Start a conversation →
Talent

Attrition is a delivery metric

Four percentage points of attrition on a fifty-person team is four to eight lost sprints a year. Why retention architecture belongs in the build plan, not the HR review.

Attrition gets reported to HR and reviewed in HR meetings, which quietly teaches every other function in the business to think of it as an HR problem. For a GCC, that framing is a mistake with a real cost attached, and the cost shows up on the engineering roadmap, not the HR dashboard.

The arithmetic nobody runs

Take a fifty-person engineering team at eighteen percent annual attrition — roughly the Tier 1 Premium average. That is nine departures a year. Each departure triggers a recruitment cycle, typically six to ten weeks. A notice period, often thirty to ninety days depending on seniority. An onboarding ramp for the replacement, four to six weeks before they are contributing at the level of the person they replaced. And a knowledge-transfer gap that no amount of documentation fully closes, because a meaningful share of what a senior engineer knows about your product was never written down anywhere.

Move that same team to a Tier 1 Stable location at roughly fourteen percent attrition, and you lose seven people instead of nine. Two fewer departures does not sound like much. It is eight to sixteen weeks of recovered engineering capacity a year — four to eight full sprints — without hiring a single additional person.

Why GCCs already outperform on this metric

The sector-wide data is genuinely favorable: GCCs record 14.1 percent attrition against 20.5 percent for IT services and twenty-four percent for financial services, according to EY's most recent sector analysis. GCCs are structurally better at retention than the alternatives, largely because engineers doing product work with real ownership stay longer than engineers executing tickets for a rotating client roster.

But sector-average performance is not the same as your performance, and the gap between a deliberately designed retention architecture and a default one is still worth four to six percentage points — the difference between a good year and a genuinely disruptive one.

What retention architecture actually means

Not perks. Named, published progression criteria so engineers know exactly what the next level requires. Real access to AI and platform upskilling from month one, which the market currently values more than almost any other career investment available to mid-career Indian engineers. Direct, regular contact with your leadership — not a monthly newsletter, an actual session where the India team sees the product and the customer, not just the ticket queue.

The engagements that get this right build the retention plan before the first offer letter goes out — not after the first wave of departures forces the conversation.