Web accessibility statistics for 2026 tell an uncomfortable story: after six straight years of slow improvement, the web went backwards. WebAIM’s February 2026 analysis of the top one million homepages found 95.9% with at least one detectable WCAG 2 failure — up from 94.8% a year earlier — while average errors per page rose 10.1% to 56.1.
The regression landed in the same twelve months that US accessibility litigation reached 4,928 filed lawsuits (UsableNet, January 2026), the FTC collected a $1 million penalty from an accessibility-widget vendor, and French courts issued the European Accessibility Act’s first two rulings — one for the plaintiffs, one for the defendant. More tooling, more automation, more enforcement; worse outcomes. That paradox is the story of this dataset.
This reference compiles the numbers that survived source verification: the WebAIM Million 2026 edition, UsableNet’s 2025 year-end lawsuit report, FTC enforcement records, and the first EAA court decisions. Every figure is attributed and dated in-line, and Section 07 covers the widely recycled statistics we refused to publish — including one the original source itself has since revised upward by roughly 38%.
- 01The first WCAG regression in six years.95.9% of the top one million homepages had a detectable WCAG 2 failure in the February 2026 WebAIM Million report, up from 94.8% in the 2025 edition. Average errors per page rose 10.1% to 56.1.
- 02The same six failures cause 96% of all errors.Low contrast (83.9% of pages), missing alt text (53.1%), missing form labels (51%), empty links (46.3%), empty buttons (30.6%), and missing document language (13.5%) — the top six for seven consecutive years, per WebAIM.
- 034,928 US lawsuits in 2025 — and overlays didn’t help.UsableNet counted 4,928 US web accessibility lawsuits in 2025, with federal filings up 27% year over year. About 22.6% of H1 2025 suits hit sites that had an accessibility overlay installed.
- 04Europe’s first EAA rulings split both ways.Carrefour was ordered on June 4, 2026 to make its site and app accessible within six months; the parallel Auchan case was dismissed on May 5, 2026. EAA enforcement is real but not uniformly pro-plaintiff.
- 05The most-quoted market stats are zombie numbers.The “$13 trillion disability market” figure dates to a 2020 report whose own publisher revised it to $18 trillion in 2024. Section 07 lists every recycled stat we refused to print, with vintages.
01 — The HeadlineThe first regression in six years.
The WebAIM Million — WebAIM’s annual automated evaluation of the top 1,000,000 homepages — is the closest thing web accessibility has to a census. The February 2026 edition found 95.9% of homepages with at least one detectable WCAG 2 failure, up from 94.8% in the 2025 edition. That is the first year-over-year increase after six consecutive years of small improvements. Detected errors climbed in step: 56.1 average errors per homepage, up 10.1% from 51.0, with 56,114,377 distinct errors found across the full sample (WebAIM Million, February 2026).
WebAIM’s own explanation for the reversal is the most quotable line in the report: the organization attributes rising errors to “increased reliance on 3rd party frameworks and libraries, and automated or AI-assisted coding practices” — adding its own parenthetical, “vibe coding” (WebAIM Million, February 2026). The supporting data points in the same report make the mechanism plausible rather than speculative: pages are getting heavier and more ARIA-laden faster than they are getting fixed.
Avg elements per homepage
Up 22.5% in a single year and nearly double the element count of 2019. WebAIM explicitly links rising page complexity to rising error counts (WebAIM Million, Feb 2026).
Avg ARIA attributes per page
82.7% of pages now use ARIA, averaging 133.6 attributes per page — up 27% year over year. ARIA adoption keeps climbing without the underlying issues being fixed (WebAIM Million, Feb 2026).
Pages with ARIA vs 42.0 without
Homepages using ARIA averaged 59.1 detected errors versus 42.0 on pages without ARIA — the long-documented pattern of ARIA increasing risk when misused (WebAIM Million, Feb 2026).
Our reading of the trend: this is not a knowledge problem. WCAG 2 has been stable for years, the failing checks are the same ones automated scanners have flagged for a decade, and the fixes are documented everywhere. What changed in the 2025–2026 window is the volume of shipped markup per team — AI-assisted coding and component libraries let organizations produce far more UI surface than their review processes can audit. When output scales and accessibility review doesn’t, error counts rise mechanically. One accessibility consultant, writing at accessibility.chat in April 2026, put it precisely: “The 2026 WebAIM Million data reveals an operational capacity failure, not a knowledge failure.”
