K
KnowMBAAdvisory
OperationsAdvanced8 min read

Captive vs Outsourced Decision

The captive vs outsourced decision determines whether work delivered offshore or nearshore is performed by your own employees in your own facility (a captive center, also called Global In-House Center or GIC) or by a third-party provider (BPO/ITO). India alone hosts ~1,800+ captives employing ~1.7M people (2024) — for many global firms (JPMorgan, Goldman Sachs, Citi, American Express, Deutsche Bank, Microsoft, Cisco, Walmart, Target) the captive in India is now larger than headquarters. The captive model trades the immediate cost arbitrage of BPOs against three structural advantages: (1) full IP and data control, (2) embedded institutional knowledge that doesn't churn out at the end of a contract, (3) ability to push the delivery center up the value stack into core engineering, R&D, and analytics — work no BPO can deliver. The economic crossover: BPOs are typically 10-25% cheaper per FTE in years 1-3, but captives match or beat BPOs on 5-7 year TCO when scale is sustained, because the captive doesn't pay BPO margin (~12-15% operating margin extracted by Concentrix, Teleperformance, etc.) and doesn't suffer renewal escalation.

Also known asCaptive Center vs BPOGIC vs BPOGlobal In-House Center DecisionBuild vs Buy Operations

The Trap

Treating the captive vs outsourced decision as primarily about cost. The headline year-1 cost gap is real but narrow; the strategic differences dominate the decision. BPO is right for: commodity transactional work, variable demand (you can flex up and down), when you don't have scale to justify a captive (typically need 200+ FTE minimum), and when you don't want to manage operational complexity in a foreign country. Captive is right for: differentiated work that touches IP, work where institutional knowledge compounds (engineering, analytics, R&D), regulated work where data residency matters, and when scale exceeds 500-1000 FTEs. Second trap: hybrid done badly. 'We'll do a captive AND use BPOs' becomes a sprawling mess if not governed — captive and BPO start competing for the same work, governance overhead doubles, and neither model gets the scale benefits. Third: setting up a captive when you don't have the scale to justify it. A captive needs ~250-300 FTEs minimum to amortize the management overhead, facility cost, country leadership, HR/finance/IT operations, and compliance functions; below that scale, BPO is structurally cheaper.

What to Do

Use a four-quadrant decision matrix on two axes: SCALE (under 250 FTE / 250-1000 / 1000+) and STRATEGIC NATURE (commodity / differentiated / core IP). Commodity at any scale → BPO. Differentiated under 250 FTE → BPO with strong vendor management. Differentiated 250-1000 → hybrid (captive for core, BPO for surge/commodity adjacent). Differentiated 1000+ or core IP at any scale → captive. For captive setup: budget 18-24 months from decision to first work delivered (not 6-12 as some consultants pitch); plan for 12-18 month payback at minimum; pick a delivery city based on 5-year talent depth, not year-1 cost; install country leadership BEFORE you scale headcount. For BPO: never sign a contract for work you haven't already documented and standardized — outsourcing chaos hard-codes chaos in a contract.

Formula

5-Year Captive vs BPO TCO = Captive Setup + (Captive Run Cost × 5) vs (BPO Year-1 × 5 with 4-7% annual escalation) + Vendor Management Cost

In Practice

GE Capital pioneered the modern captive in India: GE Capital International Services (GECIS) opened in Gurgaon in 1997, scaling from 20 employees to 11,000 by 2004 — at which point GE sold a 60% stake to private equity (Genpact was born). The GE captive originally delivered finance and accounting back-office work; over time it grew into analytics, risk modeling, and high-value research. GE's logic for going captive: the work touched financial controls and customer data they wouldn't trust to a third party. JPMorgan now operates ~50,000+ employees in India across Mumbai, Bangalore, Hyderabad, and Pune — the bank's largest single-country employee base outside the US. JPMorgan India does everything from compliance operations to quantitative research, machine learning, and core engineering. Goldman Sachs' Bangalore captive (Goldman Sachs Services) employs ~10,000+ doing engineering, analytics, and operations — Goldman publicly states the Bangalore center is 'no longer offshore, it's another office.' The captive trajectory is consistent: start with back-office work for cost, expand into mid-office for control, eventually deliver core front-office work for capability.

