The question to answer
A GCC decision is rarely "should we go offshore?". Most companies already buy offshore work from service providers. The real question is: for this scope of work, over five years, which way of sourcing it costs least and carries the risk we can live with? Four options cover nearly every case:
- Outsource: a managed service provider delivers the work from its own centre at a price per FTE or per outcome.
- Build, operate, transfer (BOT): a partner builds the centre, runs it for 18 to 36 months, then transfers the people, the assets and usually the entity to you.
- Captive: you set up your own entity and build the centre yourself from day one.
- Hybrid: you own the core (engineering, data, cyber) and buy the run (support, service desk, transactional operations).
Compare them on the same scope, the same years and the same baseline: what the work costs today if nothing changes.
From salary to loaded cost
A salary survey gives you base pay. The centre pays much more than that. Build the loaded cost of one FTE in five steps:
| Step | India | Philippines |
|---|---|---|
| 1. Base pay | Annual cost to company (CTC), usually quoted in lakh of rupees | Monthly base pay, paid 13 times a year because the 13th month pay is mandatory |
| 2. Variable pay | 5% to 15% of base, rising with seniority | Same logic, often lower for operations roles |
| 3. Employer on-costs | Gratuity, insurance and statutory costs not in CTC, around 9% in the model | Employer social contributions and health cover, around 11% in the model |
| 4. Seat and technology | Rent, maintenance and utilities per seat (0.85 seats per FTE with hybrid working), plus laptops, licences, network and security tools | |
| 5. Running the centre | Recruitment, attrition replacement, training, travel to headquarters, and the enabling staff (HR, finance, facilities, centre IT) a captive needs: about 7 per 100 FTE | |
Then convert at a rate your treasury agrees, and model the drift: wages in India have been rising faster than at headquarters while the rupee has tended to weaken against the dollar. Both belong in the model as inputs, not as a single number frozen on the day you write the case.
Cost per FTE by city
The table shows the fully loaded annual cost of one FTE for a typical technology and operations mix (engineers, analysts, service desk and finance operations, with their managers), in USD, for a captive centre, against an outsourced price for the same mix and the onshore cost in Western Europe. Year 1 prices, mid scenario.
| City | Captive, loaded | Outsourced price | Onshore today | Best known for |
|---|---|---|---|---|
| Bengaluru | 41,600 | 50,400 | 108,000 | Deepest engineering, product and AI talent; highest pay and attrition |
| Hyderabad | 39,500 | 49,400 | 108,000 | Large technology and engineering pool, strong for data and cloud |
| Pune | 37,700 | 48,900 | 108,000 | Engineering and manufacturing software, ERP, automotive |
| Chennai | 37,300 | 48,900 | 108,000 | Engineering, automotive, finance and operations at scale |
| Metro Manila | 32,800 | 47,900 | 108,000 | Service desk, finance and business operations, customer-facing work, growing technology |
| Cebu | 28,300 | 45,400 | 108,000 | Operations and support at lower cost, a smaller senior bench |
Illustrative rates as at November 2026. Salaries, seats and vendor prices move every year and differ by employer and skill: replace them with current quotes and survey data before you decide.
Two things stand out. First, the gap between onshore and offshore is large everywhere, so the choice of city matters less than the choice of model and the speed of the ramp. Second, the cost advantage is not the same for every role: it stays large for junior, operations and finance roles, and narrows for architects, AI and cyber specialists and senior leaders, who are paid close to international rates in every hub.
