A research briefing compiled from regulatory monitoring data, academic literature and sector research

See Me Please

Prepared July 2026 — updated to reflect court outcomes as of publication

A Quick Note on How We Built This

Before we get into it: here's how we chose what to include, because a briefing like this is only as trustworthy as the sourcing behind it, and we'd rather show our working than ask you to take it on faith.

We prioritised primary sources wherever they exist — the European Commission's own EAA and WAD publications, GOV.UK/GDS's published PSBAR monitoring reports, WebAIM's own Million reports (not summaries of them), Ireland's National Disability Authority, Eurostat, and peer-reviewed academic research. Where a figure could only be traced to a vendor blog or a compliance-tool marketing page — several of which sell accessibility testing and auditing services and have a direct commercial interest in the scale of non-conformance looking as large as possible — we've either cross-checked it against a primary source before including it, or flagged it explicitly as industry commentary rather than independent research. We'd rather have a shorter list of solid numbers than a longer one padded with figures we can't stand behind.

We also went back through everything with a genuinely adversarial eye, asking the same questions a sharp competitor or a sceptical reader would ask: does this number actually appear in the source it's attributed to? Does mixing years anywhere make a trend look more uniform than it really is? Is anything here already out of date? That process caught real problems in an earlier draft — including a stale enforcement section and a set of year-mixed statistics — and the corrections are reflected throughout what follows.

A note on timing: this briefing was prepared in July 2026. Regulatory and legal developments in this space move fast, particularly EAA enforcement. We've dated the specific claims that are most time-sensitive so you can judge for yourself whether anything's moved on since.

Executive Summary

Two regulatory regimes now govern digital accessibility across Europe and the UK: the EU's Web Accessibility Directive (2016) and European Accessibility Act (enforceable since 28 June 2025) on one side, and the UK's Public Sector Bodies Accessibility Regulations (2018) and Equality Act 2010 on the other. More than a decade into legal obligation, independent monitoring still finds that the large majority of public and private websites fail basic conformance. The WebAIM Million study found 94.8% of the top one million home pages had detectable WCAG failures in 2025 — and that figure went back up to 95.9% in 2026, reversing several years of slow progress. UK government monitoring under PSBAR found accessibility issues on almost every website it tested between 2022 and 2024. Ireland's National Disability Authority found average accessibility scores of only 46–55% across public bodies reviewed in 2024–2025.

The pattern holds across sectors: government services, banking, insurance and universities all show recurring, well-documented friction points — many of them the same six technical failures that have shown up in every WebAIM Million report for seven years running. But “they haven't fixed it” isn't quite the full story for every category, and we've corrected that nuance below: four of the six got worse between 2025 and 2026, but two — missing alt text and missing language declaration — actually improved. The human impact, meanwhile, is genuinely heterogeneous: screen reader users, low vision users relying on magnification and contrast, Deaf and hard of hearing users needing captions, neurodivergent users affected by cognitive load and sensory overload, and older adults facing compounding age- and disability-related exclusion each experience different, sometimes conflicting, barriers.

On enforcement: this is the section that changes fastest, and where we've made our most significant correction. By the time this briefing was prepared, two EU court outcomes already existed that don't appear in most accessibility briefings still circulating — one case dismissed on procedural grounds, one that went the other way and resulted in a binding compliance order. We've covered both in Part V, because leaving them out would make this look like a mid-2025 briefing wearing a 2026 date.

Part I — The Legislative Evolution

1.1 The EU Track: From Web Accessibility Directive to the European Accessibility Act

The EU's accessibility law didn't arrive all at once — it built in stages, and the stage we're in now (private-sector enforcement) is genuinely new. Here's the sequence:

  • 2016 — the Web Accessibility Directive (EU) 2016/2102 is adopted, covering public sector websites and mobile apps only.

  • 23 September 2018 — WAD compliance deadline for public sector websites.

  • 2019 — the European Accessibility Act (Directive (EU) 2019/882) is adopted: the first EU-wide private sector accessibility law, covering e-commerce, banking, transport ticketing, e-books, telecoms and more.

  • 23 June 2021 — WAD compliance deadline for public sector mobile apps.

  • 28 June 2022 — deadline for member states to transpose the EAA into national law; all 27 states completed transposition.

  • 28 June 2025 — EAA enforcement begins. All newly published digital content covered by the Act must comply.

  • July 2025 — France issues the first formal EAA-related legal notices, to major retailers including Auchan, Carrefour, E.Leclerc and Picard Surgelés.

  • October 2025 — Sweden launches market surveillance activity.

  • 12 November 2025 — the first EAA-related lawsuits are filed, as emergency injunctions in French court; the Dutch ACM begins active enforcement for e-commerce and electronic communications.

  • May and June 2026 — the first two EAA-linked court rulings land, with opposite outcomes. We cover both in full in Part V, because this is the part of the story most other briefings are currently missing.

  • 28 June 2027 — compliance deadline for service contracts concluded before June 2025.

