Offshore, Nearshore, Onshore: The Decision Minus the Sales Pitch

Offshore, nearshore, onshore. Three words that get treated as a strategic framework and are in fact a marketing taxonomy, invented mostly by people who wanted a category to sell you.

Think about what they actually mean. Nearshore means "geographically near you." Offshore means "far away." Onshore means "here." None of those describe anything you care about operationally. You do not care about kilometres. You care about whether you can get an answer before end of day, what the work costs, and whether your lawyer is going to have opinions.

So let us throw the words out and rebuild the decision from the variables that actually determine whether this works.

The four variables that matter

Every genuinely useful version of this decision comes down to four things. Everything else is decoration.

1. Timezone overlap, measured in hours. Not "similar time zone." Hours. This is the single most predictive variable and the one that gets described most vaguely in vendor materials.

2. Communication bandwidth the work requires. Some work needs constant back and forth. Some needs a brief and a deadline. These are different jobs and they want different setups.

3. Cost. Obvious, frequently overweighted, and almost never calculated properly. The rate is not the cost. We went through that in detail in the hidden costs of outsourcing.

4. Regulatory and contractual exposure. Data residency, employment law, IP assignment, contractor misclassification. Boring right up until it is the only thing anyone is talking about.

Notice that geography appears in none of these. Geography is a proxy for time zone, and a bad proxy at that, because longitude matters and latitude does not. Cape Town and Berlin are eight thousand kilometres apart and share a working day. Los Angeles and Bangalore share almost nothing and everyone still calls one of them "offshore" as though that were the useful fact.

Reframe it: how many hours of overlap does this specific work need?

This is the question. Ask it about the specific piece of work, not about your company in general, because different functions in the same company have wildly different answers.

Work that needs four or more hours of overlap

Tightly coupled product work. Anything where requirements are still being discovered. Early-stage engineering where the design changes based on what you learn on Tuesday. Incident response. Sales where the outsourced person is talking to your customers during their business hours. Any function where the outsourced person is embedded in a team rather than delivering to it.

The tell is whether progress depends on a chain of small clarifications. If a typical week involves eight moments where someone needs a quick human answer to keep going, and your overlap is ninety minutes, you have built a system that produces one day of progress per week of calendar time. Everyone will blame the team. The team is fine. The structure is broken.

Work that needs one to three hours

Most ongoing operations. Bookkeeping, support with defined escalation paths, CRM hygiene, reporting, recruiting coordination, marketing execution against an agreed plan. Mature engineering work on a stable codebase with good documentation.

An hour or two a day is enough for a standup, a decision, and an escalation. The rest is async and it is genuinely fine.

Work that needs almost none

Well-scoped delivery projects with clear acceptance criteria. Design work with a solid brief. Data processing. Annotation. Content production against a style guide. QA against a written test plan. Anything you can specify in advance and evaluate on receipt.

For this category, timezone overlap is nearly irrelevant and optimising for it is a waste of money. If you are paying a nearshore premium to have a designer three hours from you produce assets against a brief you wrote in full, you have bought convenience you are not using.

Here is the uncomfortable version of this. Most companies think their work needs four hours of overlap. Most of it does not. It needs one hour of overlap and a better scope document, and the reason it currently needs four is that nobody wrote things down. That is a documentation problem you are solving with a geography budget, which is an expensive way to solve a documentation problem.

The map, in overlap hours

Assume a standard 9 to 6 working day at both ends, with a reasonable person willing to shift an hour either way but not to work nights. Overlap is approximate and moves an hour with daylight saving.

