What it costs to test a market before opening an entity
Setup, running commitment, exit cost and management attention — the four numbers that decide whether a market test is affordable.
Somebody asks what it would cost to put two people in a market for a year to find out whether the opportunity is real. The answer comes back as a monthly rate per head, everyone does the multiplication, and the number looks either fine or impossible. Both readings are wrong, because the rate is the smallest of the four costs involved.
A market test has four cost components. Only one of them appears on a quote.
Setup cost and setup time
What you spend, and what you wait, before the first person can legally start work. Time is the expensive half: an opportunity that needed coverage this quarter does not care that the entity completes next quarter.
Running cost
Salary or fee, plus employer contributions, benefits, tooling, and the local administrative overhead of the structure you chose — statutory accounting and filings if it is your own entity, a per-employee fee if it is an EOR, a management margin if it is a provider.
Exit cost
What it costs to stop if the answer is no. Notice periods, severance, unwinding a lease or a registration, and the tail of filings an entity keeps generating until it is properly closed.
Management attention
Who inside your company now owns payroll questions, employment law they have never read, and a time zone that eats their evenings. This never appears in a budget and is frequently the largest real cost of a small market test.
Why exit cost decides more tests than entry cost
A market test is only a test if you can conclude it. The point of the exercise is to buy information cheaply — whether customers here will pay, whether the delivery model works, whether the talent is available in depth. If ending the arrangement is expensive or slow, you have not run a test. You have made a commitment and called it a test, and the sunk-cost reasoning that follows is predictable: the arrangement continues because unwinding it is awkward, long after the answer is known.
So model the exit before the entry. Ask what happens on a negative result at month nine: what notice applies, what severance is owed, what remains open, and how long full closure takes. If those answers are uncomfortable, the structure is wrong for a test even if the monthly rate is attractive.
How the four routes compare on shape
Absolute numbers depend entirely on country, role and seniority, and anyone quoting you a universal figure is guessing. The shape of each option, though, is consistent enough to be useful.
Your own entity
Highest setup cost and the longest setup time. Running cost is competitive at scale and poor at two people, because the fixed administrative floor does not shrink. Exit is the slowest and least predictable of the four. Management attention is high and permanent.
Employer of record
Fast to start and low to set up, with a predictable per-person running cost that includes the compliance work. Exit follows local employment law rather than a contract you negotiated, so check notice and severance for that specific country before signing. Management attention is moderate: employment is handled, everything else — finding the person, judging the market, integrating them — is still yours.
BPO or outsourcing
Low setup, running cost carries the provider's margin and usually a minimum commitment. Exit is contractual rather than statutory, which is easier, but the knowledge leaves with the provider's staff, so a negative result and a positive result both cost you the learning. Management attention is low day to day; the trade is control.
Managed arrangement
Setup is fast because the location and model decisions are made with you rather than by you. Running cost sits between an EOR and a provider and includes sourcing, assessment and the employment structure. Exit is contractual and defined up front. Management attention is low, because one relationship replaces a country's worth of administration.
What to actually put in the model
Five lines are enough for a decision at this size. Time to first productive day. Fully loaded monthly cost per person, including employer contributions and the administrative overhead of the structure. Total commitment if you stop at month nine, including notice and severance. Named internal owner and an honest estimate of their hours per month. And the value of the information — what the opportunity is worth if the answer is yes.
That last line is the one that reframes the exercise. A test that costs a fraction of the first year of revenue it would unlock is not an expense to be minimised; it is cheap. The instinct to minimise the monthly rate frequently produces the opposite result: a structure that is cheap per head, slow to start, and expensive to leave.
Two ways this goes wrong
The first is testing with infrastructure. An entity established to find out whether a market is worth entering answers the question at maximum cost and minimum reversibility, and it biases the answer, because by the time the data arrives the organisation has committed.
The second is testing too small to learn anything. One person, no local support, no coverage when they are on leave, and a workload that makes them a bottleneck rather than a signal. The result is not a negative answer about the market; it is no answer about the market, at nine months' cost. Size the test to produce information, then keep the structure reversible.
Buy the information, not the infrastructure. If the arrangement cannot be ended cleanly at month nine, it was never a test.