Transfer Pricing Benchmarking Study in India (Guide)

Transfer Pricing Benchmarking in India (Guide)
HomeInsightsTransfer Pricing › Transfer Pricing Benchmarking Study in India
Transfer Pricing
AS
Ankit Sarawagi|Founder, CFOmatrix·August 2026·10 min readTransfer Pricing

Your Indian entity bills the US parent cost plus a markup, and someone, the auditor signing Form 3CEB or the officer reviewing it, will ask one question: how do you know that markup is arm’s length?

The answer is a transfer pricing benchmarking study. It is the piece of analysis that takes your intercompany margin and tests it against what independent companies doing similar work actually earn. This guide walks through the full mechanics: characterising the tested party through a FAR analysis, picking the method and profit level indicator, searching comparables in a database like Prowess or Capitaline, applying filters, building the 35th to 65th percentile range, and what happens to your tax bill when your margin falls outside it.

Benchmarking study at a glance
What it proves
That your intercompany price or margin is at arm’s length under Sec 165 (old 92C).
Core steps
FAR analysis, choose method and PLI, search comparables, apply filters, build the range.
The range
35th to 65th percentile of the comparables (when 6 or more comparables exist), under Rule 10CA.
If outside
Arm’s length price is the median; profit is adjusted up to that median.
Where it lives
In your TP study / Local File (Sec 171, Rule 10D) and behind your Form 3CEB.
35th-65thPercentile range that is treated as arm’s length
6+Comparables needed for the range to apply
MedianThe adjustment point when your margin is outside

1 What a transfer pricing benchmarking study is

A transfer pricing benchmarking study is the documented analysis that supports the price or margin on your international transactions with an associated enterprise. The law (Sec 165, old 92C) says related-party transactions must be priced at arm’s length, meaning the price two unrelated parties would have agreed. Benchmarking is how you demonstrate that with real, market data instead of an assertion.

For most Indian startups the transaction is a captive set-up: an Indian development or support centre billing a foreign parent on cost plus a markup. The benchmarking study asks a simple question with a rigorous answer, which is what margin independent Indian companies doing comparable work actually earn, and does your margin sit inside that band.

NoteThe study is not optional paperwork. It is the backbone of your TP study / Local File under Rule 10D and the number your CA relies on when signing Form 3CEB. No credible study means no real defence in an assessment.

2 Step one: characterise the tested party (FAR)

Before you compare anything, you have to know what you are comparing. That starts with a FAR analysis: the Functions performed, the Assets employed, and the Risks assumed by each party to the transaction. FAR is what tells you which side is the simpler one to benchmark and what kind of independent companies are a fair match.

You then pick the tested party, normally the entity with the less complex functions and no valuable, unique intangibles, because that is the side you can reliably find comparables for. In a captive set-up the Indian entity is almost always the tested party: it writes code or runs a process to the parent’s specification, it does not own the IP, and it does not carry market or entrepreneurial risk.

Example

Brewly runs a 40-person engineering team in Bengaluru that builds features for its US parent’s SaaS product. FAR analysis shows the India team performs routine development functions, uses no unique intangibles (the IP and customer contracts sit with the US parent), and carries no market risk (it gets paid whether or not the product sells). Conclusion: Brewly India is a low-risk captive service provider, and it is the tested party.

CFO lensThe FAR write-up is where disputes are won or lost, more than the arithmetic. If your intercompany agreement and your actual conduct both describe a low-risk captive, a modest markup is easy to defend. If your team quietly owns roadmap, hiring and product risk, an officer can argue you deserve far more than a routine markup. Read our captive dev-centre transfer pricing guide for how to keep conduct and contract aligned.

3 Step two: choose the method and the PLI

Indian rules recognise five methods: CUP, Resale Price Method, Cost Plus Method, TNMM and Profit Split. You must select the most appropriate method for the transaction. For captive service centres, where you rarely have an internal or external price for the exact same service, the Transactional Net Margin Method (TNMM) is the workhorse, because it compares net operating margins rather than prices.

Under TNMM you pick a Profit Level Indicator (PLI): the ratio you will compare across companies. For a cost-plus captive the PLI is the net cost plus markup, that is Operating Profit divided by Operating Cost (OP/OC). A 15% PLI simply means the entity earns operating profit equal to 15% of its operating costs.

