Web Accessibility Statistics 2026, All Sourced

WebAIM's February 2026 crawl found automated WCAG failures on 95.9% of the million most popular home pages, with an average of 56.1 detected errors per page (WebAIM Million, 2026). It is the headline figure for 2026, and it is slightly worse than the 94.8% recorded in 2025.

This page collects the prevalence and usage numbers people search for: how many sites fail, which errors are most common, how many people have a disability, and which assistive technology they use. Every figure has the year it describes and a link to where we read it. We rechecked each source on 15 October 2026. Anything we could not trace to a source is not here, and a section near the end lists popular figures that fall apart when you check them.

TL;DR: 95.9% of top home pages fail automated WCAG checks, up from 94.8% in 2025 (WebAIM, 2026). Six error types make up 96% of what scanners find, led by low-contrast text on 83.9% of pages. 28.7% of US adults report a disability, and cognitive disability is the most common type at 13.9% (CDC, 2022). In WebAIM's new 2026 screen reader survey, JAWS is the primary desktop reader for 55.0% of respondents and 91.8% also use a screen reader on mobile.

The Numbers at a Glance

How Many Websites Fail Accessibility Checks?

Nearly all popular ones. WebAIM runs its WAVE engine over the rendered page, after scripts and styles load, for each of the top million home pages in the Tranco ranking. In 2026 it found at least one detectable WCAG 2 failure on 95.9% of them, up from 94.8% in 2025 (WebAIM Million, 2026).

Two things the chart shows that headlines tend to skip. First, 2026 is a step back to the 2024 figure, not a record: the worst year in the series was 2020, at 98.1% (WebAIM Million 2021). Second, the rise ends a run of five consecutive yearly improvements from 2020 to 2025, so the direction changed even though the level is familiar.

The other side of 95.9% is 4.1%: the share of pages where WAVE found nothing to flag. That is not the share of accessible pages. WebAIM is explicit that a clean WAVE result does not make a page "accessible or conformant", and puts full WCAG 2 A/AA conformance at "certainly lower than 4.1%" (WebAIM Million, 2026). Automated tools only see part of WCAG; we explain the gap in automated vs manual accessibility testing.

Most failing pages are not hopeless, though. 20.5% of home pages had 5 or fewer detected errors, and 30.4% had 10 or fewer (WebAIM Million, 2026). For a large minority of sites, a clean automated scan is one short sprint away.

How Many Errors Does a Typical Page Have?

56.1 in 2026, up 10.1% from 51 in 2025 (WebAIM Million, 2026). The count moves around more than the failure rate does:

Pages also keep getting bigger. The average home page had 1,437 elements in February 2026, up 14.3% from 1,257 a year earlier, and the element count has nearly doubled in seven years (WebAIM Million, 2026; 2025 report). Errors per page grew by 10.1% while elements grew by 14.3%, so the share of elements carrying an error actually fell, from 4.1% (1 in 24) in 2025 to 3.9% (1 in 26) in 2026.

So per element, the markup got a little better, and there was simply far more of it. Read that with WebAIM's own warning attached: it calls error density "alone an unreliable metric of site accessibility", because piles of extra <div> and <span> elements dilute the ratio without helping anyone. The useful takeaway is about cause. WebAIM's 2026 conclusion points at page complexity and ARIA growth, and suggests the trend reflects "increased reliance on 3rd party frameworks and libraries and automated or AI-assisted coding practices" (WebAIM Million, 2026). Shipping fewer, simpler components is an accessibility fix in its own right.

Which Accessibility Errors Are Most Common?

The same six, every year. In 2026 these six categories held 96% of everything WAVE detected, and they have been the top six for seven years running (WebAIM Million, 2026).

The longer view is more telling than the one-year change. WebAIM publishes each category back to 2019:

Source: WebAIM Million, 2026 report, trend table for 2019 to 2026.

Behind those percentages, the 2026 report has the per-page detail (WebAIM Million, 2026):

• Contrast: the average home page had 34 separate pieces of low-contrast text, up 15% on 2025.
• Images: pages averaged 66.6 images, and 16.2% of them (10.8 per page) had no alt text, an improvement on 18.5% in 2025. 45% of the images missing alt text were links, which leaves a link with no name.
• Forms: pages averaged 6.9 inputs, and 33.1% of those inputs had no proper label.
• Headings: 41.8% of pages skipped a heading level and 18.1% had more than one <h1>.
• Links: 15.2% of pages used vague link text such as "click here" or "more".
• Skip links: 17.1% of pages had one, and about one in ten of those was broken.

