How to Build a Service That's Expensive to Leave

A standard contact center operation can be transitioned to a new provider in 90 days. LanguageLine's interpretation services take years to replicate: you need the linguist network, the per-minute pricing infrastructure, the enterprise relationships. A client switching faces 12+ months of rebuilding and real quality risk. The market prices that difference accordingly.

Teleperformance's Specialized Services unit earned 31.9% EBITA in 2022. Its core contact center business earned 12-13%. That gap doesn't close through better sales technique or stronger relationships. It reflects how long it would take a client to leave and rebuild what you provide elsewhere.

The companies that moved up the margin stack did it through one of three routes: accumulating domain depth competitors couldn't replicate quickly, moving into client segments where incumbents weren't present, or acquiring the position rather than building it.

EXL's exit was capability-led. It reorganized around domain-specific analytics for insurance, banking, and healthcare, and built proprietary AI platforms trained on years of client-specific process data. That institutional knowledge doesn't transfer to a competitor through a sales pitch; it takes years of running the process at scale to accumulate. Data and AI-led services grew to 57% of total revenue, margins moved from roughly 17% to 19.4%, and revenue nearly doubled over four years. -> EXL analytics-led model

TaskUs took the segment route. Rather than competing for the same enterprise buyers as the incumbents, it built entirely around digital-native clients -- companies the established BPO players hadn't reached with traditional positioning. Trust and Safety and AI Services grew into lines with few qualified competitors, minimal pricing pressure, and 23.2% EBITDA margins at $960M in revenue. -> TaskUs digital-native model

WNS acquired the foothold. The $95M HealthHelp acquisition bought clinical accreditations, a proprietary care management platform, and payer relationships that would have taken three to five years to build. Instead of competing from scratch in healthcare, WNS entered with the infrastructure already in place. Healthcare grew to 17.7% of revenue. -> WNS HealthHelp acquisition

Genpact, TTEC, and Concentrix each ran versions of the same logic across different service lines and geographies, building or acquiring offerings where the switching cost was high enough to sustain a pricing conversation. -> Genpact · -> TTEC · -> Concentrix

---

Acquisition delivers a foothold, not a compounding advantage. WNS healthcare peaked at 17.7% of revenue and fell to 11.1% by FY2025 after losing a large client. The platform and accreditations are table stakes -- the expertise still has to accumulate organically after entry. Buying a position is faster than building one. It doesn't substitute for building one.

Specialization concentrates risk. Deep domain expertise in one vertical means exposure to that sector's cycles. The companies that managed this effectively -- Genpact across insurance, banking, healthcare, and manufacturing -- built genuine depth in multiple verticals rather than broad presence in many. Each new vertical should leverage existing knowledge, not just add new clients.

The Capita case sits at the opposite extreme. More than 100 acquisitions, overlapping business units, no visibility into contract-level profitability, G&A overhead so heavy the company could no longer bid competitively. An 80%+ share price decline before restructuring began. Inorganic growth without integration discipline doesn't produce specialization -- it produces complexity. -> Capita restructuring

---

Start with a simple question: what would it cost a competitor to replicate your three most important service lines -- not in a pitch deck, in reality? If the answer is less than 18 months and less than $10M, you're in commodity territory regardless of what your positioning says.

If you have domain depth that isn't yet differentiated in the market -- processes run at scale, proprietary data, institutional knowledge clients depend on -- the route is packaging it into a named offering with distinct pricing and a focused sales motion. The capability already exists; the commercial structure doesn't.

If you don't have it yet: building takes three to five years and requires concentrated investment and the discipline to turn down adjacent work that would dilute the focus. Acquiring is faster, but only worth it if the target brings accreditations, platforms, or relationships that would take you that long to build organically. If not, you're paying a premium for headcount.