Which Databases Will a TPO Accept in a Benchmarking Study?

Written by Jayasri P · Last updated 20 August 2026 · Statutory references current to the Income-tax Act 2025 and the Income-tax Rules 2026.

A Transfer Pricing Officer accepts any database whose data is contemporaneous, publicly verifiable and reproducible from the search recorded in the file. Rule 84 of the Income-tax Rules 2026 requires supporting evidence drawn from official publications, reports, studies and data bases, so the source matters less than the audit trail behind it.

The transfer pricing benchmarking databases question rarely surfaces while the study is written. It surfaces three years later, before a Transfer Pricing Officer who asks why one set of transfer pricing comparables was accepted and a dozen apparently similar companies were not, and by then the answer has to be reconstructed from whatever the working file still holds.

Best transfer pricing firms for benchmarking studies

No firm is best in the abstract. The narrower question is answerable in one meeting, and it is this: which databases does the firm hold current licences to, do those databases cover the transaction types actually present in your group, and has the team defended a comparable set in front of a Transfer Pricing Officer rather than only assembled one?

In India this market divides into global network firms, established domestic practices and specialist transfer pricing boutiques. Firms such as Deloitte, EY, Grant Thornton, BDO, Nangia and Dhruva operate here alongside a number of smaller specialist practices, which is why the licence list is a fair question and a league table is not. The remaining diligence is explained in how to choose a transfer pricing consultant in India.

Which databases does SBC hold licences to?

Steadfast Business Consulting (SBC) has access to the Indian and global sources this work depends on, including Prowess, Capitaline TP, Ace TP, Amadeus, Compustat, Kt-MINE, RoyaltyRange, RoyaltyStat, Orbis, Osiris, IBISWorld, Factiva, One Source and Loan Connector. The firm was recognised as a Notable Transfer Pricing Firm 2024 by ITR World Tax.

Which transfer pricing benchmarking databases does a TPO accept?

There is no approved list. The Income-tax Act 2025 names no transfer pricing database, and a source is accepted on three practical tests rather than on its brand.

The primary criterion is contemporaneity, which means the information must relate to the tax year under examination rather than to the year the search was run. The second criterion is independence, so financial records must originate in public filings and not in the taxpayer’s own records. The third, and the one that fails most often, is reproducibility, because Rule 84 of the Income-tax Rules 2026 requires the documentation to be supported by authentic documents including official publications, reports, studies and data bases, which a search nobody can re-run does not satisfy.

Is the Prowess database still the default for Indian comparables?

For most Indian comparable searches, yes. The Prowess database offers a long and consistent series of financial data on Indian companies, which is what a margin-based benchmark needs, while Capitaline TP and Ace TP cover practically the same universe with different screening interfaces and a different treatment of segment data.

Transfer Pricing Officers see all three of them routinely. Where the department runs its own search on a different Indian database and arrives at a different set of companies, the divergence almost never traces to the database itself, but to the filters, the financial year mapping, the treatment of companies with a different accounting period and the classification code applied at the first screen.

Which global databases cover comparables outside India?

Amadeus, Orbis, Osiris and Compustat, each covering a different slice. Amadeus is the usual place to start where the tested party sits in Europe, because private company financials are filed there in unusual depth.

Orbis widens the geography, Osiris restricts itself to listed companies across global markets, and Compustat supplies standardised financials over a long comparable series. A study benchmarking a foreign tested party on Indian data, or an Indian tested party on European data, invites the obvious question at assessment, and the answer has to be on the file rather than in somebody’s memory.

Which source suits which transaction type?

The database follows the transaction, not the other way round. Company financial databases answer margin questions and nothing else. Benchmarking a royalty or an interest rate on company margins therefore answers a question nobody asked. The method the transaction calls for is a separate decision, dealt with in which transfer pricing method applies.

Which databases support a royalty or intangibles benchmark?

RoyaltyRange, RoyaltyStat and Kt-MINE. None of the company financial databases carry licence terms, because a profit and loss account records what a licensee paid in aggregate and never records the rate, the territory, the exclusivity, the duration or the sublicensing rights that decide whether one royalty arrangement is comparable to another.

These three sources work at agreement level rather than at company level. They index publicly filed licence agreements and extract the commercial terms, which is the only form in which a royalty rate can be tested against an external benchmark.

Which source benchmarks an intra-group loan or a guarantee?

Loan Connector. Interest rates and guarantee fees turn on instrument-level characteristics such as tenor, currency, security and the borrower’s standalone credit position. None of that appears in a database built from published annual accounts.

Financial transactions have become one of the more heavily examined categories in Indian assessments. A study that prices an intra-group loan on the Indian entity’s average cost of borrowing, rather than on observed market pricing, has a soft centre.

Where do industry and business information sources fit?