02 — Failure TypesSix failures, 96% of all errors.
The distribution of failures is remarkably concentrated. Per the February 2026 WebAIM Million, six failure types account for 96% of all detected errors — and they have been the same top six for seven consecutive years: low-contrast text, missing image alt text, missing form input labels, empty links, empty buttons, and a missing document language attribute. None of these requires exotic tooling to find or fix; all six are caught by free automated scanners.
The six failure types behind 96% of detected errors
Source: WebAIM Million, February 2026 edition — % of top-1M homepages affectedLow-contrast text remains the most prevalent failure by a wide margin — found on 83.9% of homepages, up from 79.1% in the 2025 edition, averaging 34 distinct low-contrast instances per page, a 15% year-over-year increase (WebAIM Million, February 2026). If you run exactly one automated check on your site this quarter, make it contrast. If you want the full manual-plus-automated pass, our WCAG 2.2 audit checklist maps every criterion to a concrete test.
03 — AttributionDating the data correctly.
Here is the part most coverage gets wrong. The 95.9% failure rate and 56.1 errors-per-page figures belong to the February 2026 edition of the WebAIM Million. The 94.8% and 51.0 figures belong to the 2025 edition. During our research pass we found multiple 2026-dated articles quoting the new numbers as “2025 data,” and others still citing the older, better numbers as if current — which quietly erases the entire headline finding, because the difference between those two pairs of numbers is the regression.
| Report year | Pages with WCAG 2 failures | Avg errors / page | YoY change | Note |
|---|---|---|---|---|
| 2019 | 97.8% | — | — | Earliest edition in this trend line (approximate, secondary-sourced) |
| 2024 | 95.9% | 56.8 | — | Aggregated from WebAIM’s annual editions (approximate, secondary-sourced) |
| 2025 | 94.8% | 51.0 | −1.1 pts · −5.8 errors | Sixth straight year of small gains |
| 2026 | 95.9% | 56.1 | +1.1 pts · +5.1 errors | First regression in six years |
This is a stats-hygiene pattern worth generalizing: annual reports are named for their publication year, not the year of the data window, and secondary coverage conflates the two constantly. We apply the same dating discipline across our broader 2026 website statistics and our Core Web Vitals pass-rate benchmarks — every figure names its edition.
04 — US Litigation4,928 lawsuits, and e-commerce takes 70% of them.
UsableNet’s 2025 Year-End Digital Accessibility Lawsuit Report (published January 8, 2026) counted 4,928 US web accessibility lawsuits filed in 2025, a total commonly rounded to “over 5,000” once all jurisdictions are included. The cleanest year-over-year comparison is the federal docket: 3,117 federal filings in 2025, up 27% from 2,452 in 2024 (UsableNet year-end reports, January 2026 and January 2025). We use the federal-only comparison deliberately — secondary sources routinely conflate the 2023 and 2024 all-court totals, so the federal series is the least ambiguous single trend line.
US suits filed in 2025
All jurisdictions, commonly rounded to 5,000+. Federal filings rose 27% year over year to 3,117 (UsableNet 2025 Year-End Report, Jan 2026).
Of suits target e-commerce
Retail and e-commerce sites remain the dominant target class, and 35.8% of the top-500 e-commerce retailers received at least one accessibility lawsuit in 2025 (UsableNet, Jan 2026).
Suits against repeat defendants
About 29% of 2025’s suits targeted companies already sued before; within federal court, 46% of cases involved repeat defendants. Settling once does not end exposure (UsableNet, Jan 2026).
Geography is just as concentrated as sector. New York and California state courts combined saw nearly 2,000 filings in 2025, and New York alone accounts for more than one-third of all state-level suits (UsableNet, January 2026). For online retailers the compound risk is straightforward: the sector with the highest litigation share is also the one where checkout friction has direct revenue cost — context we cover in our 2026 ecommerce statistics. An inaccessible checkout is simultaneously a legal exposure and a conversion leak.
05 — OverlaysOverlays bought no protection — and drew the FTC.
The most consequential finding in the litigation data concerns accessibility overlays — the one-line-of-code widgets marketed as instant compliance. Per UsableNet’s 2025 midyear report, about 22.6% of H1 2025 lawsuits — roughly 456 cases — targeted websites that had an overlay installed. The 2024 full-year picture was consistent: 1,023 lawsuits, a quarter of that year’s total, explicitly named an overlay or widget as a barrier, and in January 2025 alone 85 defendants were sued while running third-party overlay tools (Law Office of Lainey Feingold analysis of UsableNet data, February 2025). Sites with widgets were sued at a materially similar rate to sites without them.
