Bad offshore engagements rarely announce themselves. The pitch is polished, the rates are attractive, the case studies are glossy — and the problems that will cost you a year are all visible in week zero, if you know where to look. Here are twelve of them. None requires technical depth to check; all of them are cheaper to catch before the signature than after.
People and pricing
1. The pitch team is not the delivery team
An impressive architect runs the sales calls, then vanishes the day the contract is signed, replaced by people you have never met. Ask directly: who exactly will write our code, and can we meet them before signing? If the answer is evasive, the answer is no.
2. Blended rates
One rate for the whole team sounds simple. It also means the vendor’s margin grows with every senior they quietly replace with someone cheaper. Named people at named prices align the incentive; a blended rate is a licence to downgrade.
3. No tech-lead interview offered
If you cannot interview the person accountable for your technical decisions, either that person has not been chosen yet or the vendor suspects they would not survive the interview. Neither is acceptable. It is why our clients interview the tech lead before anything is signed.
Ownership and evidence
4. Your code in their repos
“We’ll transfer everything at the end” turns migration into a favour you negotiate when your leverage is lowest. Repositories belong in your organisation from the first commit. There is no operational reason for anything else — only a commercial one.
5. Production in their cloud accounts
If billing, root access and IAM sit with the vendor, they hold your production environment — politely, helpfully, and completely. “We manage hosting for convenience” is convenience today and hostage terms later. Your accounts, their access, revocable by you.
6. No review gates or CI evidence
Ask to see a real pull request from a real project, redacted as needed: substantive review comments, checks that block a merge, a named sign-off. A shop that cannot produce one does not have review discipline — it has a story about review discipline.
7. A portfolio without living systems
Screenshots of launches tell you nothing about year two. Ask what they still run in production today, and for how long they have run it. Building systems and keeping them alive are different disciplines, and you are buying the second one.
8. Guarantees and buzzwords
“Guaranteed delivery”, “world-class talent”, “one-stop AI transformation”. A vendor who guarantees outcomes they cannot control is showing you how they handle truth under pressure. Serious engineering firms describe practices and trade-offs; marketing departments issue guarantees.
Continuity and exit
9. No continuity policy
Ask what happens when your lead engineer leaves the vendor. “We have plenty of engineers” means key-person risk is yours to absorb. You want a written answer: how knowledge is spread across the team, and what the transition obligation actually says.
10. Vague seniority definitions
If “senior” means three years of experience somewhere, the title is a pricing device, not a competence claim. Ask what senior means at this vendor, in writing, and which people on your proposed team meet that definition by name.
11. No written handover obligations
Handover offered as goodwill is handover that evaporates when the relationship sours. The notice period should buy specified deliverables — runbooks, recorded walkthroughs, credential rotation, an open-issue register — listed in the contract before you sign, not improvised at the end.
12. Pricing that punishes leaving
Discounts that claw back on early exit, “setup fees” that become repayable, source code released only after a long payment tail. When leaving is engineered to be expensive, the vendor’s retention strategy is the exit bill rather than the work. Cheap exits are what confident vendors offer.
Incentives, not villains
Almost none of these flags mean you are dealing with bad people. They mean you are dealing with incentive structures that will produce bad outcomes regardless of who staffs them — a blended rate makes downgrading rational, vendor-held infrastructure makes lock-in rational, unwritten handover makes abandonment rational. The fix is to select for vendors whose incentives already point your way, which is a checkable property, not a feeling; we wrote about how to evaluate offshore quality in exactly those terms.
If you are mid-evaluation and want to run a real vendor — including us — against this list, book an hour with a tech lead. A technical call, not a sales call.