Most guides to hiring an offshore development team in India are written to make offshore sound frictionless. It is not frictionless. It is a management change as much as a cost change, and the projects that fail almost always fail on the client side, on things that were decided before anyone wrote code. This is the sequence, in order, with the parts that go wrong marked.
The short version
Pick the engagement model before you shortlist. It decides who you should even be talking to.
Ask which stack a vendor would refuse. A firm that refuses nothing has no specialism.
Get the overlap in your timezone, in writing, as hours and not adjectives.
Pay for a two week pilot on a real ticket. Never take a free trial.
Settle IP assignment and the W-8BEN-E before the first invoice, not during diligence two years later.
Budget two to four hours a week of your own senior time. If you cannot, buy delivery instead of engineers.
What an offshore development team actually is
Strip the brochure language and there are three components: engineers working your board from another timezone, a contract that assigns the code to you, and one named person accountable when a date slips. Everything else is packaging. The word "team" is doing a lot of work in most proposals, so establish early whether you are buying named individuals or access to a pool.
The distinction that matters most is dedicated versus shared. A dedicated engineer works one client at a time. A shared one is billed to three and will be reassigned without a conversation. Both get sold as "your team", and the difference shows up in month two.
Ask about people
Which named engineers, and how many other clients does each carry right now?
Ask about paper
MSA, SOW, IP assignment on payment, NDA before code access, W-8BEN-E on file.
Ask about hours
Start and end time in ET or GMT. "Flexible" is not an answer to this question.
Decide the model before you shortlist
This is the step people skip, and skipping it is why shortlists end up comparing a two person studio against a six hundred person firm as though they were the same purchase. Four models get sold under the same words.
Our own bias, stated plainly: we are the second row, so this table is not neutral. If you need ten engineers in six weeks, the third row is the correct answer and we would be wasting your time. The wider comparison sits on our offshore web development page, including the cases where offshore is the wrong call entirely.
Where to look, and what a shortlist should cost you
Directories are a starting point and nothing more. Clutch and similar sites rank partly on review volume, which correlates with sales effort rather than engineering quality. Read the reviews for specifics — a named project, a metric, a problem that was handled — and discard the ones that only say the team was communicative.
Better sources, in order: someone in your network who has actually shipped with the vendor, public code you can read, and conference talks or writing by the named engineers. If a firm publishes nothing and shows no code, you are buying entirely on the sales conversation.
Three to five vendors is the right shortlist. Ten wastes six weeks and produces worse decisions, because nobody reads ten proposals carefully.
Score the answers as you get them. Memory flatters whoever presented last.
The eleven questions that sort the field
These are the questions we are asked by the buyers we most want to work with, and the ones that make weaker vendors uncomfortable. Ask all eleven on the first call and take notes.
Question eight is the one that separates most reliably. A vendor with a real process can send you the questions in ten minutes because the questions already exist. A vendor without one sends a slide about culture fit.
"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."
Run a paid pilot, not a free trial
A free trial gets staffed with whoever is spare, because nobody assigns their best engineer to unpaid work. A paid two week pilot gets staffed with the person who will stay on the account. That inversion is the single highest leverage thing in this whole process, and it costs about one percent of a year of engagement.
-
01
Pick a real ticket, not a sample project
Something from your actual backlog that touches your real codebase, your real API and your real deployment. Sample projects test nothing you care about.
-
02
Agree the exit criteria in writing first
Merged to main, tests passing, reviewed by your senior developer, plus a written note on what they found in the codebase. Ambiguous criteria make a pilot unfalsifiable.
-
03
Give repo access on day one
If you cannot get an external contributor into your repository in a day, that is your bottleneck and better to discover it now.
-
04
Name one reviewer with a 24 hour target
A PR sitting from Friday to Tuesday costs three days of a two week pilot. This is the most common way clients sabotage their own trial.
Where our interests differ from yours
We want the pilot fee. You want the option to walk. Resolve it by agreeing up front that the fee is refunded if you do not continue, and that you keep the merged code either way. Any vendor unwilling to write that down is telling you something.
Contracts, IP and the W-8BEN-E
None of this is exotic. It is listed because most vendor sites do not list it, and your legal and finance teams will ask in week one regardless.
Two clauses matter more than the rest. First, IP assigns on payment and the individual engineers sign the same assignment — a company level clause with contractors who never signed anything is the gap that surfaces during diligence. Second, exit: thirty days notice either side, with the final week as handover, documented and recorded.
On payment, a W-8BEN-E on file before the first invoice lets a US company pay an Indian entity without withholding surprises. It takes ten minutes and it removes a finance objection that otherwise appears at the worst moment. Repository and cloud accounts should be yours throughout, with the vendor added as a member rather than an owner — GitHub organisation roles make that distinction explicit.
Onboarding, week one
The first week sets the ceiling for the whole engagement. Front load access and context, because every hour of week one blocked on a credential is an hour billed for waiting.
What goes wrong, and the early signals
Four of these five are client side. That is not vendor deflection, it is what the failures actually look like from inside.
What it costs, honestly
Rate is the number everyone compares and the least useful one. Total cost is rate plus your review time plus ramp up plus rework, and the cheapest rate reliably produces the largest last two terms.
Marketplace
$8 to $12 / hr
Shared engineer across several accounts, high turnover, minimal vetting. Expect to rewrite.
Agency
$18 to $28 / hr
Mid tier agency, real process, account manager between you and the code.
Studio
$28 to $48 / hr
Dedicated senior engineer, one client, direct contact. Our own band, stated for comparison.
Onshore
$90 to $160 / hr
US or UK contractor. Same timezone, no overlap problem, four times the cost.
Add two to four hours a week of your own senior engineer to whichever row you pick. At a US salary that is a real line item, and it is the one that makes the bottom row closer to the middle rows than the rate table suggests.
Questions people ask
Where to go next
If you have decided offshore is right and you want the vendor side written out — how firms are structured, which services are worth buying, and the four situations where offshore is the wrong answer — that is our offshore web development page. If you are mid shortlist, the eleven questions above are the fastest way through it. Send us the ticket you would hand over first and you will have a written approach inside two working days, from the engineer who would do the work.
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
Scored within Group A only. No engineering posts surface here, by design, so a procurement reader is never handed a React internals piece.