An auditor identifies what scans cannot see: meaning, context, logic, and the actual experience of using a page with assistive technology. Automated scans evaluate HTML, CSS, and ARIA against pattern-based rules and flag approximately 25% of accessibility issues. The remaining 75% requires human evaluation. An auditor reviews whether content makes sense, whether interactions work with a keyboard, whether screen reader output is accurate, and whether the page meets WCAG conformance at the level claimed.
| Capability | What It Covers |
|---|---|
| Meaning Review | Whether alt text, labels, and link text describe the actual purpose |
| Keyboard Logic | Whether focus order, focus visibility, and interactions work without a mouse |
| Screen Reader Output | Whether NVDA, JAWS, and VoiceOver announce content accurately |
| Visual Inspection | Content reflow at 200% and 400% zoom, layout integrity, reading order |
| Coverage | The 75% of WCAG criteria that scans cannot reliably assess |
Meaning and Context
A scan can detect that an image has an alt attribute. It cannot evaluate whether the alt text describes the image. An auditor reads the alt text, looks at the image, considers the surrounding content, and decides whether the description is accurate and useful.
The same applies to link text, button labels, form labels, and headings. A scan confirms the element exists. An auditor confirms the element communicates the right thing to a person who depends on it.
Keyboard Interaction
Keyboard testing is a human activity. An auditor moves through the page using Tab, Shift+Tab, Enter, Space, and arrow keys, then records what works and what does not. Custom widgets, modal dialogs, dropdowns, carousels, and date pickers often pass scans while failing keyboard testing.
Focus order, focus visibility, focus traps, and the ability to operate every control without a pointing device fall outside what an automated scan can verify.
Screen Reader Output
Screen reader testing covers how NVDA, JAWS, and VoiceOver announce content in Chrome and Safari. An auditor listens to the actual output and compares it to what a sighted user sees. Headings, landmarks, form errors, live regions, and dynamic content updates are evaluated based on what the user hears, not what the markup suggests.
A page can have correct ARIA attributes and still produce confusing or incorrect screen reader output. The auditor catches that mismatch.
Visual and Layout Inspection
An auditor zooms the browser to 200% and 400% to check whether content reflows, whether information is lost, and whether controls remain operable. The auditor reviews reading order, content grouping, and whether the visual layout matches the underlying structure.
Conformance Judgment
WCAG conformance at 2.1 AA or 2.2 AA requires meeting every applicable success criterion. A scan reports rule violations. An auditor renders a conformance judgment against each criterion, including the ones that depend entirely on human review. That judgment is what supports an accessibility statement, an ACR, or a remediation plan.
Scans serve as a fast review layer. The conformance picture comes from the manual audit that surrounds them.