What an offshore development team in India really costs

Every comparison we are asked to take part in starts with hourly rates in a spreadsheet, and almost none of them predict what the engagement actually costs. Rate is the most visible number and the least useful one. What follows is the full picture, including the costs that land on your team rather than your invoice, and a worked example where the cheaper rate ends up more expensive.

The rate bands, stated plainly

These are the bands we see quoted in 2026 for Indian vendors selling to US and UK buyers. Our own band is in there and marked, because a cost article that hides the author's price is not much use.

The spread inside India is wider than the spread between India and Eastern Europe, which is the first thing that should tell you the country is not the variable doing the work. What you are buying at the top of the range is a senior engineer who is not shared, and a shorter path between you and the person writing the code.

The four costs the rate does not include

None of these appear on an invoice, which is precisely why they get left out of the comparison and then dominate the outcome.

The arithmetic, worked through

Take one engineer for six months, 160 hours a month, and count your own senior engineer at a fully loaded $110 an hour. The only variables are the rate, the share of output that needs redoing, and how much of your time gets consumed.

The marketplace column is not a trick. It is genuinely cheaper in absolute terms, and for a bounded task with a strong internal reviewer it can be the right answer. What the arithmetic shows is that the four-times gap in the rate card becomes a 1.6-times gap in reality, and that the difference is paid in your senior engineer's calendar rather than your budget line, which is usually the scarcer resource.

"An eight dollar an hour developer is a deferred invoice. You pay the difference later, in rewrites, in review time, or in the six weeks it takes to find out."

What actually moves the number

When we quote a range and then quote a narrower one after a scoping call, these are the things that moved it. None of them are the number of screens.

Making it genuinely cheaper

These are the levers that reduce total cost rather than moving it somewhere less visible. We would rather you used them even though several of them reduce our invoice.

The lever that does not work is squeezing the rate. Below roughly $25 an hour for a dedicated senior engineer in India, something is being subsidised: the engineer is shared, junior, or will not be there in four months. That is a fine trade for some work and a poor one for a core product, but it should be a decision rather than a discovery.

The first month costs roughly double

This is the single most under-modelled part of any offshore budget, and it is not a criticism of anyone. Month one produces perhaps half the output of month three at the same invoice, and that gap is real work rather than slack.

Two consequences follow. First, a three-month engagement is a much worse deal per unit of output than a nine-month one, because the fixed ramp-up cost is amortised over a third of the time, which is why we will usually say that a six-week project is better given to whoever already knows your codebase. Second, and more importantly, engineer turnover is expensive in a way the rate card cannot show you: every replacement pays this month again.

When you compare vendors, ask what their average engineer tenure on a single client is. At the cheap end the honest answer is often four to six months, which means you buy month one twice a year, at full price, forever.

Payment terms, currency and the small leaks

None of these will change a decision on their own. Together they move a budget by a few per cent, and they are all resolvable in one conversation before the first invoice rather than five conversations after it.

The payment-terms line is worth more attention than it usually gets, because it is one of the few places where being a good client is directly cheaper. A vendor who is paid reliably on net 15 has no working capital risk to price in. A vendor chasing invoices at net 60 does, and it shows up in the rate at renewal.

Questions people ask

Where to go next

If the total-cost view makes sense and the question is how to test a vendor before committing six months, read running a two week paid pilot. If you are weighing a vendor against building your own team offshore, the captive centre comparison has the headcount threshold where that flips. Our own rates and what sits behind them are on the offshore web development page.