  • 28 June 2030 — final deadline: all existing digital content, not just newly published content, must be fully EAA-compliant.

The EAA and WAD are complementary rather than duplicative: the WAD covers public sector websites and apps, while the EAA extends binding, functional accessibility requirements to the private sector for the first time — e-commerce, banking, insurance-adjacent financial services, transport, e-books and telecommunications. Both anchor to the harmonised technical standard EN 301 549, itself aligned with WCAG (currently 2.1 AA, with EN 301 549 v4.1.1 incorporating WCAG 2.2 expected). The European Commission has estimated that fragmented, country-by-country accessibility requirements cost companies and member states approximately €20 billion in 2020 — the core efficiency rationale for a single harmonised EU regime.

Enforcement is deliberately decentralised: each of the 27 member states designates its own market surveillance authority and sets its own penalty regime. That produces a genuinely wide fine range — approximately €60,000 in Ireland versus approximately €900,000 in Sweden — and means a single non-compliant pan-European digital service can face separate investigations in every jurisdiction where it operates. The EAA's population rationale cites an estimated 87–101 million people over 16 living with some form of disability in the EU, around 25% of the adult population.

1.2 The UK Track: Public Sector Regulations, the Equality Act, and Patchwork Enforcement

The UK runs two separate legal instruments side by side. The Public Sector Bodies (Websites and Mobile Applications) (No. 2) Accessibility Regulations 2018 (PSBAR) require public sector websites and apps to meet WCAG 2.2 AA (via EN 301 549) and publish an accessibility statement. The Government Digital Service (GDS, formerly the Central Digital and Data Office) monitors compliance and refers persistent non-compliance to the Equality and Human Rights Commission (EHRC) or the Equality Commission for Northern Ireland. Separately, the Equality Act 2010 (the Disability Discrimination Act 1995 in Northern Ireland) imposes an anticipatory duty on all UK service providers — public and private — to make reasonable adjustments, including for websites, which the EHRC's statutory Code of Practice explicitly identifies as services.

Since Brexit, the EAA doesn't directly apply in the UK, but UK businesses selling into the EU are caught by it regardless of domestic legislation. UK legal commentary consistently describes enforcement of the Equality Act for digital services as “patchy at best,” hindered by the absence of a dedicated accessibility regulator and by the cost and complexity of UK court procedure. No landmark UK court ruling on website accessibility under the Equality Act has been delivered yet, though the RNIB has initiated and settled multiple cases, and legal commentators widely regard a high-profile test case as a matter of “when, not if.”

Here's where we need to flag something explicitly. Our first draft of this briefing repeated a specific, often-cited claim: that the EHRC has “sent enforcement letters to 93 public sector bodies in England and Wales since January 2022, securing compliance with 66,” alongside an anecdote about a college facing a threatened £200,000 fine after its audit correspondence went to spam. We went back to check both figures directly against EHRC, GDS and AbilityNet material and couldn't verify either one in any primary source — which means, however widely they've been repeated elsewhere in the sector, we're not going to present them as confirmed fact. We'd rather tell you what we can actually stand behind:

What's actually verified from GDS's own published data

  • GDS's second PSBAR monitoring report (2022–2024, published December 2024) tested 1,151 websites and found accessibility issues on all but 10 of them.

  • 69 websites failed to respond to multiple monitoring contact attempts and were referred directly to the equality bodies — this is the genuine, sourced version of the escalation pathway, even without the specific 93/66 figures we couldn't confirm.

  • 6% of organisations don't respond to monitoring correspondence at all — the exact failure mode that a spam-filtered or ignored audit email would produce, whatever the outcome of any individual case.

If you've seen the 93/66/£200,000 figures used elsewhere and want to cite them, our honest advice is to get them confirmed directly with EHRC first — we couldn't, and we're not willing to put our name behind a number we can't trace.

1.3 Comparing the Two Regimes

Rather than a side-by-side grid, here's the shape of the difference in plain terms, because the two regimes genuinely diverge in scope, standard and enforcement style.

The EU side (EAA + WAD)

  • Scope: public sector under the WAD, plus private sector under the EAA — e-commerce, banking, transport, telecoms, e-books, TVs.

  • Technical standard: EN 301 549, itself built on WCAG 2.1 AA and moving toward 2.2.

  • Regulator: decentralised — one market surveillance authority per member state, so enforcement style and pace vary considerably by country.

  • Penalties: from roughly €60,000 in Ireland up to approximately €900,000 in Sweden, with some regimes empowered to fine up to 4% of annual revenue.

  • Micro-enterprise exemption: yes — under 10 employees and €2m turnover or balance sheet.

The UK side (PSBAR + Equality Act)

  • Scope: public sector under PSBAR, plus — in principle — all service providers, public and private, under the Equality Act.

  • Technical standard: WCAG 2.2 AA directly.

  • Regulator: GDS handles public-sector monitoring; the EHRC handles enforcement; there's no dedicated private-sector accessibility regulator.

