Nearshore vs offshore staff augmentation: the collaboration cost nobody puts in the rate

Every offshore rate card wins the first meeting. A senior engineer in Pune or Manila at half the Mexican rate, a quarter of the US rate, and a vendor with twenty years of references. The rate is real. What the rate card leaves out is a set of costs that never appear on an invoice: the standup nobody attends, the pull request that waits a day for a reply, the manager who now works two shifts, and the engineer who leaves in month nine. This post puts numbers and sources on those costs, compares Mexico with the offshore hubs on each, and ends with a framework for deciding which model is right for a given team. We sell nearshore, so weigh the argument accordingly; every third-party figure is linked and dated.

What is the difference between nearshore and offshore staff augmentation?

Nearshore staff augmentation places engineers in a country that shares most of your working day and is a short flight away; for a US company that means Mexico or, more loosely, Latin America. Offshore places them eight to thirteen time zones away, typically India, the Philippines or Eastern Europe. The model is identical; the collaboration economics are not.

Both models add engineers to your team who take direction from your managers, which is what separates staff augmentation from managed services and outsourcing. The distinction between nearshore and offshore is purely geographic, and geography drives four costs: overlap, rework, management load and attrition. Everything below is an attempt to price those four.

How much cheaper is offshore than nearshore, really?

On the hourly rate, offshore is typically 40 to 60% below nearshore Latin America and nearshore is 30 to 57% below a comparable US hire. On total cost the gap narrows or inverts once you add the unbilled hours: async delay, rework, management overtime and replacement after attrition. Which side wins depends on how much live collaboration the work needs.

Start with what the rate cards say, because the offshore case is strong on its own terms. Published all-in nearshore rates cluster in a band; CodersLink Staff Augmentation, for example, starts from $6,500 (mid-level), $7,500 (senior) and $10,500 (staff+) per engineer per month, all-in by role family. Clients report savings of 30 to 57% against comparable US hiring, and the basis of that comparison is the fully loaded US cost for the same seniority. Offshore vendors in India and the Philippines generally publish or quote rates well below that band; the exact discount depends on seniority and vendor, and we deliberately do not quote a third-party rate here because the ones in circulation are unsourced.

The honest framing is this: if the work can be done asynchronously and the team is already managed by people who are awake when the engineers are, offshore is cheaper and you should buy it. The rest of this post is about the cases where those two conditions do not hold, which in our experience is most product engineering.

How many working hours actually overlap?

Central Mexico is UTC-6 year-round, so a Mexico engineer shares seven to eight hours of the day with US Central time and a similar window with Eastern and Pacific. India (UTC+5:30) shares zero to two hours with US business hours; the Philippines (UTC+8) shares roughly none; Poland and Ukraine share two to four. Overlap is the single biggest driver of the other three costs.

Sources: Mexico's abolition of daylight saving is the Ley de los Husos Horarios published in the Diario Oficial de la Federación on 28 October 2022; a strip of border municipalities still follows US time. Flight times are published nonstop block times, checked September 2026 (Dallas to Monterrey about 1 h 50, Dallas to Guadalajara about 2 h 40, Los Angeles to Guadalajara about 3 h 15). TN eligibility is from USCIS, TN USMCA Professionals.

Two precisions matter. First, because Mexico no longer shifts and the US does, the Mexico window with Eastern time is closer to six hours in US summer and with Pacific closer to six in US winter; seven to eight is the Central-time figure and the one CodersLink commits to. Second, offshore vendors close the gap with a night shift, which is real, but it moves the cost from your side to the engineer's, and it shows up later in the attrition row.

What does a nine-hour offset cost in rework and delay?

With near-zero overlap every question waits until the next day, so a task that needs three clarifications takes three days longer than the same task with a live answer. Teams compensate with heavier specification, longer handover documents and a duplicated management layer, all of which are unbilled engineering time on your side.

