Skip to main content

How we audit — and what we will not claim

WCAG 2.1 AA has 50 success criteria. Our automated checks can prove failures on 22 of them — only those can count in the score. Another 21 are examined by advisory checks or AI review and reported without scoring, and 7 are left to a guided manual checklist. Every report states exactly which criteria were scored on that run.

Three kinds of results, never mixed

Every finding in a report carries one of three labels, and only the first one affects the score:

  • Machine-verified

    An automated check reproduced the failure in a real browser, and the evidence meets that check's strict rule for proof: either we observed the problem directly (for example, text whose measured contrast is too low), or we examined every element of a clearly defined kind and found what is required missing (for example, a page with no title at all). Every such check is itself tested in a real browser on examples that pass and examples that fail. Only these findings lower the score.

  • AI-assessed

    An AI model looked at screenshots and the page structure and flagged a likely problem — for example, alt text that does not describe the image, an unclear link name or a confusing visual structure. These are useful, clearly labelled hints, not proof, and they never affect the score.

  • Needs verification (candidate)

    An observation whose evidence cannot prove the failure on its own — a missing caption track on a video whose audio may contain no speech, for example. Reported with the exact verification step a human should take, excluded from the score.

Only what we tested is scored

The score counts only the criteria that one of our proof-grade automated checks actually tested on your pages — at most 22 of the 50. Criteria that only the AI reviewed, or that had nothing to test on your pages (a site without video gives the caption check nothing to examine), are listed in the report but not scored. If a check fails to run, its criteria are left out of the score instead of being counted as passed. A scored criterion with no confirmed problems gets full marks. With confirmed problems, its worst problem decides how much it keeps: 75% (minor), 50% (moderate), 25% (serious) or 0% (critical). The four WCAG principles are weighted 30% perceivable, 30% operable, 25% understandable and 15% robust; when a principle has no scored criteria on a run, its weight is shared among the others.

Each report also shows the possible score range: what the score would be if every unconfirmed risk turned out to be a real failure. You see both ends, not just one flattering number.

Sampled checks and their limits

Some checks look at a sample rather than at every element. Focus visibility, for example, is judged by comparing screenshots with and without keyboard focus on up to 40 elements per page, and up to 5 linked PDF documents are checked per page. When a result comes from a sample, the report says so and gives the exact numbers. A clean result then means no problems in the sample — not proof that every element is fine.

Real browser interaction, not just DOM scanning

Besides scanning the code with axe-core, the audit interacts with your site in a real browser, as a visitor would: it moves through the page with the Tab key, tries to get out wherever keyboard focus seems stuck, submits forms empty to see how error messages appear, narrows the window to 320 pixels and enlarges the text. For every page it records how the page was actually loaded — final address, HTTP status and load result.

Known limitations

We would rather tell you what an automated audit cannot do than let a number imply it did:

  • Human-judgment aspects — whether alt text is accurate, whether an error message helps, whether the focus order is meaningful — are left to AI review or the manual checklist and never counted as proven.
  • Pages behind login and flows requiring credentials are recorded as unreachable, not silently skipped.
  • Multi-step journeys (checkout, registration) are tested in their initial states; full journey completion needs a human tester and the report includes step-by-step instructions for it.
  • Some technologies can only be partly inspected — closed shadow DOM, canvas or WebGL applications and content embedded from other sites. The report says where this applies.

The reports practice what they audit

Every PDF report — including the public samples — is produced as a tagged PDF/UA-1 document, the accessibility standard for PDF documents: structure, document language and title are built in. Reports carry their scoring-model version and generation date, and the narrative is cryptographically bound to the finalized evidence it describes.

Scores from different scoring-model versions are never comparable; every report states its version. Methodology last updated September 2026.