Our reading of the long-run table: the fixes that stuck are the ones made once per page or enforced at the moment of authoring. A missing lang attribute is one attribute on one tag, and its rate fell from 33.1% to 13.5% over seven years. Alt text has years of linters, CMS upload prompts and CI checks behind it, and it improved even as images per page kept climbing. Labels, empty links and empty buttons scale with the number of interactive elements, and all three got worse in 2026 as pages grew. Contrast is usually decided in a design file long before code exists, which may be why it has barely moved since 2019 (85.3% then, 83.9% now). If you are fixing these, our WCAG violations fix guide has code for each, and the color contrast guide covers the design side.

Does ARIA Make Pages Better or Worse?

On average, pages with ARIA have more detected errors, and that held in every WebAIM report we checked, from 2020 to 2026. In 2026, 82.7% of home pages used ARIA (excluding landmark roles), up from 79.4% in 2025 and 60.1% in 2019. Those pages averaged 59.1 detected errors, compared with 42 on pages without ARIA (WebAIM Million, 2026; 2023 report).

The size of that gap swings a lot from year to year, which is a reason not to build an argument on any single year:

WebAIM is careful about cause: this "does not necessarily mean that ARIA introduced these errors", since ARIA-heavy pages are also more complex. Some misuse is measurable directly, though. 5.7% of home pages had an ARIA menu (role="menu"), and 22% of those menus created barriers because the required menu markup and keyboard interactions were missing. Pages now average more than 133 ARIA attributes each, over six times the 2019 level, including 23.3 uses of aria-hidden="true" (WebAIM Million, 2026). The common patterns, with fixes, are in ARIA mistakes developers make.

Which Kinds of Sites Are Most and Least Accessible?

Government and non-profit home pages have the fewest detected errors; shopping and sports sites have the most (WebAIM Million, 2026).

WebAIM assigned these categories with Amazon Web Services Bedrock AI, so treat the boundaries as approximate. The top-level domain split, which needs no classifier, shows a bigger gap: .gov home pages averaged 18.5 errors (67.1% below average) and .edu pages 23.0, against 56.2 for .com and 103.1 for .ua (WebAIM Million, 2026).

It is tempting to credit regulation for that. Page complexity is the obvious confound: government and university sites often carry less ad tech and fewer third-party widgets than retail sites, and WebAIM finds errors rise with page complexity. The data cannot separate the two explanations. For US public bodies the regulatory pressure is real either way; the dates are in our ADA Title II deadline guide. Lawsuit and court data has its own page: ADA website lawsuits.

How Many People Have a Disability?

Between about one in six and more than one in four, depending on whose definition you use. These four official figures do not contradict each other. They measure different things.

WHO's "significant disability" is a much higher bar than the CDC's self-reported functional difficulty, and the CDC figure covers adults only. When you quote one of these, name the definition along with the number.

Two trends are worth knowing. In the UK, the number of disabled people rose from 12.9 million (20%) in 2014-15 to 16.7 million (25%) in 2024-25 (DWP, published 26 March 2026). In the EU, the share climbs steeply with age, from 7.1% of 16 to 24 year olds to 72.3% of people aged 85 and over (Eurostat, 2024). An aging customer base is a growing disabled customer base. If you sell into the EU, the European Accessibility Act guide covers what that means in law.

Which Disabilities Are Most Common?

In US adults, cognitive disability is the biggest group, at 13.9%, ahead of mobility at 12.2% and more than double vision at 5.5% (CDC DHDS, BRFSS 2022, age-adjusted). CDC's own disability infographic shows the same ranking.

The cognitive figure is also the one that has moved most. In the same CDC series it was 10.8% in 2016, 12.8% in 2021 and 13.9% in 2022, while mobility stayed between 11.1% and 12.9% over the whole period (CDC DHDS, BRFSS 2016 to 2022).

The UK numbers point the same way. Among disabled working-age adults, 46% reported a mental health impairment and 40% a mobility impairment; among disabled pensioners, mobility led at 68% (DWP Family Resources Survey, 2024-25).

Compare that with where accessibility effort usually goes. Alt text, ARIA and screen reader testing mostly serve blind and low-vision users, the 5.5%. The 13.9% with cognitive disabilities depend more on plain language, consistent navigation, forgiving time limits and clear error recovery, which are the hardest things to test automatically. Two caveats: the CDC's cognitive category ("serious difficulty concentrating, remembering or making decisions") is broad and self-reported, and a bigger group does not mean bigger barriers per person. None of this is a case for less screen reader work. It is a case for not treating "accessible" as a synonym for "works with a screen reader".

How Many People Need Assistive Technology?

More than 2.5 billion people need one or more assistive products, from glasses, hearing aids and wheelchairs to captioning and speech recognition software, and WHO expects that to reach 3.5 billion by 2050 as populations age and noncommunicable diseases rise (WHO assistive technology fact sheet, 2 January 2024). Access is very uneven: as few as 3% of people in some low-income countries have the products they need, against 90% in some high-income countries. For hearing alone, WHO counts 1.5 billion people with hearing loss and says hearing aid production meets less than 10% of global need.

