Most India talent reports cannot answer the question you actually have. Here is what to demand of one before you plan a build around it.
This is not a data release. We hold current compensation benchmarks by role, level and city from our own hiring, and we share them inside engagements under NDA rather than publishing them. That is a deliberate choice: a salary number without its role definition, its level definition, its salary basis and its collection date is not a benchmark, it is a rumour with a decimal point. What follows is the more useful thing we can put in public, which is how to read the reports you are already being sent.
The question a national average cannot answer
The question an operator actually has is narrow. Can I hire eleven people with a specific skill profile, at a specific seniority, in a specific city, inside four months, and hold them for three years. Almost every published India talent report answers a different question: what does a category of role cost on average across the country.
Those two questions have different answers, and the gap between them is not small. Within a single city, the spread between the twenty-fifth and seventy-fifth percentile for the same nominal title is wide enough that the midpoint tells you very little about what your offer needs to be. Across cities, the same title can describe materially different scopes of work. A national average sits on top of both of those distributions and reports their centre, which is the one number in the report that describes nobody you are going to interview.
Five questions to put to any report
1. What is the sample, and who is in it? Sample size on its own is not the question. The question is composition. A survey of two hundred multinational captives and a survey of two hundred domestic IT services firms will produce different numbers for the same title, because the work is different and so is the competitive set. If the report cannot tell you the mix of employer types, the numbers cannot be positioned against your own situation.
2. How is the role defined? Ask for the level definition behind the title, not the title. Senior Analyst means one thing in a company where it sits two levels below a manager and something else where it sits four. Without the level definition, you are comparing labels. This is the single most common reason a compensation benchmark misleads a hiring plan, and it is almost always fixable by asking.
3. What is the salary basis? Fixed pay, total cash, or total cost to company are three different numbers, and in India the gap between them is significant enough to change a business case. Total cost to company includes statutory contributions and benefits that fixed pay does not. A report that says "salary" without saying which one it means cannot be compared with your own cost model, and mixing bases across a comparison is how a plan comes in over budget while every individual offer looks correct.
4. Which cities, and defined how? Bengaluru, or Bangalore, is one label covering several distinct submarkets with different commute realities and different employer density. Hyderabad, Pune, Chennai, the National Capital Region and Ahmedabad each have their own composition. A report that treats a metropolitan region as one point of data is averaging across submarkets that compete for different people. Ask whether the city definition is administrative or functional.
5. When was it collected? Not when it was published. The two can be six months apart, and in a market where hiring conditions move within a quarter, a six-month-old collection date presented as current data will set your offer level in the wrong place. If the collection window is not stated, treat the report as directional at best.
What actually decides a city
Cost is the input operators lead with and the one that matters least to whether the build works. In our own hiring across our locations, the decision has consistently turned on four other things.
Depth in the specific roles. Not the size of the talent pool, but the depth of it at the seniority you need. A city with an enormous pool of entry-level candidates and a thin layer of people who have run the function before is a city where you will fill the team and struggle to lead it. The senior layer is the constraint in almost every build we have done.
The competitive set for those roles. Who else is hiring the same profile in that city, and what are they. Competing with global captives for a profile is a different problem from competing with domestic services firms, and it changes both the offer and the retention plan.
Client access and time zone overlap. If the function needs live overlap with a US team, the practical working window differs enough between cities and roles to matter. This is an operating constraint, not a preference, and it is easier to design for at the start than to fix later.
Operating risk. Infrastructure reliability, commute patterns, and how concentrated your exposure becomes if the whole function sits in one location. A multi-city posture costs more to run and removes a category of single-point risk, and whether that trade is worth making depends on what the function does.
What changed going into 2026
The change worth naming is qualitative and we can state it without pretending to have measured it. The profile that has become harder to hire is the senior individual contributor who can own a workflow end to end and exercise judgment inside it, because that is the profile an AI-native operating model needs more of and a headcount-shaped operating model needed less of. Demand for that layer has risen faster than the supply of people who have actually done the work.
The corollary is that the entry-level hiring problem has eased in exactly the functions where governed AI roles now absorb the routine processing. This is not a cost story. It is a shape story, and it means a hiring plan built from a two-year-old org chart will be recruiting for the wrong layer.
A practical way to use this
Before you commit to a city or a headcount plan, write down the eleven roles you actually need with their level definitions, then ask for benchmarks against those definitions rather than against titles. Ask for total cost to company. Ask when the data was collected. If the answer to any of those is unavailable, you have learned something useful about the report.
And do the org-chart work before the hiring plan. The most expensive version of this mistake is not paying too much for a role. It is hiring eleven people into a shape that a redesign would have made into six.