TP Benchmarking Databases Compared (India and Global Sources)

The comparable pool is only as good as the data behind it, and the data is only as good as the source. Indian studies draw on a small set of standard sources, each with a different coverage, a different definition discipline and a different failure mode. Choosing the source — and knowing what it does not contain — is part of the search design, not a background decision.

The standard Indian sources

Source	What it is	Strength	Gap
Prowess (Prowess IPM)	The standard Indian corporate financial database (filings-based)	Full coverage of listed Indian companies, 10+ years, segment data where disclosed, NIC mapping	Definitions must be rebuilt to the study’s operating cost/profit lines; unlisted companies absent
Capitaline	Indian financial database, widely used for TP screening	Comparable coverage to Prowess, strong ratios and peer-group tooling	Same filings-based constraints; segment granularity varies by company
BSE/NSE + MCA filings (primary)	The raw annual reports and financial statements	The ground truth — every number in a database comes from here	Manual, slow, and the segment/related-party detail lives in the notes, not the face of the accounts
Big-4 industry studies (PwC, KPMG, EY, Deloitte)	Annual sector benchmarking studies (KPO, pharma, IT, distribution)	Curated PLI distributions by sub-sector, published methodology	The pool and definitions are the firm’s, not yours — usable as a cross-check, not as your Accept-Reject matrix
Global databases (Orbis, Capital IQ, AmicUS, Refinitiv)	Cross-border company financials	Non-Indian comparables for cross-border fact patterns; US/global pools for US-side files	Indian definition mapping is worse; local-currency translation and period mismatches; coverage of private companies is patchy

The pattern for an Indian study: Prowess or Capitaline for the candidate list, primary filings for the numbers that matter (segment revenue, related- party sales, the operating cost build-up), and a published industry study as the sanity check on where the range should sit.

What a TP-grade source must give you

A screening database is a TP data source only if it supports the full quantitative and qualitative screen:

1. Industry classification — the NIC-2008 5-digit code (or NACE for foreign comparables), not just a free-text industry label. See search design for why the code is the load-bearing filter.
2. Multi-year financials — at least the tested party’s period plus the adjacent years, so multi-year averaging and trend checks are possible.
3. Size proxies — revenue, total assets, and where available employee count — for the size screen.
4. Segment data — where a company does the tested function and other things, the segment (or at minimum the related-party revenue split) is the difference between a comparable and a false comparable.
5. Related-party disclosures — revenue and transactions with affiliates, from the notes. A company whose revenue is 40% related-party is not an “independent” comparable on the revenue lines.
6. Auditability — every number traceable to a filing or a stated source. The TPO examines the matrix company by company; a number you cannot point to a source for is a number you cannot defend.

The definition problem

Every database computes its ratios on its own operating definitions. A database “EBIT margin” is not your study’s OP/OC: the operating profit line may include or exclude interest income, other income, prior-period items, and the operating cost base may or may not include depreciation or employee benefits the way your study defines it. The discipline is:

• Pull the lines, not the ratios. Rebuild the PLI from the income statement lines on your study’s definitions, for the tested party and every comparable, identically.
• Document the mapping — database line → study line, for each company where the mapping was not automatic (and most mappings are not).
• Reconcile at least the pool’s top companies to primary filings, where the database value and the filing diverge, the filing wins and the database value is corrected in your working file.

Public filings + entity data as a source

Where the database coverage is thin (unlisted comparables, cross-border companies), the fallback is primary: annual reports, exchange filings, and entity-level enrichment — the registered legal name, LEI, ownership links and business description pulled from registries such as GLEIF, plus the company’s own website for the product and function description. That is the source chain behind a defensible qualitative screen for companies the databases handle poorly — and it is exactly the enrichment step a benchmarking platform runs automatically (entity resolution, LEI linkage, business- description capture) so the qualitative review works on verified company identities rather than on fuzzy database names.

Choosing the source for the study

Study characteristic	Source posture
Listed-industry Indian services (IT, KPO, pharma)	Prowess/Capitaline pool + Big-4 study cross-check
Unlisted or niche industry	Primary filings + entity enrichment; smaller pool, documented
Cross-border (US/EU comparables needed)	Global database for the foreign pool + Indian pool for the local anchor
Distribution with thin reseller pool	Primary reseller financials where possible; consider GMM on a small pool with the n-threshold disclosed

The source choice is recorded in the Local File — which database, which year, which definitions — because a pool you cannot reconstruct from the stated source is a pool the TPO will treat as a construction.

See also

• Benchmark search design
• Quantitative screening
• Qualitative screening
• The benchmarking pillar guide