Skip to content

How we audit your site

We record what an audit measured, what it reported, and what it could not check.

What we audit

We audit representative pages from your site's templates, or a page you choose.

The report names the pages and screen sizes it covered. A representative page does not establish what happens on every other page.

What the report distinguishes

We report a measured failure only when a deterministic check has evidence for it. Model-written observations and advisory suggestions are kept separate from measured failed checks.

Each finding shows its rule, the page it was found on and the evidence we recorded.

How we count failed checks

We count a failed check once per rule and page, even when it affects several elements. That count is not the number of affected elements.

What remains unmeasured

If a check could not run, we report it as not measured. We do not treat missing measurements as passes.

A report with no failed checks does not establish that everyone can use your site.

We do not run a screen reader or test assistive technology. Human usability research and assistive-technology testing remain separate work.

How we check a reported change

Marking a fix applied records your progress; it does not resolve the task.

A later audit confirms a fix when the same check runs again and no longer finds the violation. That doesn't prove your code change is correct.