Pro Tips

  • 01

    Captives have a 'sustainable scale floor' around 250-300 FTEs. Below that, fixed overhead (country leadership, HR/finance/IT, facility, compliance) dominates the unit economics and BPO wins. Above ~1000 FTEs the captive can amortize overhead well below BPO loaded cost.

  • 02

    The 'BOT' (Build-Operate-Transfer) model lets you have it both ways: a BPO partner builds and operates a dedicated captive for you, then transfers ownership at year 3-5. Useful when you want captive economics long-term but don't have country setup expertise. Mature in India; slowly emerging in Mexico.

  • 03

    Don't underestimate captive setup time. Realistic timeline: 6 months for legal entity / facility / country leadership; 12 months for first 200 FTEs delivering meaningful output; 24 months for steady-state scale. Plans that promise 6-month captive ramp routinely overrun and underdeliver.

Myth vs Reality

Myth

BPOs always cost less than captives

Reality

Year 1: yes, BPOs are typically 10-25% cheaper. Year 5+: captives at scale beat BPOs because (a) captives don't pay the BPO's 12-15% operating margin, (b) captives capture all productivity gains internally instead of sharing them with the provider, and (c) captives don't face renewal escalation. Industry studies (Everest Group, Hackett) consistently find captive 5-year TCO at scale is 5-15% below BPO TCO.

Myth

Captives can't deliver high-end work

Reality

Captives have moved decisively up the value chain. JPMorgan India delivers quantitative research, machine learning, and core engineering. Goldman Sachs Bangalore runs trading systems engineering. Microsoft's India captive runs significant portions of Azure and M365 engineering. The 'captives only do back office' belief is 15 years out of date.

Try it

Run the numbers.

Pressure-test the concept against your own knowledge — answer the challenge or try the live scenario.

🧪

Knowledge Check

You're considering offshoring 600 FTEs of engineering + analytics work to India. The work touches your core IP and proprietary algorithms. The CFO prefers BPO for cost; the CTO prefers captive for control. What's the right call?

Industry benchmarks

Is your number good?

Calibrate against real-world tiers. Use these ranges as targets — not absolutes.

Captive vs BPO 5-Year TCO Delta (at 500+ FTE scale)

Offshore captive vs BPO comparison, 500+ FTE scale, 5-year horizon

Captive substantially cheaper

Captive 15-25% cheaper

Captive moderately cheaper

Captive 5-15% cheaper

Roughly equivalent

± 5%

BPO cheaper (sub-scale or low-value work)

BPO cheaper

Source: Everest Group / Zinnov GIC vs BPO Benchmark 2023

Real-world cases

Companies that lived this.

Verified narratives with the numbers that prove (or break) the concept.

⚙️

GE Capital / Genpact (GECIS)

1997-2005

success

GE Capital opened GECIS (GE Capital International Services) in Gurgaon, India in 1997 with 20 employees doing accounting back-office work. By 2004 GECIS had scaled to 11,000 employees delivering finance, analytics, and customer service work for GE businesses globally. Critical to the captive logic: the work touched GE's financial data, customer data, and operating data — work GE wouldn't trust to a third party. In 2004 GE sold 60% of GECIS to private equity (General Atlantic + Oak Hill) for ~$500M, and GECIS rebranded as Genpact in 2005. Genpact went public in 2007 and is now an $4.5B+ revenue BPO serving 800+ clients globally. The GE captive trajectory pioneered the playbook: start captive for control, scale aggressively, then optionally monetize once mature.

GECIS Founding Year

1997 (20 employees)

GECIS Headcount at PE Sale (2004)

11,000

PE Sale Value (60% stake)

~$500M

Genpact Revenue (2024)

~$4.5B

Captives can become standalone businesses. GE's captive scaled to a value that made spinning off worth more than continuing to operate it captive — a path now studied in PE circles as the 'GIC monetization' play. For most companies the lesson is more direct: build the captive for control AND optionality.

