Investigation · Talent & GCC
Headcount Moved to India. Decision Rights Mostly Didn’t.
Investigation · GCC · 6 min read · June 2026
Thesis
Localising headcount is not the same as localising authority, and the gap is where the real GCC story lives.
The Question
I spent a quarter close to the GCC hiring market and came away with a nagging mismatch. Every report celebrates that centres in Bengaluru and Hyderabad do senior, strategic work now, and the headcount supports it. Mid-to-senior roles are most of the hiring. AI and platform roles are growing fast. But when I looked at who actually owns the decision, the picture went quiet. So the question I wanted to push on, after the earlier Hiring Pulse: across functions, which ones are genuinely getting decision rights moved to India, and which are still executing a call made somewhere else?
The Hypothesis
My starting position, and I will admit it was a hunch before it was a finding, was that localisation is being measured by the easy thing, how many people and how senior the titles, and not the hard thing, who signs off. A centre can be full of principal engineers and still have every architecture call routed back to New Jersey. The hypothesis: a few functions are getting real mandate and the rest are getting bigger, better-paid execution. The reporting line is the tell, not the title.
How I Looked At It
I leaned on the public landscape work, the Nasscom and Zinnov GCC reports and the EY GCC Pulse, for the macro shape, and on what I could observe directly in role design during the quarter, the job descriptions, who a role reports to, where the budget sits. A title says "Head of." The reporting line says whether "Head of" means anything. Where the two disagreed, I trusted the reporting line.
What I Found
The compensation tell comes first, because it gives the game away. Mandate-carrying roles are being priced unpredictably, all over the place, the way a scarce thing is priced. Execution roles, however senior, are priced like a market. When you see a "Head of" role paid like a market rate, you are usually looking at execution wearing a leadership title.
Engineering and product is where real localisation has happened. Architecture and roadmap authority has moved, partly because the talent depth makes it indefensible not to. When the best people for the decision sit in Bengaluru, the decision tends to follow them eventually.
Data and AI is the interesting middle. The headcount is exploding and the language is strategic, but a lot of the actual prioritisation, what gets built and what gets killed, still happens at headquarters. India is trusted to build the model more than to decide which model is worth building. That may be a transition state. It may also be where it settles.
Finance, legal, and large parts of compliance are the honest laggards. Big centres, genuinely senior work, decision rights that have barely moved. I do not think that is an accident. These are the functions where headquarters feels the cost of being wrong most personally, so they hold the pen.
The number that stuck with me from the earlier work: a large majority of India GCCs still have under ten percent of their leadership physically in India. You can move thousands of senior people and still not move the few dozen who decide.
I want to be careful. I cannot prove the function-by-function read with clean data, because nobody publishes "where does sign-off actually happen." This is pattern reading from role design and reporting lines, and it could be wrong in specific companies. The direction I am confident about. The precision I am not.
What It Means For Business
If you run a GCC, stop measuring your maturity by headcount and title seniority, and start counting how many decisions of consequence actually get made in the building over a year. That is the number that predicts whether your best people stay, because people with options do not execute someone else's call forever. The Monday version of this: pick three recurring decisions that currently route to headquarters, and move one of them, with the budget attached, to the person you already pay like a leader. If you cannot bring yourself to move even one, you do not have a capability centre, you have a large, expensive execution floor, and your competitor down the road is already hiring away the people who noticed.
The Decision
What I’d Test With More Time
I would want a structured way to score decision-rights localisation by function, something more rigorous than reading reporting lines, maybe tracking where specific recurring decisions get approved across a year. I would also want to know whether the AI function moves into real mandate or stays in well-paid execution, because that one is genuinely open and I would not bet hard either way yet.