A useful starting point
What automation does well
Automated tools can find repeatable, machine-detectable defects quickly across many pages. Examples include missing form labels, duplicate identifiers, invalid relationships and some colour-contrast failures.
They are especially useful during development because the same rules can run after every meaningful code change.
What a score misses
A tool cannot reliably decide whether alternative text communicates the purpose of an image, whether focus order follows the intended task or whether an error message helps someone recover.
It also cannot complete every real customer journey using a keyboard or screen reader. A high score may therefore coexist with a menu, checkout or authentication step that remains impossible to use.
What an accessibility audit adds
A meaningful audit combines automation with knowledgeable human evaluation. The sample should represent templates, shared components, different content structures and complete user processes.
- Keyboard-only navigation and visible focus
- Screen-reader names, roles, states and announcements
- Zoom, reflow and text-spacing behaviour
- Forms, validation and error recovery
- Dynamic components and overlays
- Complete task and transaction journeys
How to use scores responsibly
Treat an automated score as a signal, not a compliance result. Use it to find common defects, monitor regressions and guide investigation.
A better question
Instead of asking whether the score is high, ask which critical tasks were tested, which assistive technologies were used and what evidence supports the conclusion.
A practical next step
Run an automated check, then manually test the main navigation, one form and one complete customer process with only a keyboard. Record where focus moves, whether changes are announced and whether errors can be corrected.
If the website serves customers or supports important transactions, use a representative WCAG 2.2 AA audit to establish a reliable baseline.