Region	UTC offset	Overlap with Western Europe (CET)	Overlap with US East (ET)
UK and Ireland	UTC+0/+1	8 hours	5 hours
Iberia, Western Europe	UTC+1/+2	9 hours	4 hours
Poland, Romania, Baltics	UTC+2/+3	8 hours	3 hours
North Africa (Egypt, Morocco)	UTC+0 to +3	7 to 8 hours	3 to 4 hours
East Africa (Ethiopia, Kenya, Rwanda)	UTC+3	7 hours	3 hours
Turkey, Georgia	UTC+3	7 hours	3 hours
West Africa (Nigeria, Ghana)	UTC+0/+1	8 hours	4 hours
South Africa	UTC+2	9 hours	4 hours
UAE, Gulf	UTC+4	6 hours	2 hours
Pakistan	UTC+5	5 hours	1 hour
India	UTC+5:30	4 to 5 hours	1 hour or less
Vietnam, Thailand	UTC+7	3 hours	none
Philippines	UTC+8	2 hours	none
Mexico, Colombia	UTC-5/-6	2 to 3 hours	7 to 8 hours
Brazil, Argentina	UTC-3	4 to 5 hours	7 hours
US and Canada	UTC-5 to -8	2 to 4 hours	8 hours

Look at that table without the labels attached and the conventional wisdom starts to look silly. India is described as offshore for everyone, but it has more overlap with Central European time than Mexico does, and Mexico is the canonical nearshore recommendation. Vietnam and Poland get filed under the same heading in half the vendor decks you will read, and one of them shares three hours with your day while the other shares eight.

The arbitrage nobody is using

Here is the part I find genuinely strange.

East Africa sits at UTC+3. For a company anchored in London, Amsterdam, Berlin, Paris, Madrid or Stockholm, that is a seven to eight hour shared working day. Full standups. Same-day answers. A colleague you talk to rather than a mailbox you write to. Operationally, that is nearshore. It is a better overlap than a European company gets with Mexico, and roughly the same as they get with Warsaw.

And it is priced as offshore. Because someone decided decades ago that Africa is far, and "far" got translated into a pricing tier, and nobody has updated the spreadsheet.

That gap is the most underused piece of geographic arbitrage available to a European company right now. You get nearshore collaboration economics with offshore cost structure, in English, with a large university-educated talent pool that has not been bid up by fifteen years of foreign companies competing over the same few thousand people.

Two honest caveats, because I am not selling anything.

First, the vendor ecosystem is younger than in Poland or India. There are fewer companies with twenty years of process maturity, so you have to be more careful in selection and less able to rely on brand as a shortcut. That is a real cost in due diligence hours. Some serious operators exist across the continent, from Andela and Gebeya on the talent marketplace side to Sama, CCI Global, Techno Brain and Sand Technologies in delivery and BPO, but the layer beneath the recognised names varies a lot.

Second, infrastructure varies by country and city, and you should ask specific questions about connectivity and power redundancy rather than assuming. A serious provider will have a specific answer involving backup power and secondary connections. A vague answer is itself the answer.

For US East Coast companies the picture is different but not useless. Three hours of overlap with UTC+3 means an East African team's afternoon meets a New York morning. That is thin for embedded product work and completely adequate for operations, delivery and anything with a written brief. For a US company that genuinely needs six or more hours of overlap, Latin America is still the correct answer and I would not pretend otherwise.

Cost, honestly

Broad, directional ranges. These move, they vary enormously by seniority and specialism, and anyone quoting you a precise figure for an entire continent is guessing.

The pattern you consistently see: US and Western European rates at the top, Eastern Europe and Latin America clustered in the middle at roughly half to two thirds of Western rates, and South Asia, Southeast Asia and Africa at the lower end, often a third to a half of Western rates for comparable seniority.

Two things worth knowing about those numbers:

The gap narrows sharply at senior levels. A genuinely excellent senior engineer or finance lead is expensive everywhere, because they have global options and they know it. Arbitrage is largest at mid-level and shrinks as seniority rises. If your plan depends on hiring a world-class principal engineer at a quarter of the price, your plan is a wish.

Rates in a region track how long foreign buyers have been competing there. Eastern European rates are what they are because two decades of Western demand reset local expectations. The same process is running everywhere, just at different points on the curve. That is the entire arbitrage: buying before the repricing, not after.

Regulatory exposure, the part that gets skipped

This is where the offshore and nearshore labels are most misleading, because the legal picture has nothing to do with distance.

Three questions decide most of it.

Where does the data live and who may touch it? If you handle EU personal data, GDPR follows the data, not the border. Transfers outside the EEA need a legal basis and appropriate safeguards. This is workable in almost any country. It is not workable if you have not thought about it and your vendor's answer is "we are very secure."

