FOCUS for AI spend: what the open billing standard actually covers
FOCUS — the FinOps Open Cost and Usage Specification — is an open, vendor-neutral schema for billing data, maintained by the FinOps Foundation. The problem it solves is simple to state: AWS, Azure, GCP, and every SaaS and API vendor ship cost data in their own proprietary shape. A FinOps platform trying to show one unified view of spend across all of them either builds a custom parser per vendor, or asks vendors to export in a shared format it already understands. FOCUS is that shared format.
For AI spend specifically, this matters because most teams already have — or are building — a FinOps practice around cloud infrastructure, using tools like CloudZero or Apptio that expect FOCUS-shaped data. An AI cost tool that only exports its own bespoke CSV forces a second, disconnected reporting pipeline. One that exports FOCUS-conformant data plugs into the reporting your finance team already has.
“FOCUS-aligned” is often a vague claim. Here is what we actually tested.
A lot of vendors say their export is “FOCUS-aligned” without saying what that was checked against. FOCUS publishes an official validator — a tool that runs a real export through the specification's actual conformance rules and reports pass/fail per rule, not a marketing gut-check. We run our export through that official validator directly, and we're publishing what it actually says, including the parts that aren't fully resolved yet.
What we found testing against the official validator
- •FOCUS 1.2 — the version we target. Testing a real export against the official validator, every genuine rule violation we could find has been resolved. A remaining handful of reported failures were investigated individually and confirmed to be validator-engine false positives — for instance, several relate to columns we verified are present in the export via a direct database query, with zero actual violations, but the validator's dependency-resolution logic cross-references them against unrelated rules and reports a failure anyway. We're not hiding that these show up in a raw validator run; we're telling you why they're not real gaps, and re-testing periodically to make sure that stays true.
- •FOCUS 1.1 — currently untestable. The official validator dropped support for checking against 1.1 entirely. We make no claim of 1.1 conformance, because there is currently no way to verify one either way.
- •FOCUS 1.3 — not yet met. It's a newer, stricter version of the spec than 1.2, and we're further behind on it. It's not a near-term claim we're making.
Why we're stating it this way instead of a badge
A compliance badge implies a permanent, binary state. Real conformance testing against a living specification, run by a validator that itself gets updated, isn't binary or permanent — it's a result from a specific test run, against a specific validator version, on a specific export. We'd rather tell you exactly what we tested, when, and what came back, than compress that into a single word that stops being precise the moment either the spec or the validator changes.
What this means for your export
Cognocient's FOCUS export is available from the dashboard and designed to plug directly into FOCUS-aware FinOps tooling — the same reporting pipeline you likely already use for cloud infrastructure spend, extended to cover AI API spend without a second, disconnected export format.