Limits

What automated accessibility checks can and can't find, and where Monitoring's job ends.

What automated checks can find

Automated rules — including axe-core, the engine Monitoring runs — find a minority of accessibility barriers when run as a single tool; combined, or measured by volume, they find more. How you measure it changes the number. In a 2017 test, the UK Government Digital Service ran ten tools against 143 deliberately introduced failures on one page: combined, the tools found 71%, and the best single tool found 37% (GDS, 2017 (opens in a new tab)). Deque, who make axe-core, reported in 2021 that their automated tests found 57.38% of issues by volume across more than 2,000 audits — a vendor-reported figure (Deque, 2021 (opens in a new tab)). These figures don't measure the same thing, so treat any of them as a statement about a ceiling, not a score to hold to two significant figures.

In practice: a clean report means the automated rules found nothing on the pages that were sampled. It is not a claim that your site is accessible. Monitoring widens what gets tested — real states, real journeys, real visitors — but it doesn't raise that ceiling. Manual testing with real assistive technology, by the people who use it, remains the only thing that does.

See Why automated scans miss real accessibility issues for the full picture.

What Monitoring does not do

See the Monitoring overview for what Monitoring does not do — no remediation, no assistive-technology detection, no compliance claims.