Are these contractors or employees? The single most common expensive mistake in cross-border hiring. If you direct someone's daily work, set their hours and they work only for you, most jurisdictions consider that employment regardless of what your contract calls it. Misclassification penalties are real and they compound.

Who owns the intellectual property? IP assignment must be explicit, in writing, valid under the law of the country where the work is performed. A US-style assignment clause does not automatically do what you think it does everywhere.

You have three ways to handle all of this. Use an employer of record such as Deel or Remote.com, which solves compliance completely and charges per person per month indefinitely. Set up a local entity, which makes sense past roughly fifteen people in one country and is otherwise a distraction. Or work with a partner who already runs payroll, contracts and compliance in that market as part of the arrangement.

That third option is worth understanding properly because it is priced very differently. A recruitment-led partner such as Zemenay Tech, which operates out of Addis Ababa across African and global markets, handles payroll, compliance, onboarding and time tracking while charging a one-time placement fee rather than a per-head monthly subscription, which means the compliance overhead is bundled into a cost that stops rather than one that runs forever. Whether that beats an EOR depends almost entirely on headcount and duration: EOR economics are excellent for two people for eighteen months and poor for twelve people for five years. Do the multiplication before you pick, because the two curves cross earlier than most people expect.

What you must not do is improvise it. "We just pay them as contractors and see how it goes" is a strategy that works perfectly until the day it does not, and the day it does not is usually the day you are trying to close a funding round or an acquisition.

A decision procedure that takes twenty minutes

1. Write down the specific function. Not "engineering." Specific.
2. Count the overlap hours it genuinely needs. Be honest and be sceptical of your own first answer. Ask what would break with two hours instead of five. If the answer is "people would have to write things down," that is not a reason.
3. Filter the table by that number. You now have three or four candidate regions instead of a world map.
4. Apply your regulatory constraints. Data residency and IP rules will often remove a region outright. Better to find out now.
5. Compare cost across what is left, loaded, not quoted.
6. Then pick the vendor, which is a harder decision than the region and is where projects actually fail. The vendor selection process is worth doing properly.

If you do this and end up somewhere conventional, good. The point is not to be contrarian. The point is to know why, so you can defend the decision when it gets hard in month seven.

Frequently asked questions

Is nearshore always better for collaboration?

No, it is better for time zone overlap specifically, which matters only in proportion to how much synchronous collaboration the work genuinely requires. If your work is well specified and delivered against acceptance criteria, you are paying a premium for a feature you are not using. Buy overlap when the work needs it, not by default.

We are a US company. Does East Africa make any sense for us?

For operations, back office, delivery work and anything driven by a written brief, yes. Three hours of overlap covering your morning is adequate for that, and the cost difference is significant. For tightly coupled product engineering where your team is discovering requirements as it builds, three hours is thin and Latin America is the better answer. Match the region to the work, not to the company.

How much overlap do we actually need for a support function?

It depends entirely on whether the support is live. If customers expect real-time chat during your business hours, the team has to be awake during your business hours, full stop. If support is ticket-based with a response time target, you need enough overlap for escalation and coaching, and one to two hours is workable. Some companies deliberately use the offset to cover nights, which turns the drawback into the feature.

Is onshore ever the right answer?

Yes, more often than outsourcing enthusiasts admit. When the work requires deep regulatory context, physical presence, security clearance, or an in-person client relationship, keep it local. Also when the function is genuinely core to your differentiation and you would be renting out the thing your company is supposed to be uniquely good at. Cost is not the only variable and pretending otherwise is how you save fifty thousand dollars and lose a capability.

Does a mixed model work, or is it a mess?

It works, but only with clean boundaries. One region owns one function end to end, with its own definition of done. Splitting a single tightly coupled workflow across two distant time zones is where it turns into a mess, because now the handoffs are the job. Split by function, not by task.

The one-line version

Stop asking which of the three categories you want. Ask how many hours of overlap the work needs, filter by that, check the legal exposure, then compare loaded cost.

Do that and you will notice, as most people eventually do, that the map everyone else is using was drawn by someone selling seats on it.