Set-upUsual methodPLI
Captive software / ITeS on cost-plusTNMMOP / OC (net cost plus markup)
Distributor buying finished goods to resellRPM or TNMMGross margin, or OP / Sales
Identical product with an outside priceCUPPrice itself
TipThe method and PLI are not a free choice you make later to fit the answer. Document why the chosen method is most appropriate and why the others were rejected. For the full comparison see our guide to the five transfer pricing methods in India.

4 Step three: search comparables and apply filters

Now you find the independent companies. Indian studies run the search on a commercial database of audited financials, most commonly Prowess (CMIE) or Capitaline. You start with a broad pool by industry keyword or code, then narrow it with a documented set of filters until only genuinely comparable companies remain.

1
Set the search universe
Pull companies in the same broad activity (say, software development services) from Prowess or Capitaline for the relevant years.
2
Apply quantitative filters
Drop companies with insufficient data, different year-ends, persistent losses, or export/related-party revenue outside your chosen thresholds.
3
Apply qualitative filters
Read business descriptions and reject companies with different functions, owned intangibles, or product versus service models. This is where FAR pays off.
4
Compute each comparable’s PLI
Calculate OP/OC for every surviving company, using multiple-year data where the range applies (current year and up to two preceding years).
5
Build the arm’s length range
Rank the margins and derive the 35th to 65th percentile band. This is the set you test your own margin against.
Note on multiple-year dataWhere the range applies, each comparable’s margin is taken over the current year and, where relevant, the two preceding years (a weighted average), so a single distorted year does not swing the result. The tested party’s own margin is generally tested on the current year.

5 Step four: the 35th to 65th percentile range

This is the heart of the study. Under Rule 10CA, once your dataset has six or more comparables, you do not use a simple average. You build an arm’s length range that runs from the 35th percentile to the 65th percentile of the comparables’ margins. If your tested party’s margin falls anywhere inside that band, the price is accepted and no adjustment is made.

If your margin falls outside the band, whether below the 35th or above the 65th, the range benefit is lost. The arm’s length price is then deemed to be the median (the 50th percentile) of the comparable set, and your taxable margin is adjusted to that median.

Worked example: inside the range

Brewly India is benchmarked as a captive software developer. The final comparable set gives an arm’s length range of 12% to 18% OP/OC, with a median of 14.5%. Brewly’s own margin for the year is 15%. Because 15% sits between the 35th percentile (12%) and the 65th percentile (18%), the margin is within the range and accepted. No transfer pricing adjustment, no extra tax.

Worked example: below the range, adjusted to median

Now take the same comparable set (range 12% to 18%, median 14.5%), but suppose Brewly billed thin and earned only a 9% margin. 9% is below the 35th percentile of 12%, so it is outside the range. The arm’s length price is deemed to be the median, 14.5%, and Brewly’s operating profit is adjusted up from 9% to 14.5%. On an operating cost base of, say, ₹20 crore, that is extra operating profit of (14.5% − 9%) × ₹20 crore = ₹1.1 crore added to Indian taxable income, plus interest, and a potential penalty exposure.

Watch outThe adjustment is always to the median, not to the nearest edge of the range. Being just 1% below the 35th percentile does not get you adjusted to the 35th percentile; it drags you all the way to the 50th. That asymmetry is exactly why setting your markup a hair too low is one of the four classic captive mistakes.

The tolerance band, and when the range does not apply

The 35th to 65th range only kicks in with six or more comparables. If you end up with fewer than six, you fall back to the older mechanism: the arithmetic mean of the comparables, with a tolerance band. If your price is within the tolerance band of that mean, no adjustment is made. The band is 1% for wholesale trading and 3% for all other cases. If you are outside it, the arithmetic mean itself becomes the arm’s length price.

Number of comparablesArm’s length benchmarkIf tested margin is outside
6 or moreRange: 35th to 65th percentileAdjust to the median (50th percentile)
Fewer than 6Arithmetic mean, with a 1% / 3% tolerance bandAdjust to the arithmetic mean
TipDo not confuse the two. The 1% / 3% tolerance band belongs to the small-sample arithmetic-mean method. When the range applies, there is no tolerance band: you are either inside the 35th to 65th band or you are pulled to the median.

6 How the pieces fit together

Put end to end, a benchmarking study is a short, logical chain. Each link has to hold, because a weak FAR or a poorly filtered comparable set undermines everything that follows.

