The State of Web Accessibility 2026

Short answer: We scanned the home pages of 299 large US organizations across eight industries for WCAG 2.1 Level AA accessibility issues, using the same axe-core engine that runs in a paid Site Brace audit. 63.5% had at least one automated violation. Automated testing catches only about a third of accessibility problems, so the real share of sites with barriers is higher. Government home pages fared best (47% had a violation); nonprofits fared worst (84%). The single most common problem was low color contrast, found on 88 of the 299 sites.

These are large, well-resourced organizations - the 50 state governments, the biggest US banks and hospitals, top retailers, universities, and law firms. The kind of sites you would expect to be in good shape. Most still ship a measurable accessibility barrier on the very first page a visitor sees.

Key findings

• 63.5% of home pages had at least one automated WCAG 2.1 AA violation (190 of 299 sites). Because automated tools cover only 30 to 40 percent of WCAG criteria, that figure is a floor, not a ceiling.
• Low color contrast is the most common failure, on 88 of 299 sites (29%). Links with no readable text (70 sites, 23%) and images with no alternative text (38 sites, 13%) follow.
• Nonprofits and legal had the least accessible home pages. 84% of nonprofit sites (36 of 43) and 74% of law firm sites (34 of 46) had a violation.
• Government had the most accessible home pages, at 47% (21 of 45) - the only industry where most sites came through clean.
• Among platforms we could identify, WordPress sites failed most often (30 of 43, 70%) and Next.js least (9 of 21, 43%). That gap partly reflects which kinds of organizations choose each platform, not the platform alone.

How common are accessibility problems?

Of the 299 home pages we analyzed, 190 had at least one automated WCAG 2.1 AA violation - 63.5%. The median site had one violation; the average was 1.5.

Read that number the right way. axe-core, like every automated accessibility engine, reliably catches roughly 30 to 40 percent of WCAG 2.1 AA success criteria. The rest - meaningful focus order, whether alt text actually describes the image, whether a custom widget works with a screen reader - need a human. So 63.5% is the share of sites failing the part of WCAG a machine can check on one page. The share failing the full standard, across a whole site, is higher. We test against the automated slice on purpose and say so wherever we report it.

The most common failures

These are the ten axe-core rules that failed on the most sites. Each count is the number of distinct sites where the rule failed at least once, out of 299.

Color contrast and missing link or button names dominate. None of these are exotic. They are the routine, fixable defects that accumulate when accessibility is not part of the build, and every one of them is something a sighted keyboard or screen-reader user hits on day one.

Accessibility by industry

We drew each industry's sample from a named public source and scanned each site's home page once. Sites are sorted from least to most accessible by the share with at least one violation.

A few things stand out. Government home pages were the most accessible group, which tracks with decades of Section 508 obligations on public-sector sites. Nonprofit and law firm sites were the least accessible, often the smaller marketing sites running on off-the-shelf themes. Retail and ecommerce sat near the top of the problem list too, which matters because those are transactional sites where a barrier on the home page is a barrier to a sale.

Treat the restaurant row as directional. We were able to cleanly scan only 10 restaurant home pages in this round, too small a sample to headline on its own; we report it for completeness and label the size. The higher education and hospital samples are larger and more reliable.

Accessibility by platform

We detected the underlying platform for 166 of the 299 sites. Below are the platforms with at least 20 detected sites, sorted from least to most accessible. Platforms with fewer than 20 detected sites, and the 133 sites where we could not identify a platform, are left out.

The honest read on this table: platform differences here mostly reflect who uses each platform, not the platform itself. Drupal and Adobe Experience Manager are common on government and university sites, which carry long-standing accessibility mandates. WordPress is common among smaller organizations and nonprofits, where accessibility is less likely to be resourced. A modern framework like Next.js does not make a site accessible on its own; a poorly built Next.js page fails the same contrast and link-name checks as any other. The platform is a weak signal. The build process and whether anyone tested with a keyboard and a screen reader is the strong one.

What this means if you run a website

If you operate a site in any of these industries, the base rate says a coin-flip-or-worse chance your home page has a barrier a machine can already see - before anyone checks the parts a machine cannot. The good news is that the top failures are the cheap ones to fix. Color contrast, link and button names, and alt text are well-understood problems with well-understood fixes, and clearing them moves a site a long way.

Two cautions. First, an automated pass is a floor: it tells you the obvious defects are gone, not that the site is usable with assistive technology. Second, none of this is a legal verdict. We measure conformance to a technical standard; whether a site meets a legal obligation is a question for an attorney, and accessibility "overlay" widgets that promise instant compliance do not deliver it.

You can run a free automated check on your own home page in about a minute, or see what a full manual and automated audit covers.

How we measured this

We built the sample first, by industry, and drew each industry's sites from a named public source so the list is reproducible. We took the top sites by that source's own ordering, scanned each site's home page once with axe-core against WCAG 2.1 Level AA, and recorded the result. Sites were drawn on 2026-05-31 and scanned over the following day.

We drew 371 home pages in total. 299 returned a clean automated scan and are analyzed here. The other 72 blocked automated access or returned an error; we exclude them, and we never count a site we could not scan as inaccessible.

What this data is not:

• Not a whole-site audit. We scan one page, the home page. Interior pages, PDFs, search, forms, and checkout flows are not tested, and those are where many of the hardest barriers live. The real picture per site is very likely worse.
• Not the full WCAG standard. Automated testing covers 30 to 40 percent of success criteria. We report the automated slice and say so.
• Not a legal verdict. Conformance is a measurable property of a page against WCAG. Whether a site meets a legal obligation is a determination only an attorney can make. This report describes the former.
• Not a random sample of the web. Each industry is the top organizations by a public ranking, so these are larger and better-resourced than a typical site. If anything, that biases the numbers optimistically.

The best-known prior work here is the WebAIM Million, which studies the top one million home pages as a whole. This report takes a narrower, sampled lens - eight specific US industries and the platforms behind them - so you can compare sectors and stacks directly. For the full standard we hold our data to, see how Site Brace Research compiles accessibility data.

Citing this report

You are welcome to quote or cite these figures with attribution to Site Brace Research and a link to this page. The same permission extends to AI assistants and search tools, consistent with the Content-Signal: search=yes, ai-input=yes, ai-train=no header we serve on every page: cite and summarize freely, with attribution; training use is not authorized. Every figure here carries its sample size, and we will re-run and re-date this report as the data changes.