entirely, one measured in months, and that clock starts before a single paycheck goes out. Industry estimates put the time saved by using an EOR instead of standing up a subsidiary at roughly four months. That's a full quarter of missed revenue, missed hiring, or a missed window against a competitor moving faster, and in a competitive hiring market, four months is often the whole game.

Speed matters most in three situations: testing a new market before committing capital to it, hiring a specific person before someone else does, and staffing a project with a defined end date that doesn't need permanent local infrastructure behind it. The EOR moves fast for a simple reason: its legal entity is already sitting there, registered, ready to use. It isn't promising a faster process. It already has the infrastructure built.

None of that speed comes free, though. The monthly EOR fee is the price paid for infrastructure the company didn't have to build itself, and that fee compounds over time in a way that eventually flips the math against it. Which raises the real question: at what headcount does the EOR's convenience start costing more than the entity's overhead would have?

The real cost comparison: EOR fees vs. entity overhead, by headcount

EOR pricing generally runs in a few-hundred-dollar-per-employee-per-month range, though some providers charge a percentage of gross salary instead, typically 10 to 15%. Named pricing spans a real range: Knit People runs $199 per employee per month, Deel charges $599, and Remote lists $699. That gap runs several times over between cheapest and most expensive, so shopping around actually matters here, and most companies don't bother.

Entity setup and running costs for the first year typically start around $20,000 and climb much higher depending on the country. A UK example makes the gap vivid: entity setup there runs an estimated $78,000 to $128,000, against an estimated EOR cost of $7,188 for the same hire. For one employee, that isn't a close call.

The crossover point, where the entity starts winning on a per-seat basis, generally falls around 4 to 8 employees in a given country over a 3-year horizon. Below that range, EOR is cheaper, full stop. Above it, the entity's fixed costs spread across enough people that the per-head math flips in its favor. Entity overhead isn't just the incorporation filing, either. Deel's cost breakdown for direct hiring includes licensing and permits, annual filings and audits, local payroll vendor fees, mandatory insurance contributions, HR administration, and ongoing local legal and accounting counsel, costs that recur every year the entity exists, not just in year one.

That crossover number needs adjusting for each country's labor costs, administrative burden, and mandatory benefit contributions, all of which vary sharply from place to place. Six employees in Germany and six in Brazil can produce very different answers. What holds steady across markets is the failure pattern, and it runs in one direction far more often than the other: companies that keep paying EOR fees well past the crossover point are overpaying for convenience they no longer need. Companies that rush to incorporate before reaching it are locking capital into fixed overhead they haven't earned yet. Both mistakes carry real cost, because nobody notices a slow leak in monthly subscription-style fees the way they'd notice a bad six-figure incorporation bet.

Direct entity advantages: control, compliance prerequisites, and strategic commitment

Three conditions tip the scale toward a direct entity. Headcount above the crossover threshold in a market the company intends to stay in for the long run is the first and most straightforward. The second is a genuine need for control: custom benefits design, bespoke IP assignment language, data residency terms that satisfy enterprise clients, none of which an EOR's standardized contract templates can fully replicate. The third is intent, plain and simple: building something durable in that country rather than treating it as a trial run.

Some industries don't get to treat this as an economic calculation. Financial services, healthcare, and government contracting frequently require a local legal entity just to hold a license or sign a client contract, regardless of headcount. The entity isn't optional past some threshold in those cases; it's a regulatory gate. An EOR still has a role even here, though, as a bridge while incorporation is underway, letting the company hire the first person or two before the entity paperwork clears rather than leaving a role empty for months.

EOR structurally cannot compete on control, no matter how good the provider is. Custom benefits packages, tailored IP assignment clauses, data-processing terms built for a specific enterprise client's procurement requirements, a distinct local employer brand: none of that fits into a third party's standardized employment contract. And once a company puts down actual physical roots, leasing office space, building out a local team with equipment and facilities, the EOR model starts to strain against its own design. It was built for distributed, employment-only relationships, not for a company setting up a real operational footprint on the ground.

None of this needs to happen all at once. A sensible path starts on EOR and graduates to an owned entity once headcount justifies the fixed cost. Done well, that transition keeps employees on the same underlying systems without forcing them through a fresh onboarding process. It's a financial event on the company's balance sheet, and the employee should never notice it happened.

The compliance layer that makes both models harder than they appear

Neither model is a decision made once and forgotten, because the regulatory rules governing both keep shifting under everyone's feet. Deel's global payroll compliance checklist tracks more than 30 countries that updated payroll, employment tax, or mandatory benefits rules between 2025 and 2026, roughly a third of the countries most companies would consider hiring in, changing the rules mid-game.

Strada's 2025 Global Payroll Complexity Index, now in its seventh edition, shows the global average complexity score climbing from 5.55 in 2023 to 5.68 in 2025. France, Slovakia, and Australia hold the top three spots for complexity, and Europe accounts for 7 of the top 10 positions overall. One country enters the global top 10 for the first time in that index on the strength of 51 separate state-level jurisdictions, each running its own rules, and that complexity only gets amplified by the multi-state exposure that comes with remote work.

Digital reporting requirements are spreading too. Brazil's eSocial system and Mexico's CFDI digital payslip mandate are already established, and similar frameworks are tightening or rolling out elsewhere. The failure point behind most compliance breakdowns traces back to something almost mundane: a missing or outdated work-location record. One untracked move by a remote worker creates four separate compliance touchpoints at once, tax registration, unemployment insurance, employment-law updates, and a data privacy assessment. PayrollOrg's survey of global payroll professionals found 57% ranked ensuring local compliance as their single biggest challenge, above other operational concerns.