“One-line-of-code overlays do not make a website accessible to disabled people. They do not protect companies that license them from lawsuits under the Americans with Disabilities Act.”— Lainey Feingold, disability rights attorney, February 2025
Regulators reached the same conclusion. On January 3, 2025, the FTC announced that accessiBe would pay $1 million over deceptive claims that its AI-powered accessWidget could make any website WCAG-compliant, plus deceptively formatted third-party “reviews” that concealed the company’s involvement; the order was finalized in April 2025. Samuel Levine, then Director of the FTC’s Bureau of Consumer Protection, said: “Companies looking for help making their websites WCAG compliant must be able to trust that products do what they are advertised to do. Overstating a product’s AI or other capabilities without adequate evidence is deceptive, and the FTC will act to stop it” (FTC press release, January 3, 2025).
Read the two datasets together and the conclusion writes itself: overlays neither fix the six failure types driving 96% of errors nor reduce the statistical probability of being sued — and marketing them as if they do is now an FTC enforcement matter.
06 — EuropeEAA enforcement arrived — with split verdicts.
The European Accessibility Act became enforceable on June 28, 2025, requiring digital products and services sold to EU consumers — including e-commerce — to meet accessibility requirements regardless of where the vendor is based. Enforcement started fast. French disability organizations ApiDV and Droit Pluriel, working with the Intérêt à Agir legal collective, sent formal notices to Auchan, Carrefour, E.Leclerc, and Picard Surgelés in July 2025, then filed emergency injunctions in French court on November 12, 2025 after judging the responses inadequate (TestParty case coverage, 2025).
The first two rulings split. On June 4, 2026, the Tribunal judiciaire de Caen ruled against Carrefour France, ordering the company to make both its e-commerce site and its mobile app fully accessible within six months, with a per-day fine accruing after the deadline (Droit Pluriel statement, June 2026; corroborated by Silktide and Deque coverage). But the parallel Auchan case — transferred to Lille and heard March 3, 2026 — was dismissed on May 5, 2026, with the plaintiff organizations calling the reasoning “a manifestly erroneous interpretation of the legal texts” in reported statements. The first EAA scoreboard reads one-for-one: enforcement is real, but outcomes are not uniformly pro-plaintiff. Meanwhile Sweden’s Post and Telecom Authority began EAA market surveillance in October 2025, inspecting devices and opening its first e-commerce regulatory cases (reported PTS coverage, 2025–2026).
| Member state | Maximum fixed fine | Turnover-linked ceiling / scope note |
|---|---|---|
| Turnover-linked penalty regimes | ||
| Netherlands | Up to €900,000 | Up to 10% of revenue |
| Hungary | Up to ~€1,260,000 | Up to 5% of annual turnover |
| Italy | Up to €40,000 | Up to 5% of turnover |
| Fixed-ceiling penalty regimes | ||
| Spain | Up to €1,000,000 | Tiered from €30,000 for “very serious” violations |
| Germany | Up to €100,000 | Per violation |
| France | Up to €250,000 | Non-accessible public-facing platforms |
| Ireland | From €60,000 | Entry-level exposure |
07 — House FormatStats we refused to publish.
Accessibility content has a zombie-stat problem: numbers that were defensible once, got detached from their vintage, and now circulate as if current. The sharpest example in this dataset is self-correcting: the “$13 trillion global disability market” figure traces to the Return on Disability Group’s 2020 Annual Report — and the same organization’s 2024 Annual Report revised it to $18 trillion. Most 2026-dated posts still quote the 2020 number, which means they disagree with their own source by roughly 38%. Here is every recycled figure we deliberately left out of this reference, and why.