Which Screen Readers Do People Use in 2026?

JAWS, by a wider margin than two years ago. WebAIM ran its eleventh Screen Reader User Survey in July and August 2026 and received 1,780 valid responses (WebAIM Screen Reader Survey #11, results updated 29 September 2026). For primary desktop or laptop screen reader:

In the 2024 survey, JAWS (40.5%) and NVDA (37.7%) were close (WebAIM Screen Reader Survey #10, 2024). The 2026 split is regional: JAWS led NVDA by 72.5% to 15.6% in North America, the two were level in Europe (42.8% and 42.3%), and NVDA led in Africa and the Middle East (83%), Asia (77.4%) and South America (67.2%). North America supplied 54.1% of 2026 respondents, though WebAIM says that alone does not explain the JAWS rise.

Other findings from the 2026 survey that matter when you build or test a site:

• People use more than one reader. 70.3% use more than one desktop screen reader. Narrator is the main reader for only 0.4%, yet 36.1% use it commonly. Test with at least JAWS and NVDA.
• Mobile is routine. 91.8% use a screen reader on a mobile device. 73.3% mainly use an iPhone or iPad and 24.0% Android, and 54.8% are more likely to use an app than a website for tasks like banking or shopping.
• Browsers: Chrome 52.3%, Edge 23.6%, Firefox 15.9%, Safari 4.8%. JAWS with Chrome is the most common pairing, at 31.2%.
• Headings are still the main way in. 67.8% look for headings first on a long page, down from 71.6% in 2024, and 88.3% find heading levels very or somewhat useful.
• PDFs are a problem; Word files much less so. 85.6% say a PDF is very or somewhat likely to cause them significant problems, up from 75.1% in 2019. For Word documents the figure is 31.1%.
• Sites matter more than software. 83.9% say more accessible websites would help more than better assistive technology, slightly down from 85.9% in 2024.
• Perceived progress is weak. 32.2% think web content became more accessible over the past year, 44.9% see no change and 22.9% think it got worse.
• AI is already part of the toolkit. 60.1% use AI to describe images, and only 8.2% use AI in none of the ways WebAIM listed.

WebAIM notes the sample "was not controlled and may not represent all screen reader users". Read the shares as a large, self-selected snapshot rather than a census.

How Accessible Are Mobile Apps?

Nobody has a WebAIM Million for apps, so there is no large, repeatable scan to quote. The best public study is small. ArcTouch and Fable tested 50 leading Android and iOS apps across five industries with four assistive technologies, and published the results on 15 April 2025 (ArcTouch, 2025; full report page). Only 2 of the 50 apps scored above 85 on ArcTouch's App Accessibility Score, and 9 got a failing grade. Android apps averaged 56 against 48 for iOS, shopping was the weakest industry at 41, and streaming was the strongest at 60. Their headline: 72% of people who rely on any of the four assistive technologies tested may have a poor or failing experience in at least one step of a typical user journey.

Treat that as a signal rather than a census: it is 50 apps, chosen and scored by the two companies that published it.

Five Statistics to Stop Quoting

Each of these is widely repeated, and each one falls apart when you trace it.

1. "$6.9 billion is lost every year to inaccessible e-commerce." The figure comes from a Nucleus Research study commissioned by Deque and announced on 15 August 2019. Per Deque's own press release, the research combined in-depth interviews with 73 blind or severely visually impaired US adults and a scan of hundreds of websites, and the $6.9 billion was a CEO quote about "potential North American e-commerce revenues" tied to a finding about the top ten online retailers (Deque, 2019). The download link on that page now reads "This report is no longer available." A seven-year-old, vendor-commissioned estimate nobody can read is not a statistic.

2. "71% of disabled users leave a site they find hard to use." The real source is the UK's Click-Away Pound survey of 2016, and it covers UK disabled internet users with access needs, not disabled people in general. The 2019 survey, a fresh sample, put it at 69% (Click-Away Pound 2019).

3. "The click-away pound is worth £17.1 billion." That is from the 2019 report (the 2016 estimate was £11.75 billion), and it is a model. The report puts the spending power of 7.15 million UK internet users with access needs at £24.8 billion, using an average online spend of £3,460 a head, then takes the 69% who say they click away (Click-Away Pound 2019). It is a market-size estimate built on 2019 data, not measured lost sales.

4. "Only 3% of the web is accessible." This most likely traces to the share of WebAIM pages with no detected WCAG failures: 2.6% in 2021 and 3.2% in 2022 (100% minus that year's failure rate), and 4.1% in 2026 (WebAIM Million, 2021 to 2026 reports). WebAIM itself says no detected errors does not mean accessible. Call it "pages with no automatically detected errors", or leave it out.