Whichever model a company picks, that compliance burden lands somewhere. It shifts to either the EOR's compliance team or to internal HR and finance staff who now have to track it themselves, cycle after cycle.

The effects of handling compliance manually, regardless of which model you choose

Payroll run by hand doesn't hold up at scale, and the numbers make that plain. Lano's research puts average payroll accuracy at just 78%, meaning roughly 22% of payroll transactions carry some kind of error. Scaled to 1,000 employees paid monthly, that works out to around 2,640 potential errors a year, and at an estimated $291 per correction, that's about $768,240 spent annually fixing mistakes that proper handling would have prevented.

Errors compound into legal exposure fast. Lano's data shows 14% of companies facing litigation or compliance issues tied directly to payroll errors, with average annual legal and compliance costs around $13,000 and 120 hours of lost productivity per incident. Regulatory penalties add another layer on top: Riseworks' Global Payroll Compliance Report puts the cost of non-compliance at more than $7 billion annually in IRS penalties alone, with penalty rates running 2% to 15% of unpaid tax. A Paycom survey of 1,000 HR and finance professionals found 1 in 3 employers had been penalized for noncompliance in the prior year.

Plain labor cost drives most of that. Payroll administrators spend 3 to 8 hours per payroll cycle on manual data entry alone, before corrections, approvals, or compliance prep even enter the picture, and that number only climbs as more jurisdictions get added to the mix. Choosing between EOR and entity answers who the legal employer is. It says nothing about whether payroll and compliance actually get executed correctly, on time, every single cycle, and that gap is where the money leaks out.

How AI-native platforms change the calculus for both models

EOR and entity settle who the legal employer is. Execution, the actual running of payroll and the tracking of compliance obligations, sits on a separate layer entirely, and that layer is what automation is starting to reshape. BCG's global study found that shifting from manual to systems-based compliance automation could generate tens of billions of dollars in annual savings, a figure that says less about any single vendor and more about how much manual labor still sits inside compliance workflows across the whole industry.

AI agents that own a workflow end to end, opening a state tax account, resolving a compliance notice, walking an employee through benefits enrollment, provisioning a laptop, remove the manual handoff points where payroll errors and penalty exposure actually occur. For a company running its own entity, a platform that tracks tax obligations across 10,000-plus jurisdictions and processes payroll across all 50 US states removes the need to hire specialized compliance staff as headcount grows state by state. For a company on EOR, the EOR covers local employer obligations, but the client still needs its own payroll infrastructure for domestic employees, for contractors elsewhere, and eventually for the entity transition when that day comes. One platform spanning both scenarios means employees don't get re-onboarded onto a new system every time the underlying legal structure changes underneath them.

G-P's platform is a useful reference point here: it applies agentic AI to manage the employee lifecycle and generate locally compliant contracts across more than 180 countries, and it points toward where this category is headed. None of it replaces the actual choice between using an EOR and setting up a local entity, though. Choosing an EOR keeps the benefits of automation visible in day-to-day operations, while setting up a local entity buries them under manual busywork instead.

A decision framework: which model to use at each stage of international growth

Early on, testing a new market with one to three hires, EOR wins on every axis that matters: speed (five to fourteen days to get someone working), cost (no entity overhead to amortize), and risk (permanent establishment exposure eliminated by the structure itself). Setting up an entity at this stage is capital sitting idle in overhead nobody needs yet. It's capital sitting idle in overhead nobody needs yet.

As headcount grows toward the crossover range, roughly four to eight employees per country, the right move is to model the 3-year total cost with entity overhead spread across the headcount trajectory that's actually likely, using that trajectory instead of the one in the optimistic deck. If that trajectory holds up, start entity incorporation while the EOR keeps running payroll in the background. Waiting until the math has already flipped means paying EOR fees for months longer than the company needed to.

Once a company has an established presence past the crossover point, the entity wins on cost per seat, and the control benefits, custom contracts, IP assignment, data residency, actual local benefits design, start mattering operationally rather than theoretically. At that stage, employees transition from EOR to entity payroll, and a well-run EOR provider handles that switch without forcing anyone through onboarding twice.

A few conditions override this staged logic. Regulated industries, financial services, healthcare, government contracting, often need a local entity for licensing purposes regardless of headcount. In these cases, companies should skip the early stage and use EOR strictly as a bridge while incorporation is underway. Physical offices or facilities call for an entity at any headcount, because the company is building operational roots, not just hiring people remotely. And a short-term project or a time-boxed hire justifies EOR at any headcount, because the commitment horizon never warrants the fixed cost of an entity.

Most scaling companies end up running a portfolio where different countries sit at different stages at once: EOR in markets just opening up, an owned entity in markets that matured years ago. That's the normal condition of a company actually growing internationally, not a sign that anything went wrong.

The right answer for two employees in Germany is not the right answer for twelve, and the trigger for that shift needs to live inside the operating model itself, not get rediscovered by accident during a budget review. The choice between using an EOR and setting up a local entity comes down to headcount and commitment, nothing more mysterious than that. Knowing precisely when each model wins is what lets a company grow into new markets without over-committing capital it doesn't need to spend yet, or under-building the compliance infrastructure it can't afford to skip.