| Stat you’ll see elsewhere | Vintage / origin | Status in 2026 | Our call |
|---|---|---|---|
| “£274 billion Purple Pound” | Scope-era UK estimate, roughly 2018–2020 vintage | University of Bristol research cited in Purple Tuesday materials (~Nov 2025) puts the same UK metric at £446 billion | Refused without a vintage label — the recycled figure runs nearly 40% below the current best estimate |
| “$13 trillion global disability market” | Return on Disability Group 2020 Annual Report (Sep 2020) | The same organization’s 2024 Annual Report (Sep 2024) revised the figure to $18 trillion — roughly 38% higher | Refused as current — superseded by its own source |
| A single “digital accessibility software market size” | Three vendor research firms, 2026 estimates | Estimates span ~$0.93B to ~$1.9B for the same year — a roughly 2x disagreement with no methodology reconciliation | Refused to pick one — we describe the range instead |
| Vendor “research” explaining the 2026 regression | accessiBe self-published analysis, republished by trade press in 2026 | The FTC ordered the same vendor to pay $1 million in 2025 over deceptive WCAG-compliance claims | Refused as neutral analysis — vendor-interested commentary |
| “X% of accessibility lawsuits are frivolous” | Pre-2024 trade-press assertions without methodology | No credible, dated primary source found in our research pass | Refused — no percentage exists worth printing |
| “EAA penalties up to 4% of revenue” | A round number circulating across compliance blogs | No confirmed member state matches 4%; sourced figures are 5% (Italy, Hungary) and 10% (Netherlands) | Refused — we publish the per-country table instead |
The “Purple Pound” case deserves one extra sentence because the correction is so large. The £274 billion UK disability spending-power figure is a Scope-era estimate from roughly 2018–2020; University of Bristol research cited in Purple Tuesday campaign materials around November 2025 puts the same metric at £446 billion. Anyone quoting £274 billion today is understating the market by nearly 40% — while appearing to make the pro-accessibility case. Zombie stats don’t just age badly; they can undersell the very argument they are recycled to support.
08 — ImplicationsWhat this means for site owners.
The data points to one strategic conclusion: accessibility risk in 2026 is structural, and the fixes have to be structural too. The six failure types behind 96% of errors are all root-cause issues in markup, components, and design tokens — none of them can be papered over by a client-side widget, and the litigation data shows widget-equipped sites getting sued at a materially similar rate anyway.
One-line “instant compliance”
About 22.6% of H1 2025 suits hit overlay-equipped sites, and the FTC fined a leading vendor $1M over compliance claims. Buys no measurable legal protection and fixes none of the top six failure types.
Automated scanner, no remediation loop
Scanners catch the six dominant failure types, but the 2026 regression happened in a world full of scanners. Detection without an owner, a budget, and a fix pipeline changes nothing.
Fix the root causes
Audit against WCAG 2.2, fix contrast tokens, alt text, labels, and accessible names at the component level, then verify with assistive tech. Addresses the failure classes that drive both errors and lawsuits.
Accessibility gates in CI
The accessibility.chat diagnosis — an operational capacity failure — implies the fix: automated checks on every pull request plus periodic manual audits, so AI-accelerated shipping can’t outrun review.
Looking forward from July 2026, the two enforcement tracks are converging rather than cooling. In the US, the federal docket has grown 27% year over year with repeat defendants at 46% of federal cases — a pattern that rewards fixing sites once, properly, over settling serially. In Europe, the Carrefour ruling established that an EAA court will order remediation on a six-month clock with daily fines, and member-state surveillance programs like Sweden’s were already opening e-commerce cases before the first anniversary of enforceability. The rational move is to treat the six dominant failure types as a defined engineering backlog now, not as a reaction to a demand letter later. That component-level remediation work is exactly what our web development team builds into client projects — accessible components by default, verified in CI, rather than retrofitted under deadline.
09 — ConclusionThe web got worse while the stakes got higher.
Accessibility regressed the same year enforcement matured on two continents.
The 2026 numbers are unambiguous: 95.9% of top homepages fail WCAG 2, errors per page are up 10.1%, and the same six fixable failure types still drive 96% of the problem (WebAIM Million, February 2026). At the same moment, US litigation reached 4,928 annual filings, the FTC turned overlay marketing into an enforcement target, and European courts issued the EAA’s first rulings — split verdicts included.
Two disciplines separate useful accessibility reporting from recycled content. First, date everything: the difference between the 2025 and 2026 WebAIM editions is the entire headline, and coverage that misdates them erases it. Second, retire zombie stats: when the Return on Disability Group has revised $13 trillion to $18 trillion and the Purple Pound’s own ecosystem now cites £446 billion, quoting the old numbers weakens an argument the data actually strengthens.
For site owners the playbook is unchanged but the urgency is not. The failure types are known, automatable, and concentrated; the legal exposure is growing on both sides of the Atlantic; and the one shortcut on the market demonstrably does not shortcut anything. Fix the structure, gate it in CI, and the same work that removes legal risk also removes the barriers the statistics keep counting.