Evidence Levels
Evidence Levels
Section titled “Evidence Levels”Not all evidence supports the same kind of claim.
Level 1 — Documentation evidence
Section titled “Level 1 — Documentation evidence”Architecture, contracts, runbooks, threat models, and design records can show intent, scope, and expected behavior.
They do not prove that a current build behaves as documented.
Level 2 — Executable evidence
Section titled “Level 2 — Executable evidence”Reproducible examples, tests, conformance runs, and verification outputs can show that a specific workflow behaved as expected under defined conditions.
See Executable Verification Examples.
Level 3 — Internal release evidence
Section titled “Level 3 — Internal release evidence”Build outputs, release checks, migration tests, backup/restore tests, and signed or hashed release artifacts can support statements about a specific release candidate or release.
The claim is bounded by the tested environments and procedures.
Level 4 — Pilot evidence
Section titled “Level 4 — Pilot evidence”A controlled pilot can show that a workflow works in a real operational setting with an identified owner, defined inputs, and observed outcomes.
A pilot is not automatically production evidence.
Level 5 — Production evidence
Section titled “Level 5 — Production evidence”Production evidence requires an actually operating workflow with a defined accountable owner, live status, and observed operational outcomes.
Level 6 — External verification
Section titled “Level 6 — External verification”Independent assessment can add assurance when the reviewer, scope, method, version, and limitations are recorded.
External verification should not be represented as broader than the performed assessment.