Every year, WebAIM runs automated accessibility tests against the home pages of the top one million websites and publishes what it finds. The 2026 edition isn’t encouraging: 95.9% of home pages had at least one detectable WCAG 2 failure, up from 94.8% the year before, according to the WebAIM Million 2026 report. That’s not a niche subset of poorly built sites. It’s the overwhelming majority of the modern web, including sites run by large, well-resourced companies with dedicated engineering teams.

What the study actually measures

The WebAIM Million pulls its list of home pages from a ranked set of popular domains and runs each one through automated testing. Automated tools only catch a portion of real accessibility barriers by design. They can flag a missing alt attribute, but they can’t judge whether the alt text someone wrote is actually useful, and they can’t tell you whether a screen reader user can complete a checkout flow. That makes the study’s numbers a floor, not a ceiling, on how much work the web genuinely has left to do.

The headline numbers

Three figures from the 2026 report are worth sitting with:

  • 95.9% of home pages had at least one detectable WCAG 2 failure, per WebAIM Million 2026.
  • The average home page carried 56.1 detectable errors, a 10.1% increase over 2025, per the same report.
  • 83.9% of home pages had low contrast text, making it the single most common failure type found.

The errors-per-page trend is the one that should concern site owners most. A failure rate near 96% has hovered in that range for years and mostly reflects how widespread basic gaps are across the web generally. But the average number of errors per page climbing, not falling, suggests the sites that do fail aren’t just failing in one small way. They’re accumulating more problems as pages get more complex, not fewer.

Why low contrast keeps winning

Low contrast text has topped this list for years, and it’s a useful case study in why accessibility problems persist even when they’re easy to describe. Contrast ratio is a purely mechanical calculation between a text color and its background, checkable with free tools in seconds, yet it remains the most common failure on the web by a wide margin. The likely explanation isn’t ignorance of the rule. It’s that contrast gets decided in a design tool by someone optimizing for visual polish, then shipped without anyone re-checking it against WCAG’s numeric thresholds once real content and real backgrounds are in place.

What this means if you run a website

If your home page hasn’t been tested, the base rate says it’s more likely than not to have a detectable WCAG failure, and specifically worth checking your text and background color combinations first, since that’s the single most common issue by a wide margin. Automated scanning tools, including the free WAVE tool WebAIM uses for this study, catch a meaningful share of issues quickly and cheaply, which makes them a reasonable starting point. They won’t catch everything: screen reader usability, logical reading order, and whether interactive components actually work with a keyboard all generally require someone testing by hand or with assistive technology directly.

That gap between what automated tools can and can’t see is why some platforms in this space pair automated scanning with a manual review layer instead of shipping detection alone. Wawsome, for instance, combines an automated widget with continuous monitoring and a human reviewed checker on the theory that a scan by itself isn’t a complete answer. Deque’s axe-core testing tools take a different but related approach, aimed at catching issues during development before a page goes live, which is a genuinely different strategy from testing a finished page after launch.

The trend line matters more than any single year

A single year’s numbers are a snapshot. The more useful signal from the WebAIM Million is the multi-year pattern: failure rates that stay stubbornly high and error counts that keep climbing, even as awareness of web accessibility has grown and more regulation, the European Accessibility Act among it, has come into force. That combination suggests the problem isn’t primarily a lack of information anymore. It’s that accessibility testing still isn’t built into most teams’ normal workflow the way a spell check or a broken link checker is, so problems compound as sites get redesigned and rebuilt rather than getting resolved along the way.

A reasonable starting checklist

None of this data is actionable without a starting point, and the study itself suggests one. Given that low contrast text is the single most common failure by a wide margin, checking body text, buttons, form labels, and placeholder text against their backgrounds is the highest-yield first step for most sites. After that, missing alt text on meaningful images and unlabeled form fields tend to be the next most common and most fixable categories, since both are quick for a developer or content editor to address once flagged. None of that requires a full redesign. It requires someone actually running a check, which the failure rate suggests still isn’t happening on the large majority of sites tested.

Reading the report critically

It’s worth treating the WebAIM Million as a directional signal rather than a precise verdict on any individual site. The methodology tests home pages specifically, not full sites, and it relies on automated detection, which means a page scoring “clean” on this particular test could still have real usability problems for a screen reader or keyboard-only user that automated tooling simply can’t detect. The report’s real value is in the pattern it reveals across a million sites at once: accessibility failures are common, they’re concentrated in a predictable handful of categories, and they’re not improving on their own without deliberate effort.