Compensation

What is compensation benchmarking? A practical guide for Indian HR teams

Compensation benchmarking (also called salary benchmarking) means comparing what you pay for a role with what the market pays for a similar role, so that offers are competitive and pay bands are fair and defensible.

Why benchmark pay

  • Make offers candidates accept, without overpaying.
  • Spot employees paid below market before they get an outside offer.
  • Build pay bands you can explain to employees, managers and the board.
  • Budget next year's increments from market movement rather than guesswork.

Step 1: level your jobs

Benchmarking compares jobs, not job titles. A "Senior Engineer" at a startup can match an SDE-II at a larger company. Define levels using scope, complexity and impact (for example IC1–IC8 for individual contributors, M1–M6 for managers), then map every role to a level before you compare pay.

Step 2: choose your peer group

Compare against the employers you actually compete with for talent. In India that usually means filtering by:

  • Industry: IT services, product/SaaS, BFSI, GCCs and others pay very differently.
  • Company stage: seed, Series A–D, late-stage or listed.
  • City: pay for the same role varies widely between metros and Tier 2 cities.

Step 3: pick your data sources

SourceStrengthsWatch out for
Salary surveys (e.g. Aon, Mercer, WTW)Structured, levelled, widely acceptedAnnual cycle; data can be months old
Employee-reported sites (e.g. AmbitionBox, Glassdoor)Large volume, freeSelf-reported, inconsistent titles
Job postingsLive, shows current offersMany Indian postings omit pay
Statutory and public filingsVerifiableLimited to certain roles and companies

Combine several sources and age older data forward with the market movement since it was collected.

Step 4: read the percentiles

  • P25: a quarter of the market pays less than this. Pay here and expect more lost offers.
  • P50 (median): the middle of the market and the most common target.
  • P75: premium positioning, often reserved for critical or scarce skills.

Decide your pay philosophy, for example "P50 for most roles and P75 for niche engineering", and apply it consistently.

Step 5: compare like with like (CTC in India)

Indian offers are usually quoted as cost to company (CTC), which bundles very different components. Benchmark each part separately:

  • Fixed pay: basic, HRA and allowances.
  • Statutory and retirement: employer PF contribution and gratuity.
  • Variable pay: target bonus. Also check how often it actually pays out.
  • Equity: ESOPs or RSUs, valued consistently (see our ESOP valuation guide).
  • Benefits: insurance, learning budgets and similar perks.

Two metrics to track

Compa-ratio = Employee's pay ÷ Midpoint of their pay band

A compa-ratio of 1.0 means the employee is paid exactly at the band midpoint. Below about 0.9 is a retention risk to review.

Range penetration (%) = (Pay − Band minimum) ÷ (Band maximum − Band minimum) × 100

How often to benchmark

At minimum, once a year before the increment cycle. For hot skills such as GenAI or cybersecurity, check quarterly. Pay for these roles can move quickly, and annual surveys miss it.

Benchmark with live data

Kynoa's salary benchmarking for India refreshes weekly and breaks pay down by role, level, company stage and city, including ESOPs and variable pay.

Frequently asked questions

What is the difference between salary benchmarking and compensation benchmarking?

Salary benchmarking usually compares base or fixed pay. Compensation benchmarking compares total rewards: fixed pay, variable pay, equity and benefits.

Which percentile should we pay at?

Most companies target the median (P50) for most roles and pay closer to P75 for scarce or business-critical skills. What matters is choosing a philosophy and applying it consistently.

How many data points do we need for a reliable benchmark?

More is better, but the match quality matters more than the count. A smaller sample of well-levelled, recent, same-city data beats a large mixed sample.

See Kynoa on your own data Book a demo (opens in a new tab)