Average detected accessibility errors per home page, broken down by content management system, JavaScript framework, library, and ecommerce platform, across one million sites.
Headline finding
Nearly every popular JavaScript widget library was associated with more accessibility errors than average, not fewer. The worst are carousels, lightboxes, custom selects, and alert replacements: the components teams install so they do not have to build them.
Source: WebAIM Million 2026. Correlation only. Automated results do not establish WCAG conformance.
Executive summary
Every mainstream content management system in the sample came in at or below the million-page average of 56.1 detected errors per home page. The libraries layered on top did not. That is the finding worth acting on, because it points at a decision teams make every sprint rather than once at the start of a project.
All figures are averages of the sites measured, not properties of the technologies themselves. The interpretation section explains what that distinction rules out.
From 29.9 to 52.8 errors across every mainstream platform. Choosing a different CMS moves you less than fixing one bad theme.
Popular widget libraries ran from 55.3 to 101.6 errors, and almost all of them sat above the average.
The lowest figure in the entire study was Astro at 9 errors, a framework whose entire premise is shipping less client-side JavaScript.
WebAIM states directly that additional errors correlated with a technology cannot always be attributed to that technology. A platform's score reflects the sites built on it as much as the platform itself. Ecommerce stores are more complex than brochure sites, and the teams choosing an enterprise CMS often have accessibility budgets that a hobbyist does not. Use these numbers to ask better questions about your stack, not to declare a winner.
Most sites on a mainstream CMS scored better than the million-page average. The platform you publish on sets a floor, not a ceiling.
46.7% below average
41.2% below average
40.6% below average
37.4% below average
31.3% below average
26.5% below average
18.6% below average
5.8% below average
89.8% above average
Source: WebAIM Million 2026
The spread here is the widest of any group, from 84% below average to 48% above it.
84% below average
27.1% below average
22.5% below average
2.6% below average
7.7% above average
15.1% above average
15.9% above average
26.1% above average
36.4% above average
47.7% above average
Source: WebAIM Million 2026
Bootstrap is on more home pages than any other framework in this group and sits above the average error count.
27.1% below average
7.3% below average
3.3% below average
2.1% above average
4.1% above average
12.8% above average
Source: WebAIM Million 2026
Almost every widely used library in the sample was associated with more detected errors, not fewer. The carousel, lightbox and custom-select widgets are the worst of them.
1.5% below average
1.5% above average
2.5% above average
15.7% above average
19.8% above average
24% above average
25.7% above average
26.1% above average
26.4% above average
31% above average
31.8% above average
42.3% above average
43.3% above average
43.6% above average
46.2% above average
59% above average
60.3% above average
81% above average
Source: WebAIM Million 2026
Every ecommerce platform measured sat above the average, consistent with the Shopping category being one of the worst-performing sectors.
33.9% above average
35% above average
155.3% above average
Source: WebAIM Million 2026
Interpretation
The useful signal is not a leaderboard. It is that the pieces you add to a page carry more risk than the platform you start from.
Every mainstream CMS in the sample came in at or below the million-page average, and the spread between the best and the worst is smaller than the spread inside any one of them. Picking a different CMS will not fix a theme, a page builder or a plugin that emits inaccessible markup.
Almost every popular library measured was associated with more errors than average. The pattern is telling: the worst are carousels, lightboxes, custom selects and alert replacements, the exact components teams install so they do not have to build them. Reach for a native element first, then a headless library that documents its keyboard and ARIA behaviour.
Every ecommerce platform measured sat above the average, and Shopping was one of the worst-performing sectors overall. Online stores also attract the most accessibility litigation, so this is the stack where the gap between measured errors and legal exposure matters most.
If the library findings match your stack, the component guides are the place to start: accessible carousels, dialogs and modals for lightbox and alert replacements, and comboboxes or listboxes for custom selects. For framework-level patterns, see the React, Vue and Angular guides, and for storefronts the ecommerce accessibility guide.
Automated WAVE evaluation of the home pages for the top one million websites in February 2026. Technology and category detection was performed alongside the accessibility scan.
Sample Size
1,000,000 home pages, of which 2,173,756 technology detections are represented across the groups above
Date Range
February 2026 sample, published 2026-03-30
Last updated: 2026-08-27
Use the research
Reuse the findings with a clear citation, and keep the link to WebAIM when publishing extracts.
The PDF contains every group as a table with the methodology note. The CSV contains one row per technology for your own analysis.
Accessibility.build. (2026). Web Accessibility by Technology Stack. https://accessibility.build/research/accessibility-by-technology
Questions about the data
What these averages support, and what they do not.
In the February 2026 WebAIM Million sample, home pages built with Adobe Experience Manager averaged the fewest detected errors at 29.9, followed by Squarespace at 33.0 and Wix at 33.3. Drupal averaged 41.2 and WordPress 52.8, against a million-page average of 56.1. Every mainstream CMS measured came in at or below that average. The important caveat is that these are averages of the sites built on each platform, not a measure of what the platform forces you to produce, and WebAIM states that errors correlated with a technology cannot always be attributed to that technology.
WordPress home pages averaged 52.8 detected errors in February 2026, which is 5.8% better than the million-page average of 56.1. It was the largest sample in the study by a wide margin at 252,302 home pages, so that figure is stable, but it also means the average blends everything from carefully built agency sites to abandoned theme installs. WordPress itself can be used to build a fully conformant site. What drives the error count in practice is the theme, the page builder, and the plugins, not the core CMS.
Astro home pages averaged 9.0 detected errors, 84% below the million-page average and the lowest figure of any technology in the study. Next.js followed at 40.9 and React at 43.5. Vue.js averaged 64.6 and AngularJS 76.6, both above average. Astro's result is consistent with what the framework does, which is ship mostly static HTML with little client-side JavaScript, but its sample of 5,472 home pages is small compared with React or Vue and its users skew toward developer-focused sites.
In this sample, nearly every popular library was associated with a higher error count than average. SweetAlert2 home pages averaged 101.6 errors, FancyBox 90.0, Select2 82.0, jQuery UI 79.9, and Swiper 74.0, against an average of 56.1. The pattern is that the worst performers are carousels, lightboxes, custom select replacements, and alert replacements, which are precisely the components teams install to avoid building them. This is an association rather than proof of cause: pages carrying many libraries are also more complex overall.
Shopify home pages averaged 75.1 detected errors in February 2026, 33.9% above the million-page average. Magento averaged 75.8 and PrestaShop 143.2. Every ecommerce platform measured sat above the average, which matches the finding that Shopping was one of the worst-performing site categories overall. Online retail also attracts the largest share of digital accessibility litigation, so the gap between a storefront's measured errors and its legal exposure is worth taking seriously.
No. Automated testing detects only a subset of WCAG requirements, so a low average error count cannot establish conformance for any technology or for any individual site built with it. The averages also describe home pages only, which means checkout flows, dashboards, forms, and authenticated pages were never evaluated. Treat platform choice as one input among many, and verify your own build with manual keyboard testing, screen reader testing, and a full WCAG 2.2 audit.
Every number on this page is available as CSV and JSON. The underlying figures are © WebAIM and are reproduced under WebAIM's terms: free reuse with full credit to WebAIM, a link to webaim.org, and the © WebAIM notice intact. Cite Accessibility.build for the analysis. Last updated .
Comprehensive tools, checklists, and guides to help you create inclusive digital experiences
Accessibility assistant
Hi, I am Curb. Ask me about accessibility, WCAG, or anything on Accessibility.build. I can also explain the page you are on.
Try asking