Source ↗
🏦

JPMorgan India

2002-2024

success

JPMorgan operates one of the largest captive footprints in India: ~50,000+ employees across Mumbai, Bangalore, Hyderabad, and Pune by 2024 — the bank's largest single-country employee base outside the US. The work has evolved from back-office operations (year 1-5) to mid-office risk and analytics (year 5-15) to front-office quantitative research, machine learning, and core engineering (year 15-present). JPMorgan publicly states India is no longer offshore but a 'core capability center.' The economic logic at this scale: the marginal cost-per-FTE is dramatically below any BPO equivalent, and the work the India teams now perform (algorithmic trading systems, regulatory modeling, core banking engineering) cannot be outsourced because no BPO has the depth.

JPMorgan India Headcount (2024)

50,000+

Cities

Mumbai, Bangalore, Hyderabad, Pune

Scope of Work

Back office → Engineering, AI, Quant Research

Position

Largest non-US JPM presence

Mature captives migrate up the value stack — and that migration is the long-term strategic payoff. JPMorgan India started as a cost play and became a capability moat. The same trajectory is visible at Goldman Sachs Bangalore, Microsoft India, Deutsche Bank Bangalore, and dozens of other captives.

Source ↗

Decision scenario

The Build-vs-Buy Operations Decision

You're COO at a $2B fintech. You need to scale offshore operations from 80 FTEs (currently with a BPO partner) to 600 FTEs over 24 months. Work mix: 40% commodity transaction processing, 35% analytics and risk modeling (touches proprietary algorithms), 25% engineering for core platform. You have three options.

Current Offshore HC

80 (all BPO)

Target HC

600

Time Horizon

24 months

Annual Operating Budget

$28M (target)

Strategic Importance

65% of work touches IP

01

Decision 1

Three options: (a) Scale BPO contract to 600 FTEs — fastest ramp, lowest setup cost, but BPO will own institutional knowledge and you'll be exposed to renewal pressure on a much larger contract. (b) Pure captive build — 18-24 month ramp, higher setup cost ($14M), best 5-year economics and IP control. (c) Hybrid: build captive for 350 FTEs of analytics + engineering work (the IP-touching scope), keep BPO for 250 FTEs of commodity transaction processing.

Pure BPO scale-up — 600 FTEs with the existing partner. Fastest, simplest, no captive setup cost.Reveal
Ramp completes in 14 months. Year 1 looks great — $24M run rate. By year 3, BPO renews at +14% citing the much larger headcount and 'capability uplift on analytics work.' Your IP-sensitive analytics work is now in the BPO's hands and they know switching is impractical. By year 5 you're paying $34M/yr (vs original $24M baseline) and the analytics team's institutional knowledge belongs to the BPO. When you eventually try to insource, you find you've lost the talent (they all work for the BPO) and have to rebuild from scratch at far higher cost.
Year-5 Annual Cost: $24M → $34MIP Control on Analytics: Erodes to BPOStrategic Optionality: Lost
Hybrid: build captive for 350 FTEs of analytics + engineering (IP-sensitive), keep BPO for 250 FTEs of commodity transaction processing.Reveal
Year 1: captive setup runs $11M, captive HC reaches 80. BPO scales to 250 FTEs running commodity work efficiently. Year 2: captive hits 280 FTEs, takes over analytics work from BPO. Year 3: captive at full 350 FTEs, IP fully internalized. Year 5 run-rate: $25.5M (vs $34M for pure BPO). More importantly: analytics talent is yours, IP is yours, and you have permanent optionality — you can scale the captive further or replace the BPO without strategic risk. Total 5-year savings vs pure BPO: ~$28M. KnowMBA POV: cost takeout that doesn't include strategic redesign comes back as either service quality decay or capability erosion. The hybrid captures both arbitrage AND structural advantage.
Year-5 Annual Cost: $24M → $25.5M (vs $34M pure BPO)5-Year Cumulative Savings vs BPO: +$28MIP Control on Strategic Work: Fully retainedTalent Ownership: 350 FTEs internal

Related concepts

Keep connecting.

The concepts that orbit this one — each one sharpens the others.

Beyond the concept

Turn Captive vs Outsourced Decision into a live operating decision.

Use this concept as the framing layer, then move into a diagnostic if it maps directly to a current bottleneck.

Typical response time: 24h · No retainer required

Turn Captive vs Outsourced Decision into a live operating decision.

Use Captive vs Outsourced Decision as the framing layer, then move into diagnostics or advisory if this maps directly to a current business bottleneck.