In 2007, NASSCOM's president Kiran Karnik told the industry what everyone in Gurgaon and Bangalore already suspected: India's software and services sector would record revenues of $36–38 billion for the fiscal year, with ITES-BPO exports crossing $8 billion. He added a prediction that turned out to be the more important one: an increase in mergers and acquisitions that would consolidate the industry and create world-scale players. This is the year the Western caricature of business process outsourcing — a room full of headsets in Bangalore, reading scripts to angry customers in Ohio — was at its most accurate and already going out of date.
What Was Actually Being Outsourced
The call centre stereotype was never the whole industry, but it was close to the majority of it. Industry analysis of the period put finance and accounting, customer interaction services and HR administration at roughly 89% of total BPO revenue. That combination tells you what buyers were optimising for. These were high-volume, rules-based, well-documented processes where output quality was easy to measure and where labour was the dominant cost. Accounts payable. Payroll processing. Order entry. Tier-one customer support. Collections. What was not being outsourced: anything requiring judgement, institutional context, or a conversation with a regulator.
Where the Industry Came From
The origin story is more corporate than entrepreneurial. General Electric established a captive operation in Gurgaon in the late 1990s to process its own back-office work — the unit that would later be sold to private equity and renamed Genpact. British Airways did something similar with the captive that became WNS. The credibility to do any of this came from Y2K. Through 1998 and 1999, Indian IT services firms proved they could execute large volumes of tedious, deadline-bound, quality-sensitive remediation work for Western clients. When the deadline passed, the delivery model, the client relationships and the trust remained — and were pointed at business processes rather than code. Labour arbitrage did the rest. A finance graduate in Chennai cost a fraction of an equivalent hire in Manchester or Ohio, and the internet had just become cheap and reliable enough to make location irrelevant for work that consisted of reading a screen and typing.
The Economics Were Already Eroding
By 2007 the model was under visible pressure, which is why Karnik's consolidation forecast mattered. Wage inflation in the major delivery cities was running at double-digit annual rates as providers bid against each other for the same experienced staff. Attrition in voice processes was the industry's structural disease — routinely cited in the 30–50% range annually. Every departure destroyed accumulated process knowledge and added recruiting and training cost. Currency moved the wrong way. The rupee appreciated sharply against the US dollar during 2007, squeezing margins for providers whose costs were in rupees and whose contracts were priced in dollars. It is an underrated lesson: an arbitrage strategy is also a currency position. Political backlash was real in the US and UK, expressed in election rhetoric, in customer complaints about accents, and in the awkward industry practice of giving agents Western pseudonyms. Each of these compressed the savings that had justified the decision in the first place.
The Risks Nobody Wrote Into the Contract
The contracts of this era were meticulous about service levels and vague about everything that determined whether the arrangement worked. Process knowledge left the building. Organizations transferred work they had never documented, discovered the documentation gap during transition, and reconstructed the process from whoever remembered it. Some of the knowledge simply did not survive. Quality measurement rewarded the wrong things. Average handling time and calls per hour are easy to measure and easy to game. First-contact resolution and downstream error rates are what actually cost money. Data protection was contractual, not legal. India had no comprehensive personal data protection statute at the time; controls depended on what the client wrote into the master services agreement and audited afterwards. A widely reported fraud at a Pune call centre in 2005, where employees used customer credentials to siphon funds from overseas accounts, showed how fast a control failure became a client's problem rather than a provider's. Accountability did not transfer. Regulators and customers always came back to the brand on the invoice. The provider's SLA credit never covered the actual loss.
From Call Centres to Capability Centres
The industry that Karnik was describing in 2007 has mutated almost beyond recognition. The voice tier commoditised, then fragmented across the Philippines, Eastern Europe, Egypt and Latin America. Knowledge process outsourcing moved up the value chain into research, analytics and actuarial work. And most significantly, the captive model returned at scale: India now hosts well over 1,700 global capability centres employing more than a million and a half people, running engineering, finance, analytics and product work for their parent companies rather than for third-party clients. That is a reversal worth noticing. The 2000s logic was that non-core processes should be handed to a specialist. The 2020s logic is that these capabilities are strategic enough to own directly, just in a lower-cost location.
What AI Is Doing to the Same Work
The processes that made offshore BPO profitable — high volume, rules-based, well-documented — are precisely the processes that automation and language models now handle without a human in the loop. Tier-one support deflection, invoice coding, document extraction, reconciliation matching and first-draft correspondence are all being absorbed by software. The arbitrage that remains is increasingly in exception handling, judgement, relationship management and control — the work that was never sent offshore in the first place. For mid-market operators the practical implication is uncomfortable: if you are still evaluating a sourcing decision purely on cost per full-time equivalent, you are running a 2007 analysis in a market where the relevant comparison is no longer between two groups of people.
Document and eliminate
Describe the workflow and remove work that need not happen.
Assess automation
Test routine steps before choosing a delivery location.
Source with an exit
Price transition, retain control ownership and define data/documentation return.
Qualitative summary of this article's source text, not a measured outcome or performance estimate.
How to Modernise a Back Office Sourcing Strategy
- Document the process before you move it. If it cannot be written down, it cannot be transferred, automated or audited — by anyone.
- Separate the three questions: what should be eliminated, what should be automated, and only then what should be sourced elsewhere.
- Price the transition honestly. Knowledge transfer, parallel running and productivity dips typically consume the first year of projected savings.
- Measure outcomes, not activity. Error rates, rework and cycle time tell you whether the arrangement is working. Handling time does not.
- Contract for exit from day one, including data return formats, documentation ownership and transition assistance.
- Keep the control function in-house. Approval authority, reconciliation sign-off and regulatory responsibility should never sit with a vendor.
Common Questions
How big was India's BPO industry in 2007?
NASSCOM projected ITES-BPO exports to cross $8 billion in 2007, within a broader software and services industry of $36–38 billion for the fiscal year.
Why did companies outsource back office work offshore?
Primarily labour cost arbitrage, enabled by cheap bandwidth and by delivery credibility built during Y2K remediation. Secondary drivers included follow-the-sun coverage and access to a large graduate workforce.
Is offshore BPO still cost effective?
It can be, but the margin is far narrower than in 2007 after two decades of wage inflation and currency movement. The stronger contemporary case is access to talent and capability, which is why captive global capability centres have grown faster than third-party BPO.
What is the most common outsourcing mistake?
Outsourcing a broken process. Transferring an undocumented, exception-heavy workflow to a third party does not fix it; it relocates the problem and adds a contract boundary that makes the problem harder to solve.
Modernize Your Back Office Strategy — Outpace reviews finance and administrative operations across eliminate, automate and source decisions, with honest transition costing. If your last sourcing review was built on a cost-per-FTE model, it is out of date.