Accessibility testing continues to find widespread, repeatable barriers across websites and mobile apps. The February 2026 WebAIM Million report found 56,114,377 distinct detectable errors across one million home pages, while UK public-sector monitoring recorded 29,787 issues in website and mobile-app tests from 1 January 2022 to 1 September 2024.
Contents
- Large-scale accessibility test results
- The most common automated WCAG failures
- Content structure and interaction checks
- Results by category, domain, and technology
- Public-sector monitoring outcomes
- Detailed website and mobile-app findings
Large-scale accessibility test results
The 2026 WebAIM scan examined one million home pages in February 2026. Its automated test engine found an average of 56.1 detectable accessibility errors per home page. WCAG failures detectable by that engine appeared on 95.9% of the tested home pages. Detected errors increased 10.1% from the 2025 WebAIM analysis. These figures describe failures detectable by the test engine; they are not a complete measure of every accessibility barrier or a substitute for manual testing.
The tested pages contained more than 1.4 billion page elements in total. They averaged 1,437 elements per home page, an increase of 14.3% in one year. Detected accessibility errors affected 3.9% of all tested page elements. The scale matters for test planning: even a modest error rate can represent millions of individual findings when pages contain large amounts of content and interface markup.
Page complexity also differed by popularity. The most popular 100,000 tested pages averaged 1,584 elements, compared with 1,318 among the least popular 100,000. This comparison is descriptive rather than causal: it shows different levels of page complexity in the scan, not that popularity itself creates accessibility errors.
The most common automated WCAG failures
Low-contrast text was the most widespread automated finding. Text below WCAG 2 AA contrast thresholds appeared on 83.9% of tested home pages, with an average of 34 distinct low-contrast-text instances per page. This makes color and contrast testing one of the highest-yield early checks in an accessibility test plan.
Image alternatives were another major source of findings. Missing alternative text appeared on 53.1% of home pages. Across 66.6 million images, 16.2% lacked alternative text, equivalent to an average of 10.8 images per home page. Linked images accounted for 45% of images missing alternative text. Among images that did have alt text, 10.8% used questionable or repetitive alternative text.
Forms and controls produced similarly common failures:
| Automated finding | Share of tested home pages | Measurement period |
|---|---|---|
| Low-contrast text below WCAG 2 AA thresholds | 83.9% | February 2026 |
| Missing alternative text for images | 53.1% | February 2026 |
| Missing form-input labels | 51.0% | February 2026 |
| Empty links | 46.3% | February 2026 |
| Empty buttons | 30.6% | February 2026 |
| Missing document language | 13.5% | February 2026 |
Missing form-input labels affected 51% of tested home pages. Empty links appeared on 46.3%, while empty buttons appeared on 30.6%. A missing document language affected 13.5%. Each percentage measures the share of home pages with at least one detected instance, so it should not be read as the percentage of all individual controls that failed.
Content structure and interaction checks
The WebAIM results show why accessibility testing needs more than a color-contrast scan. Tested home pages contained an average of 6.9 form inputs, and that average increased 36% over three years, from 2023 to 2026. One third of those inputs, or 33.1%, were not properly labeled in the February 2026 scan.
Headings were common but often structurally inconsistent. The scan detected nearly 30 million headings, averaging 29.9 headings per home page. Among pages with at least one h6, 94.9% also had skipped heading levels. Multiple h1 headings appeared on 18.1% of tested home pages. In total, the scan found 1,206,444 instances of skipped heading levels, and skipped heading levels appeared on 41.8% of home pages. No headings were present on 7.5%.
Landmarks were present on most pages, but coverage varied by landmark type. At least one region or ARIA landmark appeared on 84.3% of tested home pages. A main element or main landmark appeared on 46.1%, and a search landmark appeared on 18.8%.
ARIA usage was extensive. The scan detected 133,589,803 ARIA attributes, averaging more than 133 per home page. Home pages averaged 31.4 aria-label, aria-labelledby, or aria-describedby attributes, and ARIA excluding landmark roles was used on 82.7% of tested home pages.
However, the presence of ARIA did not correspond to fewer detected errors in this dataset. Pages with ARIA averaged 59.1 detected errors, compared with 42 errors on pages without ARIA. This is an association in the scan, not evidence that ARIA causes errors; more complex pages may be more likely both to use ARIA and to contain detectable issues.
Menus and keyboard-related markup also deserve focused checks. An ARIA menu appeared on 5.7% of tested home pages, and 22% of those menus introduced barriers through missing required markup or interactions. Tested home pages averaged 30.4 tabindex=0 or tabindex=-1 instances.
Results by category, domain, and technology
The WebAIM Million report found substantial differences in average detected errors across site categories, domain endings, and content-management or development technologies. The results below are averages from the February 2026 home-page scan, not compliance ratings for every site in each group.
| Group | Average detected errors per home page |
|---|---|
| Government category | 42.4 |
| Education category | 48.9 |
| Shopping category | 71.0 |
.gov domain | 18.5 |
.org domain | 44.2 |
.ua domain | 103.1 |
| Adobe Experience Manager | 29.9 |
| WordPress | 52.8 |
| 1C-Bitrix | 106.5 |
| Astro | 9.0 |
| AngularJS | 76.6 |
Shopping-category home pages averaged 71.0 detected errors, compared with 42.4 for government-category pages and 48.9 for education-category pages. By domain ending, .gov home pages averaged 18.5 errors, .org pages averaged 44.2, and .ua pages averaged 103.1.
Technology averages were also different. Adobe Experience Manager home pages averaged 29.9 detected errors, WordPress pages 52.8, 1C-Bitrix pages 106.5, Astro pages 9.0, and AngularJS pages 76.6. These figures are useful for prioritizing investigation, but they do not establish that a platform alone determines accessibility. Templates, components, content, configuration, and development practices can all affect a result.
Public-sector monitoring outcomes
The UK government’s accessibility monitoring report for public-sector websites and mobile apps from 2022 to 2024 covers monitoring from 1 January 2022 to 1 September 2024. Across website and mobile-app tests, the process recorded 29,787 issues.
Public-sector organisations fixed 55.3% of the issues found during this monitoring period. The process recorded 3,693 issues that remained unfixed at retest. In 67.9% of completed monitoring cases, organisations fixed all detected issues or had a short-term plan. These measures describe outcomes within the UK monitoring process and should not be generalized to every public-sector service or every accessibility issue.
The monitoring process also tracked accessibility statements. An accessibility statement was present at initial monitoring for 85% of the UK websites and apps tested. After monitoring, 97% had published an accessibility statement, and 65% included all information required by the model accessibility statement. Seventy percent of completed simplified website cases received a recommendation of no further action, up from 59% in the previous UK monitoring period from 2020 to 2021.
Detailed website and mobile-app findings
Detailed testing showed that initial compliance was not achieved by any of the 52 UK websites undergoing detailed testing during the 1 January 2022–1 September 2024 monitoring period. Those tests found between 14 and 116 issues per test, with an average of 53. Across the detailed website tests, 2,736 issues were found and 1,857, or 68%, were resolved.
The mobile-app findings followed the same pattern. None of the 21 UK mobile apps tested was fully compliant at its initial test. The tests found between 13 and 76 issues per app, averaging 42 issues per test. In total, UK mobile-app testing detected 880 issues, of which 743, or 84%, were resolved.
The website and app results illustrate why retesting is an essential part of accessibility testing. Initial testing identifies barriers, while a later test can measure whether fixes were implemented and whether the service moved toward compliance. The reported resolution percentages also cover the issues found in those specific detailed tests; they do not indicate that all accessibility barriers in the products were eliminated.