Search, then score, then compute. In that order.
A benchmarking study that starts with a desired margin and works backwards is not advisory. It is a liability. The platform's order is the OECD order: understand the tested party (FAR), design qualitative and quantitative screens, pull a set from Prowess or TP Catalyst, apply accept/reject with reasons, compute the PLI, then write the economic analysis.
Prowess covers India. TP Catalyst covers global and Middle East sets. Together they are wired end-to-end, not "export CSV and rework in Excel." More than 200 predefined comparable search filters are available. The accept/reject matrix is rebuilt from the database export so the screening trail survives a partner review and, later, an audit. If you cannot replay the screens, you do not have a study. You have a printout. That is the standard we will show on a walkthrough.
FAR comparable scoring is explicit: 0–100, four criteria, 25 points each, live web search per company, written rationale per score. That is slower than a gut-feel tick. It is also the only version of "comparability" a reviewer can re-perform without sitting next to the analyst.
If a distributor of industrial chemicals and a distributor of fashion apparel share a database industry string, they still should not share a set. The score exists so a partner can change a weight without throwing the extract away. That is slower than a tick-box. It is also how you explain the set eighteen months later without inventing a memory.
PLI computation and the arm's length range
Instant PLI accepts up to five files of 100 MB each, PDF (text and scanned, via vision) and Excel. It detects entity type (commodity trader, distribution trader, service, manufacturing, holding/investment) and applies deterministic classification rules that override the model for known line items. Every override is audit-trailed.
Twelve PLI ratios are computed. Operating profit must equal operating revenue minus operating cost or the result is blocked. Reconciliation checks run against stated P&L totals. Nonsense guards stop a ratio when there is no revenue but material expenses. The worksheet is editable with live recompute, then Excel export in Annexure-5 style plus an extended sheet. Deterministic mode is on by default.
Quartiles and the arm's length range in the Local File are code, not prose. The model may draft the economic commentary. It does not invent a 35th percentile. Eleven tested-party margin indicators can also be computed from uploaded financials for the broader benchmarking report.
UAE benchmarking reports and the vault
Where the tested party is a UAE entity, a 51-section UAE Benchmarking Report can be generated end to end, including a UAE TP Regulations chapter (Article 34 arm's length principle) and per-transaction method, PLI, and tested-party analysis. That is advisory packaging, not a cover sheet on a global set that ignored the regulation.
The Benchmark Vault stores reusable pre-computed cases so a firm that files twelve distributors in the same NACE code does not re-invent the screens every March. Consistency across engagements is a quality control issue as much as a productivity issue. Shared templates, prompts, and master data exist for the same reason.
SBC's TP advisory practice still owns method selection, risk, and sign-off. The software accelerates the mechanical half. Read the benchmarking article and the methodology page. To see Prowess and TP Catalyst on a set you choose, request a walkthrough.
Where software stops and advisory begins
Benchmarking software cannot tell you whether a loss-making tested party should be the tested party. It cannot decide if a CUP exists. It cannot weigh a business restructuring. Those are TP advisory judgements. What it can do is stop the judgement from being starved of a clean set, a reconciling PLI, and a search memo.
We publish an 80% reduction in benchmarking time (from weeks to minutes) and 70% faster turnaround versus manual drafting because those figures already appear on the public site. We will not invent a rupee ROI. If your team currently rebuilds comparables in Excel every busy season, the walkthrough will show the delta on your own screens, not on a slide.
Independence screens, persistent-loss screens, and related-party intensity still have to be written down before the extract. A set that cannot explain why Company 37 was rejected will not survive an officer. Multi-year data should match the tested party's years, not a convenient window that ignores an outlier you already knew about.
Berry ratio versus OP/OC is a characterisation question. Instant PLI will compute both. Specialists still choose. The worksheet's job is to stop a chosen PLI that does not reconcile to the P&L. That is benchmarking hygiene, which advisory cannot skip, and which a database tick-box never provided.