Automated vs Manual Accessibility Testing

Automated accessibility testing uses software to scan your website’s code for detectable WCAG 2.1 AA failures in seconds, while manual testing has a trained person use a keyboard, a screen reader, and human judgment. Automation catches roughly 30-40% of issues; you need both for real ADA compliance.

The short version

Automated tools and manual testing are not competitors — they’re two halves of the same job. Scanners are fast, cheap, and consistent, but they can only check things a machine can measure. Manual testing is slower and requires expertise, but it’s the only way to confirm a site is genuinely usable by people who rely on assistive technology. A credible accessibility audit uses automation as a starting point and manual testing as the real work.

What automated accessibility testing does well

Automated accessibility testing tools parse your page’s HTML, CSS, and rendered DOM, then flag patterns that violate WCAG success criteria. They run in seconds and scale across hundreds of pages.

The most common tools include:

• axe (axe-core) — the engine behind many scanners and browser extensions; known for a low false-positive rate.
• WAVE — WebAIM’s visual checker that overlays icons on your page so you can see issues in context.
• Lighthouse — built into Chrome DevTools; gives an accessibility score alongside performance and SEO.

These tools reliably catch machine-detectable problems such as:

• Missing alt attributes on images (though not whether the alt text is meaningful)
• Insufficient color contrast ratios
• Missing form labels and empty buttons
• Missing lang attributes and document language
• Some ARIA misuse, like invalid roles or required attributes
• Broken heading structure, such as skipped heading levels

If a scanner reports these, fix them. They’re real failures. But a clean scan is the floor, not the ceiling.

The ~30-40% coverage limit

Here’s the number every business owner should know: automated tools catch only about 30 to 40 percent of WCAG success criteria. This isn’t a knock on the tools — it’s a hard limit of what software can detect. The other 60-70% requires a human.

Why? Because most accessibility is about meaning and experience, which a machine can’t evaluate:

• A scanner sees that an image has alt="image123". It cannot tell you the alt text is useless.
• A scanner sees a heading exists. It cannot tell you the headings describe the content in a logical order.
• A scanner cannot tab through your site and discover that a modal traps keyboard focus.
• A scanner cannot listen to a screen reader announce a checkout flow and notice it makes no sense.

This is also exactly why overlays fail. An overlay is just automation wearing a costume — it patches the same machine-detectable layer a scanner finds, and nothing more. It cannot fix the 60-70% that requires judgment, and it frequently breaks the experience for real screen-reader users.

What manual testing catches that automation can’t

Manual accessibility testing is where conformance is actually proven. A trained tester works through your site the way a person with a disability would, checking the criteria that demand human judgment.

Manual testing covers things like:

• Keyboard navigation — can you reach and operate every control with Tab, Enter, and arrow keys alone, with no traps and a visible focus indicator?
• Screen reader testing — does the page make sense aloud in NVDA, JAWS, or VoiceOver? Are buttons and links announced with their purpose?
• Meaningful alt text — does each image’s description convey what matters in context?
• Logical reading and focus order — does content flow in a sensible sequence?
• Error handling — when a form is submitted wrong, is the error announced and easy to find?
• Cognitive load and clarity — are instructions clear, and is the interface understandable?

These map to the POUR principles — Perceivable, Operable, Understandable, Robust — that underpin WCAG. Automation can only verify slivers of each. Manual testing covers the whole picture, and it’s what the DOJ and courts effectively expect when they reference WCAG 2.1 AA under ADA Title III.

Side-by-side comparison

When to use each

Both methods have a place in a healthy accessibility program. Use them at the right moments:

1. Start with an automated scan to find the low-hanging fruit fast. Run a free scan or use axe/WAVE/Lighthouse yourself.
2. Add automated checks to your build pipeline so new code doesn’t regress — this is the heart of ongoing accessibility monitoring.
3. Commission manual testing for anything that matters — a real audit, a VPAT or Accessibility Conformance Report, or sign-off before launch.
4. Re-test after remediation to confirm fixes actually work in practice, not just in the scanner.

For the full step-by-step process, see our guide on how to do an accessibility audit.

Why this matters for ADA compliance

Thousands of ADA web accessibility lawsuits and demand letters are filed in the US each year, and a common thread is over-reliance on automation. A business runs a scanner, sees a green score or installs an overlay, and assumes it’s covered — then receives a demand letter anyway, because the real barriers were never machine-detectable in the first place.

WCAG conformance is measured against levels A, AA, and AAA. ADA Title III, Section 508 of the Rehabilitation Act, and DOJ guidance all point toward WCAG 2.1 AA as the practical target — and AA conformance simply cannot be confirmed by software alone. This guide is educational, not legal advice; if you’ve received a legal threat, consult a qualified attorney.

The reassuring part: every issue automation misses is fixable by hand. That’s what we do. Curbcut runs the scan, performs the manual testing scanners can’t, and then remediates the code manually — no overlays, no shortcuts.

Authoritative resources

• W3C Web Accessibility Initiative (WAI) — the source of WCAG and evaluation guidance
• ADA.gov — the Department of Justice’s official ADA resource
• WebAIM — the team behind the WAVE tool, with deep testing guidance
• Section508.gov — federal accessibility testing and conformance resources

The bottom line

Automated testing is essential but incomplete — it’s the fast first pass that handles 30-40% of the work. Manual testing is where real, legally defensible compliance is earned. Don’t choose between them; pair them. The fastest way to see where you stand is to get a real accessibility audit that combines both, or start with a free scan to see what a machine can already find.