A benchmarking study does not set your intercompany price. It tests the price you already charge, and the market, not your opinion, decides whether it survives.
CFO lensRun the study before the year-end, not after. If your live margin is drifting toward the bottom of a likely range, you can still true up the intercompany invoicing for the remaining months and land comfortably inside the band. Discover it only when the study is prepared in October, and your only options are an adjustment or an argument.

7 Your benchmarking checklist

  1. Write the FAR analysis first, and confirm the tested party is your low-risk entity (usually the Indian captive).
  2. Select the most appropriate method (TNMM for most captives) and state why the others were rejected.
  3. Fix the PLI: OP/OC for a cost-plus captive, and define operating cost and operating profit consistently.
  4. Run the comparable search in Prowess or Capitaline and document every quantitative and qualitative filter.
  5. Compute each comparable’s margin using multiple-year data where the range applies.
  6. Build the 35th to 65th percentile range if you have 6 or more comparables; otherwise use the arithmetic mean with the 1% / 3% tolerance band.
  7. Compare your tested margin to the range; if outside, model the adjustment to the median and its tax cost.
  8. File the study inside your Rule 10D documentation and reconcile it to the numbers in Form 3CEB.

Not sure if transfer pricing even applies to you?

Use our free Transfer Pricing Applicability Checker: tell us about your foreign parent or subsidiary and what flows between you, and see which filings you owe, from Form 3CEB to the TP study, Master File and CbCR.

Check my transfer pricing

8 FAQs

What is a transfer pricing benchmarking study?

A benchmarking study is the analysis that proves your intercompany price is at arm’s length. You characterise the tested party through a FAR analysis (functions, assets and risks), select the most appropriate method and profit level indicator, search a database such as Prowess or Capitaline for independent comparable companies, apply filters, and build an arm’s length range from their margins. Your own margin is then compared against that range.

What is the 35th to 65th percentile range in Indian transfer pricing?

Under Rule 10CA, when the dataset has six or more comparables, the arm’s length range is the set of values from the 35th percentile to the 65th percentile of the comparables’ margins. If the tested party’s margin falls anywhere inside that range, the price is accepted with no adjustment. If it falls outside, the arm’s length price is taken as the median (the 50th percentile) and an adjustment is made to that median.

What happens if my margin is below the arm’s length range?

If your margin is outside the 35th to 65th percentile range, whether below or above, the range benefit is lost. The arm’s length price is deemed to be the median of the comparable set, and your taxable profit is adjusted up to that median. For a captive service centre this usually means a higher markup and higher Indian tax.

Which database is used for transfer pricing comparables in India?

Indian benchmarking studies typically use commercial databases of company financials such as Prowess (CMIE) and Capitaline. These hold audited financials of Indian companies, which lets you screen for independent companies performing similar functions and compute their operating margins over multiple years.

Is multiple-year data used in a benchmarking study?

Yes. Where the arm’s length range applies, the comparables’ margins are computed using current-year data and, where relevant, the two preceding years, to smooth out one-off distortions. The weighted average margin of each comparable across those years is what goes into building the range.
Sources: Income-tax Act, 2025, Sec 165 (arm’s length price and methods, old 92C) and Sec 171 (documentation, old 92D); Income-tax Rules, Rule 10B (methods), Rule 10CA (arm’s length range, 35th to 65th percentile, median and tolerance band) and Rule 10D (documentation); Form 3CEB requirement under Sec 172 (old 92E). The 35th to 65th percentile range for 6 or more comparables, the median adjustment, and the 1% (wholesale) / 3% (other) tolerance band for smaller samples verified as of August 2026. Comparable databases (Prowess, Capitaline) are commercial tools; specific filters are a matter of professional judgement.
AS
Founder, CFOmatrix  |  Finance Strategy & Compliance

CFOmatrix helps Indian startups build finance, tax and compliance functions that stand up to investor due diligence, from process and controls to the filings and the numbers behind them.

Disclaimer: This article is general information as of August 2026 and is not tax or legal advice. Transfer pricing rules, section numbers under the Income-tax Act, 2025, and the mechanics of Rule 10CA can change and turn on the specific facts of your case. Confirm your benchmarking approach with a qualified transfer pricing professional before filing.

What do you think?

Leave a Reply

Your email address will not be published. Required fields are marked *

Insights

More Related Articles

FEMA / FDI Filing Checker

External Commercial Borrowing (ECB) in India: A Guide

Downstream Investment & Form DI in India (FEMA)