Yes, an audit is still needed even if scans are running regularly. Scans flag approximately 25% of accessibility issues. The remaining 75% requires manual evaluation by a qualified professional. Scans evaluate HTML, CSS, and ARIA attributes against programmatically detectable rules, but most WCAG success criteria cannot be assessed by code inspection alone. An audit identifies the issues scans cannot see, including problems with screen reader output, keyboard operation, focus management, and content meaning.
| Key Point | What It Means |
|---|---|
| Scan coverage | Scans detect approximately 25% of accessibility issues against WCAG success criteria. |
| Audit coverage | An audit evaluates the full set of WCAG 2.1 AA or 2.2 AA success criteria, including the 75% scans miss. |
| Evaluation method | Scans use automated rule checking. Audits use screen reader testing, keyboard testing, visual inspection, and code inspection. |
| Output | Scans produce flagged items with locations. Audits produce a report identifying issues, severity, and remediation guidance. |
| Use together | Scans support ongoing monitoring between audits. They do not replace the (manual) audit. |
What Scans Actually Evaluate
Accessibility scans load a page and run automated checks against WCAG success criteria. The checks evaluate HTML structure, CSS properties, and ARIA attributes. A scan can detect things like missing alt attributes, empty form labels, invalid ARIA roles, and improperly nested headings.
These are valuable findings. They are also the easier category of accessibility issues to identify because the rules are deterministic. The code either has the attribute or it does not.
The 25% figure reflects this reality. Scans catch what can be reduced to a rule. Most WCAG criteria cannot.
What Scans Cannot Evaluate
Many WCAG success criteria require human judgment. A scan cannot determine whether alt text accurately describes the image, whether a form error message is helpful, or whether the reading order on a page makes sense to a screen reader user.
Scans cannot assess keyboard operability across interactive components, focus management in modals and menus, or whether content announces correctly to assistive technology. They cannot evaluate whether headings reflect the document outline or whether link text conveys purpose in context.
This is the 75% that requires manual evaluation. An auditor uses screen readers like NVDA, JAWS, and VoiceOver, tests keyboard operation in Chrome and Safari, inspects code, and reviews the page at 200% and 400% browser zoom. The audit identifies issues that no scan can flag.
How Scans and Audits Work Together
Scans and audits answer different questions. A scan answers: what programmatically detectable issues exist on this page right now? An audit answers: does this page conform to WCAG 2.1 AA or 2.2 AA across all applicable success criteria?
The two work in sequence and in parallel. A scan runs as one component within the audit methodology, catching low-hanging items efficiently. The auditor then evaluates everything else manually. After the audit, scheduled scans monitor for regressions on the items scans can detect, such as a developer shipping an image without alt text or removing a form label.
Scheduled scans support ongoing monitoring. The audit produces the conformance evaluation. One does not replace the other.
What Happens If You Rely on Scans Alone
Relying on scans alone leaves the majority of WCAG criteria unevaluated. A clean scan report does not indicate WCAG conformance. It indicates that the rules a scan can check did not return errors.
Organizations that present a passing scan as evidence of accessibility are reporting on 25% of the picture. Legal risk, procurement requirements, and actual user experience all turn on the 75% that was never evaluated.
When decision-makers ask whether a product meets WCAG 2.1 AA, the supporting evidence is a (manual) audit report, not a scan dashboard. Quality accessibility programs combine both: audits to evaluate conformance, scans to monitor between evaluations. For a closer look at what manual evaluation covers that scans cannot, see the manual evaluation pillar.