The lines people forget
| Line | Why it matters | How the model treats it |
|---|---|---|
| Ramp and productivity | Nobody is fully productive in month one, and a captive takes longer to hire than a partner | Share of the team in seat and its productivity, year by year, per model |
| Parallel run | Onshore teams stay until the new team has proven it can run the work | Onshore cost released one quarter after the offshore team is in seat |
| Retained team | Product owners, architects and vendor managers stay at headquarters | 7 retained roles per 100 FTE moved for a captive or BOT, 12 for outsourcing |
| People leaving onshore | Severance is often the largest one-off cost | Share of employees in today's team, share redeployed, cost per exit |
| Attrition | Every leaver means a new hire, an agency fee and lost knowledge | Replacement hires a year by country, costed as recruitment |
| Partner fees and transfer | A BOT partner charges a management fee on cost, a build fee and a transfer fee | Fee on pass-through cost until transfer, then the transfer fee and post-transfer legal costs |
| Entity and compliance | An own entity needs audits, transfer pricing documentation, tax filings and a company secretary every year | One-off setup and an annual compliance cost per country |
Four options compared
| Outsource | BOT | Captive | Hybrid | |
|---|---|---|---|---|
| Time to first team | 3 to 6 months of transition | 2 to 4 months | 4 to 6 months (entity first) | Mixed |
| Run cost per FTE | Highest: the provider's margin, every year | Partner fee until transfer, then as captive | Lowest at scale | Between the two |
| Upfront effort | Low | Medium: you choose and govern a partner | High: entity, site, leadership, hiring | High for the core |
| Control of talent and know-how | Low | Grows to full at transfer | Full | Full for the core |
| Exit and reversibility | Contract notice and transition | Call option; harder after transfer | Hardest: an entity to close | Mixed |
Worked example
A fictional industrial manufacturer with headquarters in Western Europe wants to move 220 FTE of work in six workstreams: applications and ERP support, software engineering, data and analytics, cyber security operations, infrastructure and service desk, and finance operations. Doing nothing costs USD 126.2 million over five years. The four options, priced on the same basis:
| Option | 5-year cost | 5-year saving | Payback | Run-rate saving in year 5 |
|---|---|---|---|---|
| A. Outsource (Pune and Manila delivery) | USD 97.9m | USD 28.3m | 2.1 years | 38% |
| B. BOT: Pune hub, Manila spoke, transfer at month 30 | USD 86.9m | USD 39.3m | 1.5 years | 51% |
| C. Captive, built in-house | USD 91.1m | USD 35.1m | 2.1 years | 51% |
| D. Hybrid: own the core, buy the run | USD 96.1m | USD 30.1m | 2.2 years | 43% |
The captive and the BOT end at the same run-rate cost, because after the transfer the BOT centre is a captive. The BOT wins over five years because the partner hires faster: the onshore cost falls months earlier, which is worth more than the partner's fee. Outsourcing is quick and light but pays the provider's margin every year, which is why its run-rate saving stays lowest. In this example the knowledge transfer, travel and severance cost USD 6.4 million, more than the partner's fees and the setup costs together.
The company and all figures are fictional and illustrative. Use your own scope, rates and baseline.
What moves the answer
Test at least these before you present: a six-month delay to the ramp, wage inflation two points higher, attrition five points higher, a stronger local currency, a higher severance cost, and the partner's fee. In most cases the ramp and the onshore exit cost move the five-year saving far more than a few percent on salaries, which is why the speed and quality of the transition plan deserve more attention than the last dollar of the rate card.
What cost cannot decide
The cheapest option on paper can still be the wrong one. Score the options, with weights agreed before the scores, on control of intellectual property and talent, speed to start, execution risk, access to the skills you need, flexibility to scale up or down, exit, and the management time it will take from your leaders. A centre that will own products or global processes within three years needs a path to ownership, which favours a captive or a BOT. A centre that only runs stable, well-documented work can stay outsourced.
Also ask the question the board will ask: how much of our critical work will sit in one country or one city? Measure it on criticality and share of global volume, not headcount, and plan a second city or a tested continuity plan before you need it.
Ten tests before the board
| Done | Test |
|---|---|
| ☐ | The baseline is today's full cost of the same scope, growing at the onshore rate. |
| ☐ | Every option covers the same workstreams, years and service levels. |
| ☐ | Rates are dated, sourced and shown as low, mid and high. |
| ☐ | The ramp and productivity are different for each model, and justified. |
| ☐ | The parallel run, the retained team and the people leaving onshore are costed. |
| ☐ | Attrition replacement and recruitment are in the run cost. |
| ☐ | The BOT includes the management fee, the build fee and the transfer fee. |
| ☐ | Currency and wage drift are inputs, not frozen numbers. |
| ☐ | The tornado shows what moves the saving most. |
| ☐ | Control, risk, speed and exit are scored next to the cost. |
Questions
How many people do we need before a captive makes sense?
It depends on the work and the country, but the fixed costs of an own entity (leadership, compliance, enabling staff, setup) only spread well beyond a few dozen people. Below that, a managed team or a GCC-as-a-service provider is usually cheaper and faster; a BOT lets you start small and own the centre once it has reached a sustainable size.
India or the Philippines?
India has the deepest engineering, data and product talent and the widest choice of cities. The Philippines is strong for service desk, finance, business operations and customer-facing work, with good cultural and language fit for Western headquarters. Many companies use both: an engineering hub in India and an operations spoke in the Philippines.
Is the cost advantage disappearing?
It is changing rather than disappearing. It remains large for junior, operations and transactional roles; for niche skills and senior leaders, the hubs pay close to international rates. Centres that move up the value chain justify themselves by what they own, not only by the salary gap.
What does a BOT partner charge?
Typically a management fee on the costs it passes through (people, seats, technology, recruitment), a build fee, recruitment fees and a transfer fee. Fix the transfer price method when you sign, not at transfer, and compare bids on five-year cost, not on the headline fee.
Which rates should I trust?
None without checking. Treat any published rate, including ours, as a starting point: replace it with two partner quotes, a salary survey for your roles and a real-estate quote for the city you prefer.