Every growing company that spends enough with offshore vendors eventually builds a spreadsheet comparing vendor margin against the cost of hiring the same people directly. The spreadsheet almost always favours building your own. It is almost always missing about half the cost. That does not make the conclusion wrong. Past a certain size it is clearly right, but the threshold is much higher than the arithmetic suggests, and we lose clients to it, which is the reason to be careful about how this is written.
The short version
The threshold is roughly fifty engineers offshore. Below that, vendor margin is cheaper than the overhead of running an entity.
The costs missed in the spreadsheet are management, HR, compliance, real estate, attrition and the eighteen months before it works.
A captive centre is a company you are founding in a country you do not live in. That is the actual decision.
There are middle options: build-operate-transfer, employer of record, or a vendor with named long-term staff.
The reason to build is usually control and retention, not cost. When cost is the only argument, it is often the wrong call.
Where the line actually sits
Not legal, tax or employment advice
Opening an entity abroad touches company law, employment law and tax in a country your existing advisers may not cover. What follows is what we have watched happen from the vendor side of these decisions, and nothing more than that. The thresholds and timelines are useful for framing the question. The answer needs people qualified in the country you would be setting up in.
The vendor margin you are trying to escape is real: somewhere between twenty and forty per cent on top of salary, depending on the vendor. On twenty engineers that looks like a large annual number, and it is. The question is what that margin is buying, and whether you can produce the same thing more cheaply yourself at your current size.
Fifty is not a magic number and nobody should treat it as one. What sits behind it is that a captive centre needs its own leadership: a site lead, someone owning hiring, someone owning delivery. That layer costs roughly the same whether it supports fifteen engineers or eighty. Below about fifty, you are paying for that layer out of too few heads, which is precisely the structure vendor margin exists to spread.
Below that line the comparison is really vendor margin against your own overhead, and the vendor side of it is broken down in what an offshore team really costs.
What the spreadsheet leaves out
These are the lines we see omitted most often. None of them are exotic and all of them are recurring.
Add those honestly and the twenty-to-forty per cent margin stops looking like pure profit and starts looking like the price of not doing any of it. Past fifty heads you can do it more cheaply. Below thirty, you are usually buying yourself a second job.
The reasons that actually justify building
When clients who moved to a captive centre are happy about it two years later, cost is rarely the reason they give. These are.
Retention of context
Your own employees stay on your product for years. Vendor staff rotate, and each rotation costs domain knowledge nobody wrote down.
Control of hiring
You choose every person against your own bar, rather than accepting who a vendor allocates.
Strategic permanence
If the product is the company, having its engineering inside the company matters in ways a spreadsheet does not capture.
Where our interests differ from yours
We are the vendor. When a client crosses fifty heads and builds their own centre, we lose the account, so treat our threshold with appropriate suspicion and pressure-test it. What we would ask you to check is not the number but the reason: if the only argument is margin, the case is weaker than it looks. If the argument is retention and control, it is stronger than the spreadsheet says.
Retention of context is the honest reason to build. It is also achievable contractually, which is the third option below and something the contract checklist covers as named staff.
The middle options nobody mentions
The comparison is usually framed as vendor versus captive, which is a false binary. Three arrangements sit between them, and for most companies in the thirty-to-fifty band one of them is the right answer.
The third is worth trying before either of the others, because it addresses the actual grievance at almost no cost. Most companies who say they want a captive centre want their engineers to stop changing. That is a contract clause, not an entity.
"A captive centre is not a procurement decision. It is founding a company in a country you do not live in, and it should clear the bar that implies."
Whichever arrangement you pick, test it on real work first. A two week paid pilot costs less than a year of finding out.
What the first eighteen months actually look like
Business cases tend to model the steady state and skip the approach to it. The approach is where the money and the disappointment live, so it is worth setting out plainly.
The site lead hire deserves particular attention because getting it wrong resets the clock. A weak site lead hires a weak team, and by the time that is visible from headquarters you are twelve months in with salaries running. Companies that succeed at this usually either relocate someone trusted for a year or pay well above local market for a leader with a track record they can verify personally.
"The business case is usually right about the steady state and wrong about the eighteen months in front of it, which is where the money and the patience get spent."
Eighteen months is also long enough that the overlap question has to be answered twice: once for the vendor, once for the team you build.
Questions to answer before committing
If the answer to any of these is unclear, the decision is not ready, which is different from the decision being wrong.
The third question catches more companies than the others. A captive centre inherits whatever your engineering practice actually is, not what your careers page says it is. Teams with weak documentation, informal review and tribal knowledge tend to reproduce all three offshore and then conclude that offshore does not work.
Questions people ask
Where to go next
If the answer is to stay with a vendor for now, the thing to fix is usually the contract rather than the vendor, and the contract checklist covers named staff and exit terms. If you are still comparing vendors, the real cost breakdown has the total-cost arithmetic, and the offshore web development page sets out how we structure a long engagement.
Sources
Yogesh Jadhav
Founder of reactdevstudio. Ten years as a working full stack developer, across SaaS products and client work.
Related, Group A only
Three more on offshore work
Procurement and process posts only, scored within Group A, so a finance reader is never handed an engineering piece.