Accessibility scanning tools flag approximately 25% of WCAG issues on a given page. They detect a narrow slice of technical problems, mostly issues with code patterns a machine can parse. They cannot evaluate context, meaning, user experience, or most of what makes content accessible to people who rely on assistive technology. A scan is a starting signal, not a measure of conformance.
| Capability | Reality |
|---|---|
| Issue detection | Approximately 25% of WCAG issues on a page |
| Conformance determination | Not possible through scanning |
| Context and meaning | Cannot be evaluated by a machine |
| False positives and negatives | Both occur and require human review |
| Best use | Ongoing monitoring between audits |
Why Scans Only Catch a Fraction of WCAG Issues
WCAG covers a wide range of success criteria that depend on human judgment. Whether alt text actually describes an image. Whether a heading reflects the content beneath it. Whether a link’s purpose is clear from its text. Whether a form error message helps a person recover. These require interpretation, not pattern matching.
Automated tools work by checking code against a fixed set of rules. If alt text exists, the scan passes. If the alt text reads “image123.jpg,” the scan still passes because something is present. A person reviewing the page would mark that as an issue immediately.
This is the core limitation. Scanners verify presence, not quality.
What Do Scans Actually Miss?
The majority of accessibility issues fall outside what automation can detect. Among the most common:
Meaningful alt text: Scans confirm alt attributes exist but cannot judge whether the description is accurate or useful.
Heading structure: A scan can verify heading tags are present but cannot tell if the hierarchy reflects the page’s actual content.
Link purpose: “Read more” links pass automated checks but fail real accessibility review.
Keyboard interaction: Complex widgets, custom dropdowns, and modal dialogs need human evaluation to confirm they work without a mouse.
Focus order: Tools can detect missing focus indicators but rarely catch illogical tab order.
Form labels and instructions: A label may exist in the code but not match the visible field.
Screen reader experience: The way content is announced, grouped, and navigated requires assistive technology review.
These are not edge cases. They are some of the most frequently cited issues in ADA website compliance lawsuits.
False Positives and False Negatives
Scanning tools produce both. A false positive flags something as an issue when it is not. A false negative misses a real issue entirely. Both require a person to sort through.
Teams that act on raw scan output without review often spend time fixing things that were never broken, while leaving the real issues in place. The output of a scan is a list of suspected items, not a verified list of issues.
Why Scans Cannot Determine Conformance
WCAG conformance means a page meets the success criteria as written. The criteria require evaluation of content, behavior, and user experience. A scan cannot make that determination because it cannot evaluate meaning or context.
A page that returns a perfect scan score can still have serious accessibility issues. A page with many flagged items in a scan can still meet WCAG when reviewed properly. The scan and the conformance picture are not the same thing.
An accessibility audit conducted by a person who evaluates each success criterion is the only way to determine WCAG conformance. Accessible.org audits are always fully manual for this reason.
The Right Use of Scanning Tools
Scans are useful when treated as monitoring, not measurement. Conduct scans regularly to catch regressions, surface obvious issues before they multiply, and keep the team aware of code-level patterns that need attention.
What scans should not do is stand in for an audit, generate a conformance claim, or feed into a VPAT or ACR. Those documents require evaluation a tool cannot produce. The VPAT and ACR process is grounded in audit findings, not scan output.
Real AI vs. Inflated Claims
Some vendors market scanning products as AI that automates WCAG conformance. That claim is inaccurate. AI can speed up parts of an auditor’s workflow, surface patterns, and help map issues to fixes. It cannot replace the human judgment that conformance requires.
Accessible.org Labs is actively researching practical applications of AI in audit and remediation workflows, where the goal is making skilled practitioners more efficient, not pretending a tool can do their work.
How accurate are accessibility scanning tools?
Scanning tools flag approximately 25% of WCAG issues on a page. The remaining issues require human evaluation. Accuracy also varies because scans produce false positives and false negatives that need review.
Can a scan replace an accessibility audit?
No. A scan cannot evaluate context, meaning, or user experience, which the majority of WCAG success criteria require. An audit conducted by a qualified person is the only way to determine WCAG conformance.
Should I still use a scanning tool?
Yes, for monitoring. Scans are valuable for catching code-level regressions between audits and giving the team ongoing visibility into common issues. Treat the output as a signal, not a verdict.
Why do scan scores not match audit results?
A scan score reflects a narrow set of automated checks. An audit evaluates the full set of WCAG success criteria against the actual content and behavior of the page. The two measure different things, so the numbers will not align.
Scans tell you part of the story. An audit tells you whether the page meets the standard.
Contact Accessible.org to request an audit quote.