There is no universal multiplier, and any vendor who gives you one is guessing. What you can do is count the events. Take a typical two-week sprint for a feature with a few ambiguous edges:

• Clarifying questions. Product engineering generates several per ticket. With live overlap each is a two-minute Slack thread. With a nine-hour offset each is an overnight turn, and a ticket with three of them slips three days.
• Code review latency. A pull request opened at the end of the offshore day is reviewed during the US day and the comments are read the next offshore morning. Two review rounds cost four calendar days instead of an afternoon.
• Handover overhead. Teams with no overlap write more: longer tickets, explicit acceptance criteria, recorded walkthroughs. That is good practice and also a real cost: a senior US engineer spending an hour a day writing for the offshore team is a fifth of that engineer gone.
• The duplicated manager. The most common offshore compensation is a local delivery lead who attends your evening meetings and relays. It works, and it adds a layer, a salary and a game of telephone between your product owner and the engineer.

None of this appears on the vendor's invoice. It appears in your velocity, your senior engineers' calendars and your on-call rota. When NetProtect moved its operations from offshore Asia to a Guadalajara team, the published account of the reason was time-zone alignment rather than rate; the team grew from six to forty people over eighteen months and was transferred to the client's own payroll (case study).

What does attrition cost, and how do the regions compare?

Indian IT majors reported voluntary attrition between roughly 13 and 14% in their FY2026 filings, after peaks above 20% in 2022. Every departure costs the ramp time of the replacement plus the knowledge that left. Nearshore vendors that hold the engineer's contract and publish replacement terms move that cost onto themselves; marketplace and body-shop models leave it with you.

Attrition is the cost that arrives in month nine, after the rate has been approved and the team has ramped. The large Indian services firms disclose it, which is to their credit and makes the comparison possible. In their fiscal-2026 reporting, Infosys reported voluntary attrition around 14.1%, Wipro 13.8% for the twelve months to March 2026 (down from 15.1% the prior quarter) and TCS around 12.8 to 13.7% depending on the quarter (Wipro Q4 FY26 release; Infosys Form 6-K, FY2026). Those are firm-wide numbers for very large, well-run companies; the attrition on a specific augmented team, where the engineer is a contractor working a night shift for a foreign client, is a question to put to the vendor directly.

What the number costs you is arithmetic. A replacement engineer takes weeks to source and, on a mature codebase, two to three months to reach full productivity. On a team of ten at 14% annual attrition, that is one or two ramps a year, each costing a quarter of an engineer-year plus the reviewers' time. At 30%, which is not unusual on night-shift contract teams, it is three.

The nearshore side of that comparison is where we should be precise about our own numbers. CodersLink publishes 15% voluntary turnover and an 18 to 24 month average tenure as marketing-class figures, and an average seniority of eight years across placed engineers. The published case with a retention figure is Particle: a 19-role team built in under 120 days with 97% retention at twelve months. The more useful question for any vendor is not the attrition rate but who pays for it. On CodersLink Staff Augmentation, replacement is unlimited and free for the life of the engagement, and sourcing restarts within one business day. We do not run a free trial; our equivalent is open-ended. Several nearshore competitors publish a 90-day replacement window (TECLA, Hire South) or a 14 to 30 day trial (Revelo, Sonatafy), as read on their sites in September 2026; most offshore body shops publish nothing.

How do travel and visas change the equation?

A Mexico team is a two to four hour nonstop flight from Texas and California, which makes quarterly onsite sprints routine, and Mexican professionals can work onsite in the US under the TN category with no lottery and no annual cap. Offshore teams are a fifteen-hour flight away and depend on the H-1B, which is capped, lottery-selected and, since 2025, entangled in a $100,000 fee dispute.

Distance is the easier half. At two to four hours nonstop, a Mexico team can join a quarterly planning session in Austin and be home the next evening; a fifteen-hour flight turns the same visit into a week and a budget line. Over a multi-year engagement that changes how often the team meets your customers and your executives, which changes how embedded it becomes.

Mobility is the harder half, and it is worth stating carefully because the rules move. Mexican and Canadian professionals in USMCA-listed occupations, which include engineers and computer systems analysts, can work in the US under the TN category. There is no lottery and no annual cap; Mexican citizens apply at a US consulate, and admissions are granted in increments of up to three years, renewable while the role qualifies (USCIS, TN USMCA Professionals; 8 CFR §214.6). USMCA remained in force after the July 2026 joint review and the TN category is unchanged.

