Web accessibility statistics

WebAIM tests the home pages of the top one million websites every year. In the February 2026 edition:

• 95.9% of home pages had detectable WCAG 2 failures. That figure has risen four years running (it was 94.8% in 2025).
• The average home page had 56.1 distinct errors, up 10% on the previous year.
• Across a million pages, automated testing found more than 56 million separate errors.
• Just six error types account for 96% of everything detected, so most of the problem is concentrated and fixable.

Share of home pages affected by each issue (WebAIM Million, 2026):

• Low-contrast text: 83.9%. The most common failure by a wide margin. How to fix it.
• Missing image alt text: 53.1%. How to fix it.
• Missing form input labels: 51.0%. How to fix it.
• Empty links: 46.3%. How to fix it.
• Empty buttons: 30.6%. How to fix it.
• Missing document language: 13.5%.

None of these are hard to fix once you know where they are.

ARIA is markup meant to improve accessibility. WebAIM found the opposite in practice. Home pages using ARIA averaged 59.1 errors, against 42 errors on pages with no ARIA at all. The more ARIA a page used, the more errors it tended to have, because ARIA is so often applied incorrectly. Plain, correct HTML beats accessibility features bolted on without care.

The HTTP Archive's Web Almanac measures a much larger crawl of the web and looks at different things from WebAIM, several of which no error count captures. From the 2025 accessibility chapter, published January 2026:

• 67% of sites explicitly remove the browser's default focus outline, up from the year before. Only about a quarter use :focus-visible to put a better one back. Removing the outline with nothing in its place leaves keyboard users unable to see where they are. See keyboard navigation and contrast that breaks on focus.
• Only 30% of sites pass Lighthouse's colour contrast audit, on both mobile and desktop.
• 19% of mobile sites disable zooming or scaling, which stops people enlarging text. See font size and accessibility.
• 24% of form inputs on mobile have no accessible name at all.
• About a quarter of pages include a skip link, and about 86% declare a valid page language.

WebAIM's list stops at WCAG 2.1-era checks. WCAG 2.2 added a minimum size for things you tap or click: 24 by 24 CSS pixels (success criterion 2.5.8). We scanned 342 homepages of major US stores, hospitals, federal agencies and banks in June and July 2026 with the same engine:

• 71% had at least one serious WCAG failure on the homepage, and only 10% passed every automated check.
• Colour contrast was the most common failure, on 30% of sites.
• Undersized touch targets came second overall, on 20% of sites. Where they come from, and the exceptions that let a small control pass.

The sector-by-sector breakdown and the method are in web accessibility by industry.

Overlays are scripts that add an accessibility widget to a page instead of changing its code. The Web Almanac 2025 found one on 2.1% of desktop sites and 1.8% of mobile sites, but on only 0.2% of the top 1,000. In UsableNet's 2026 midyear figures, about 20% of the companies sued over digital accessibility had a widget or overlay installed. The overlay guide covers why they do not prevent claims.

UsableNet tracks US digital accessibility litigation. Its 2026 midyear report projects:

• About 6,000 lawsuits over websites and mobile apps in 2026, a rise of roughly 20 to 25% on 2025.
• About 80% involve e-commerce companies.
• The 10 most active plaintiff firms file about 84% of cases.
• About a quarter are against companies that had already faced a similar claim.
• New York leads in volume, followed by California, Florida and Illinois, with Illinois among the fastest growing.

UsableNet estimates businesses receive two to three times as many demand letters as there are lawsuits, though there is no public count of letters. For how the rules differ, see ADA website compliance and California's Unruh Act.

Accessibility failures are the rule. They also cluster into a small set of recurring problems, most of which an automated test catches in seconds. The rest need a person to judge, such as whether text still reads when it is resized to 200%. Tools that promise instant compliance with a single line of JavaScript do not fix any of the errors above, which is why the only reliable approach is still to scan, fix the code, and re-test.

• WebAIM Million, February 2026: an automated analysis of the home pages of the top one million websites. Published each year.
• HTTP Archive Web Almanac 2025, accessibility chapter, published January 2026. Built largely on Lighthouse audits run across the HTTP Archive crawl.
• UsableNet, figures from its 2026 midyear lawsuit report as published on its blog in July 2026. UsableNet sells accessibility services.
• Our own reports: 342 US homepages across four sectors, scanned June and July 2026 with axe-core against WCAG 2.x A and AA plus our gradient contrast resolution. Aggregate figures only.

Every automated figure here covers only the machine-checkable part of WCAG, so real failure rates, including issues that need a person to judge, are higher.