5. "1 billion people, 15% of the world, have a disability." That is the 2011 World Report on Disability from WHO and the World Bank, and WHO's page for that report now points readers to its newer data (WHO). The current fact sheet says 1.3 billion, 16%, or one in six (WHO, 2023).

Frequently Asked Questions

What percentage of websites are accessible?

No large study can tell you, because automated scans cannot confirm accessibility. What they can tell you: in 2026, 95.9% of the top million home pages had at least one automatically detectable WCAG failure, and WebAIM says full conformance was "certainly lower than 4.1%" (WebAIM Million, 2026).

What is the most common accessibility error?

Low-contrast text, found on 83.9% of top home pages in 2026, with 34 instances on the average page (WebAIM Million, 2026). Missing image alt text is second, at 53.1% of pages.

How many people have a disability?

1.3 billion worldwide (16%) by WHO's measure of significant disability (2023). 28.7% of US adults (CDC, 2022), 25% of the UK population (DWP, 2024-25) and 23.9% of EU residents aged 16+ (Eurostat, 2024) under their own national definitions.

What is the most popular screen reader in 2026?

JAWS, the primary desktop screen reader for 55.0% of respondents to WebAIM's 2026 survey, followed by NVDA at 32.9% and VoiceOver at 6.5% (WebAIM Screen Reader Survey #11, 2026). On mobile, VoiceOver is used by 72.2% and TalkBack by 29.5%.

How many people use assistive technology?

WHO estimates that more than 2.5 billion people need at least one assistive product, rising to 3.5 billion by 2050 (WHO, 2024). Need is not the same as access: in some low-income countries only 3% of people who need a product have one.

The Bottom Line

The 2026 numbers give a consistent picture. The web's failure rate went back up to 95.9%, pages grew faster than their error counts, and the same six error types still account for almost everything scanners find. On the people side, more than one in four US adults report a disability, the largest group is cognitive rather than visual, and screen reader users now lean further toward JAWS and rely on headings and mobile as much as ever.

To act on it, scan your own pages, fix the high-volume errors first using the WCAG violations fix guide, and stop regressions with an automated check in CI as described in the GitHub Actions guide. If you quote a figure from this page, quote its source and year as well.

Sources (all checked on 15 October 2026)

• WebAIM Million, 2026 report (February 2026 data, page updated 30 March 2026): https://webaim.org/projects/million/
• WebAIM Million reports for 2025, 2024, 2023, 2022 and 2021, used for the trend tables and chart: https://webaim.org/projects/million/2025, https://webaim.org/projects/million/2024, https://webaim.org/projects/million/2023, https://webaim.org/projects/million/2022, https://webaim.org/projects/million/2021
• WebAIM Screen Reader User Survey #11 (fieldwork July to August 2026, n=1,780): https://webaim.org/projects/screenreadersurvey11/
• WebAIM Screen Reader User Survey #10 (fieldwork December 2023 to January 2024, n=1,539): https://webaim.org/projects/screenreadersurvey10/
• World Health Organization, disability fact sheet, 7 March 2023: https://www.who.int/news-room/fact-sheets/detail/disability-and-health
• World Health Organization, assistive technology fact sheet, 2 January 2024: https://www.who.int/news-room/fact-sheets/detail/assistive-technology
• World Health Organization, World Report on Disability 2011 page: https://www.who.int/teams/noncommunicable-diseases/sensory-functions-disability-and-rehabilitation/world-report-on-disability
• Eurostat, Population with disability (EU-SILC, 2024 data): https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Population_with_disability
• UK Department for Work and Pensions, Family Resources Survey 2024-25, published 26 March 2026: https://www.gov.uk/government/statistics/family-resources-survey-financial-year-2024-to-2025/family-resources-survey-financial-year-2024-to-2025
• CDC Disability and Health Data System, prevalence of disability status and types (BRFSS 2016 to 2022, age-adjusted): https://data.cdc.gov/Disability-Health/DHDS-Prevalence-of-Disability-Status-and-Types-by-/s2qv-b27b
• CDC, Disability Impacts All of Us infographic: https://www.cdc.gov/disability-and-health/articles-documents/disability-impacts-all-of-us-infographic.html
• ArcTouch, The State of Mobile App Accessibility, 15 April 2025: https://arctouch.com/blog/state-of-mobile-app-accessibility
• Deque Systems press release, 15 August 2019 (origin of the $6.9 billion figure): https://www.deque.com/blog/research-shows-internet-is-unavailable-to-blind-users/
• Click-Away Pound Survey 2016 report: https://www.clickawaypound.com/cap16finalreport.html
• Click-Away Pound Survey 2019 report: https://www.clickawaypound.com/downloads/cap19final0502.pdf