AuditAble coverage methodology
AuditAble measures coverage as evaluated WCAG success criteria divided by a published denominator. This is a product capability claim, not a WCAG conformance claim, ADA compliance certification, or substitute for expert human accessibility review.
What the percentage means
Primary denominator
50
WCAG 2.1 Level A and AA success criteria, the DOJ Title II web rule reference point for public-sector readiness.
75-80% claim range
38-40
A 75-80% claim against the WCAG 2.1 A/AA denominator means 38 to 40 criteria are evaluated by the named product version.
Report behavior
Exact
Every scan report should disclose the actual criteria evaluated for that scan. Untested criteria are marked Not Evaluated.
For WCAG 2.2 planning, AuditAble maintains a separate internal crosswalk using the 55 active Level A and AA criteria. A 75-80% claim against that denominator means 41 to 44 criteria. Public-sector product claims should name which standard and product version the percentage applies to.
Counting rules
A criterion counts as evaluated only when an executed detector, guided review flow, or approved manual attestation produces evidence for that criterion.
Axe-core rules count only when the installed rule actually runs under AuditAble scan settings and maps to the relevant WCAG criterion.
Browser interaction and accessibility-tree checks count only when they run in the scan pipeline and persist evidence or findings.
AI-assisted checks count only after validation against a labeled corpus and only for product versions where that layer is enabled.
Heuristic screening, needs-review flags, and roadmap items do not count as tested criteria unless they graduate through the same evidence gate.
A criterion that is not tested is never treated as passing. It is reported as Not Evaluated or reserved for manual review.
Versioned sign-off
A specific coverage percentage becomes externally quotable only when the version record contains the denominator, numerator, scan version, benchmark date, and methodology reviewer approval. Ryan Markham's sign-off is retained in the investor diligence packet with the approved claim wording.
Evidence record fields
- Claim wording and product version
- WCAG version, level, denominator, and numerator
- Scanner commit, enabled modules, and benchmark run ID
- Reviewer, approval date, limitations, and expiration or review date
Current disclosure rule
Customer-facing reports disclose the exact criteria evaluated for each scan. If a deck, sales script, or investor note uses a percentage range, it should link to this methodology and to the signed version record that supports that range.
Benchmark comparison
AuditAble benchmarks against axe-core as a known open-source reference engine. The benchmark is not used to copy a competitor's result; it is used to show what AuditAble adds beyond the reference baseline and where human judgment is still required.
Same target
Run AuditAble and the reference tool against the same pages, browser family, viewport set, and WCAG target.
Same evidence buckets
Compare criteria mapped, issues detected, needs-review cases, false positives, false negatives, and not-evaluated criteria.
Versioned result
Record the AuditAble commit, scanner version, axe-core version, corpus version, reviewer, and date for each benchmark run.
What we do not publish
This public page intentionally avoids exposing AuditAble's full testing recipe. The complete matrix is available to investors, counsel, auditors, or enterprise buyers under appropriate diligence terms.
Important limitations
W3C guidance is clear that no accessibility evaluation tool alone can determine whether a website meets accessibility standards. AuditAble helps teams find, prioritize, document, and monitor issues, but complete accessibility evaluation still requires knowledgeable human review and, for complex experiences, assistive technology testing.
Coverage is not the same as conformance. A scan can evaluate a criterion and still find failures, and a clean automated result does not prove a site is fully accessible. AuditAble uses readiness language and reports Not Evaluated criteria instead of treating missing evidence as a pass.