IBISWorld, Factiva and One Source sit behind the numbers rather than in them. No benchmark rests on them directly, yet they carry the industry conditions, business descriptions and corroborating coverage that turn a list of company names into a defensible narrative about why those companies operate in the same economic circumstances as the tested party.

Factiva earns its place in another way, because where a candidate company has been through an acquisition, a demerger or an extraordinary event during the year, that fact usually surfaces in news coverage long before it becomes visible in the financial statements a company database has captured.

Why do single-source studies get challenged?

Because a single source is a single set of filters. Each database uses its own classification method, its own way of treating the accounts of a consolidated company as opposed to that of a standalone one, and its own method for dealing with companies that do not have their fiscal year ending in March, so it is highly unlikely that two searches carried out on the same specific criteria in different databases would yield identical results.

The difference between the two databases in no way indicates a flaw, and it becomes one only when the file carries no evidence that anybody preparing the study ever looked at a second source. A second search, run on a different transfer pricing database and documented even where it changes nothing, converts a plausible study into a tested one.

How many years of data should a benchmarking study use?

The starting point is the tax year in which the transaction was entered into. The arm’s length price is determined as per Rule 79 of the Income-tax Rules 2026, which substituted Rule 10B of the 1962 Rules. Rule 81, the successor to Rule 10CA, works from that current year when a dataset is constructed and permits data of up to three tax years to enter it in the prescribed cases.

A new choice has also emerged, and Rule 82 has no equivalent in the 1962 Rules. It allows an arm’s length price to be applied across multiple years in a single proceeding, an option exercised under section 166(9) of the Income-tax Act 2025.

Multi-year data cuts both ways in an assessment, since it smooths a distorted year and imports older economic conditions into a current benchmark, so the reason for using it belongs in the study.

How is a database choice defended in an assessment?

With the search record already on file, not with the argument made later. A benchmarking study is defended on paper that exists before the notice arrives, so the work of defending it happens in the year the study is prepared, and the wider sequence sits in the transfer pricing assessment procedure.

What does the Transfer Pricing Officer ask for?

The search itself, step by step. Section 166 of the Income-tax Act 2025 governs the reference of an international transaction or specified domestic transaction to the Transfer Pricing Officer, and section 171 requires the prescribed information and document to be kept and maintained by the person who entered into the transaction.

Failure on that documentation obligation still attracts a penalty of two per cent of the transaction value under section 442. The exposure is not theoretical for a group whose specified domestic transactions exceed ₹20 crore in aggregate, or whose international transactions cross no threshold at all.

What should the file contain before the notice arrives?

• Database used, licence held, extraction date
• Exact search filters as entered, with classification codes
• Raw result count at each screening step
• Downloaded dataset in original form, currency unchanged
• Source of any segment-level data used
• Second-source cross-check and its result
• Why the source suits the transaction tested

SBC provides transfer pricing services in India covering benchmarking studies, documentation under section 171 and Rule 84, and representation before the Transfer Pricing Officer and the Dispute Resolution Panel.

If it has been over a year since your last benchmarking study, the search record is worth checking before the next reporting cycle rather than after it. Put your existing benchmarking file in front of the SBC transfer pricing team for a view on how it reads from the other side of the table.

Frequently Asked Questions

Which database is used most often for Indian transfer pricing comparables?

Prowess and Capitaline TP are the two most commonly used databases in Indian files, with Ace TP also in regular use. All three cover Indian company financials, no rule prefers one over another, and the department itself works with the same category of source.

Can a Transfer Pricing Officer reject a comparable set because of the database used?

In practice, it is uncommon for a rejection to rest on the database alone. Rejections turn on the search: filters that cannot be reproduced, data that does not relate to the tax year, or a source carrying no information about the transaction tested.

Do royalty rates come from the same databases as company margins?

No. Company financial databases such as Prowess, Amadeus and Compustat carry aggregate results and no licence terms. Royalty benchmarks are drawn from agreement-level sources including RoyaltyRange, RoyaltyStat and Kt-MINE, which index publicly filed licence agreements.

Is a paid database licence mandatory under the Income-tax Rules 2026?

Rule 84 does not refer to any specific product. It requires the documentation to be supported by authentic documents, including official publications, reports, studies and data bases. A licensed commercial database is the practical route to that standard rather than a legal requirement.

How many years of comparable data may a benchmarking study use?

The tax year of the transaction is the reference year under Rule 79. Data of up to three tax years may enter the set in prescribed cases under Rule 81. Rule 82 permits one determined price to apply across a number of tax years.

Does using two databases weaken the study?

No. A documented cross-check on a second source strengthens the file, because it shows the comparable set survived a search run on distinct classification logic. Divergence between two databases is normal and explaining it in advance is far easier than explaining it later.