Rights and compliance · Knowledge distribution
Someone drives the critical path with a screen reader
A meaningful share of accessibility conformance is not mechanically decidable, so the critical paths have to be operated with assistive technology by a person, on a schedule, or the remainder is silently assumed to be fine.
Do this firstScan for the six failures that account for almost everything
Keyboard only, then a screen reader, through the paths that matter: sign in, search, complete the transaction, recover from an error.
The reason this cannot be automated away is the shape of what is left. A rule engine decides whether an image has an alternative; it cannot decide whether the alternative says anything useful. It sees a labelled field; it cannot tell that the focus order sends you backwards through the form. Screen-reader users consistently report the worst barriers as unpredictable interactive elements, unexpected screen changes, and nonsensical link text — every one of them a judgement rather than a rule.
Be honest about what the 57% ceiling is: an unverified claim by the people who sell the tool, and the only number available. It is the best argument for this control and it is not evidence.
The decoy
A green automated scan. The engine’s own maintainers put its ceiling at an average of 57% of issues — so a clean run certifies at most half the standard, and the half it cannot see is the half that needs judgement.
Evidence
- axe-core: the maintainers' stated ceiling on automated WCAG detection — IVThe tool's own documentation states: "With axe-core, you can find on average 57% of WCAG issues automatically. Additionally, axe-core will return elements as 'incomplete' where axe-core could not be certain, and manual review is needed." By the same claim, an average of 43% of issues are not caught by automation.
What would settle it
A measured comparison of barriers found by automated scanning against barriers found by assistive-technology testing on the same pages. The 57% figure is the tool vendor’s own claim with no disclosed corpus or methodology, and this control rests on it until someone measures the split independently.
Last reviewed 2026-08-19.