The H-1B, which is the route for engineers from India, the Philippines and most of Eastern Europe, has an annual cap of 85,000 and, since the FY2027 season, a wage-weighted registration lottery (DHS final rule, 90 FR, 29 Dec 2025). Mandatory government fees for a large employer's new petition run to roughly $3,600. Since September 2025 the category has also carried a $100,000 payment on many new petitions under a presidential proclamation; a federal court vacated it in June 2026 and an appeals court declined to reinstate it in July 2026, the government has appealed, and the policy could be revived, extended or replaced. Check current status before planning around it.

Immigration rules change. This section is general information, not legal advice; last reviewed September 2026. Sources: USCIS; US Department of State; Proclamation 10973 (19 Sep 2025); D. Mass. judgment 8 Jun 2026; First Circuit order 24 Jul 2026.

Where does nearshore lose to offshore?

Offshore wins when the work is genuinely asynchronous, when you need a very large or very specialised pool that Mexico's roughly half a million software professionals cannot supply, when you already run a follow-the-sun operation, or when the rate is the only number your budget will accept. Nearshore also loses to onshore when the work needs a security clearance or a badge.

Being honest about the limits is the only way the rest of this post is worth anything.

• Truly async work. Data labelling, test automation against a fixed spec, maintenance of a stable system with a good runbook, overnight batch operations. If nobody needs to talk to the engineer during your day, the overlap argument is worth nothing and the rate argument is worth everything.
• Follow-the-sun. If you want 24-hour coverage of a support queue or an on-call rota, an offshore team is not a compromise; it is the design.
• Pool size and rare specialisms. India's engineering workforce is an order of magnitude larger than Mexico's. Mexico has more than 500,000 software and IT professionals (INEGI ENOE, 1Q 2026; STPS Observatorio Laboral) and produces roughly 110 to 130 thousand engineering and technology graduates a year (ANUIES, Anuario Estadístico de Educación Superior). That is deep for most stacks and thin for some; if you need forty engineers with a rare combination next quarter, the larger pool wins.
• Budget as a hard constraint. If the number on the purchase order is the constraint and the collaboration costs land on someone else's budget line, offshore will be approved and nearshore will not. That is a political fact, not an economic one, and it is worth naming as such in the business case.
• Clearances and badges. Federal and defence work that requires onshore staff excludes both models; do not let a nearshore vendor, including us, tell you otherwise.

How should you decide between nearshore and offshore?

Score the work on four questions: how much of it needs a live answer, how often the team will meet your people, how long you intend to keep the same engineers, and whether the rate or the total cost is the number your organisation will manage to. Three or more answers pointing to live, frequent, long and total mean nearshore; the reverse means offshore.

1. How much of the work needs a live answer? Count the clarifying questions per ticket in your last sprint. More than one or two, and overlap is worth paying for.
2. How often will the team meet your people? If quarterly onsite time is part of how you build teams, a two-hour flight and a visa without a lottery are operating advantages, not perks.
3. How long do you intend to keep the same engineers? If the answer is years, buy from a vendor who holds the engineer's contract, publishes retention and replacement terms and can show a case where the team stayed. If the answer is one project, buy the rate.
4. Which number will your organisation manage to? If procurement manages to the rate and engineering manages to velocity, put both in the business case and let the CFO see the unbilled hours. The Mexico Tech Salaries Report 2026, with 10,246 verified respondents across 36 roles and 32 states, is the compensation baseline for the Mexico side of that model.

If you have already chosen Latin America, the next question is which country, and we make the case for Mexico with the data in best Latin American country for staff augmentation. If you are comparing vendors rather than geographies, the comparison hub lays out the categories and the vendor pages show each company's own published terms, for example CodersLink vs Turing for a global talent cloud and CodersLink vs BairesDev for a multi-country outsourcer. For what the nearshore model itself looks like from Mexico, start with Staff Augmentation. And if the answer for your team is offshore, we will say so on the call: book a 30-minute fit call.