  • Penalties: reasonable-adjustment claims via courts and tribunals, EHRC enforcement letters and binding agreements — no statutory fine schedule equivalent to the EAA's.

  • Private sector coverage: yes in principle under the Equality Act, but enforcement is reactive and complaint-driven, with almost no case law to date.

Part II — What the Monitoring Data Actually Shows, and Why

2.1 The WebAIM Million: Seven Years of Automated Evidence

WebAIM has scanned the top one million home pages on the web every year since 2019, using its own WAVE engine. Here's the headline trajectory: 97.8% of pages had a detectable WCAG failure in 2019; that eased to 98.1% in 2020, 96.3% in 2023, 95.9% in 2024, and 94.8% in 2025 — a real, if modest, six-year improvement. Then 2026 reversed it: back up to 95.9%. Average errors per page followed a similar arc — 60.9 in 2020, down to 50.0 in 2023, up to 56.8 in 2024, down to 51.0 in 2025, and back up to 56.1 in 2026. Average page element count, meanwhile, has climbed in almost a straight line — 1,173 in 2024, 1,257 in 2025, 1,437 in 2026 — and that complexity growth is central to why the reversal happened at all (more on that below).

The same six error categories have accounted for 96%+ of all detected failures every single year of the study — seven years running — which is genuinely one of the most consistent findings in web accessibility research. But here's the correction we owe you: not all six moved in the same direction between 2025 and 2026, and an earlier version of this briefing gave contrast the “it's getting worse” update while quietly leaving the others at flat 2025 figures. That made the picture look more uniformly bad than it actually was. Here's the honest, paired comparison for all six:

The same six failures, 2025 → 2026, all paired up properly

  • Low-contrast text: 79.1% → 83.9% of pages — worse, and still the single most common failure by a wide margin.

  • Missing form input labels: 48.2% → 51.0% — worse.

  • Empty links: 45.4% → 46.3% — worse.

  • Empty buttons: 29.6% → 30.6% — worse.

  • Missing alternative text: 55.5% → 53.1% — actually improved.

  • Missing document-language declaration: 15.8% → 13.5% — also improved, continuing a longer decline from 33.1% back in 2019.

So: four of six got worse, two got better. The two that improved — alt text and language declaration — have something in common worth noting: both are effectively single-decision, server-renderable fixes that benefit from CMS-level automation, whereas contrast, form labelling and empty links/buttons all depend on discipline at the point of design and development, which is a much harder thing to automate your way out of. Only around 17% of data tables, separately, are properly coded for assistive technology. Government and education sites perform best of any sector category (average scores around 53–56); travel and e-commerce perform worst (scores of 34–36, error rates above 97%).

2.2 Why: The Root-Cause Pattern Behind Stagnant Compliance

The research literature and monitoring bodies converge on a consistent causal account here — and it's not a single cause, which is exactly why a single fix won't solve it:

  • Rising page complexity is outpacing accessibility practice. Average homepage element count has nearly doubled in seven years, and WebAIM explicitly attributes the 2026 reversal to this complexity growth outpacing remediation effort. Pages using ARIA average more than twice the errors of pages that don't (57 vs 27) — a strong signal that ARIA is being misapplied more often than it's being adopted correctly.

  • There's a governance and ownership gap. AbilityNet's Attitudes to Digital Accessibility survey found that while over half of respondents rate accessibility as a significant part of their role, only 7% of senior leadership treat it as a “very high priority.” Separately, 43% of organisations cite “lack of clear ownership” as their single biggest internal barrier, and 65% have never tested their website with a real disabled user — which, if you're in the business of user testing, is the statistic that should genuinely concern you most.

  • Automated testing only catches a fraction of real barriers. Automated scanners — including the WAVE engine behind the WebAIM Million — detect roughly 30% of WCAG issues. Manual testing, particularly keyboard operability, focus visibility and screen-reader behaviour, surfaces the majority of what automated tools miss entirely. In other words: the headline numbers above are a floor, not a ceiling, on how bad things actually are.

  • Accessibility statements decay. UK PSBAR monitoring finds that accessibility statements — the transparency mechanism the whole regulation hinges on — are frequently not updated, despite being required at least every 12 months. That undermines the entire monitoring feedback loop.

  • Procurement is where a lot of this actually gets baked in. AbilityNet's research found very few organisations recognise procurement's role in delivering accessibility, meaning inaccessible third-party platforms, LMSs and vendor tools get locked into institutional technology stacks well before any accessibility team is in the room.

  • Documents are the most neglected category of all. PDFs remain one of the largest sources of accessibility failure specifically because document accessibility — tagging, structure, reading order — sits outside standard web development workflows and skillsets altogether.

2.3 UK Public Sector Monitoring: The GDS/PSBAR Findings in Detail

GDS's second PSBAR report (2022–2024, published December 2024) tested 1,151 websites via simplified review and found no accessibility issues on only 10 of them. Half of tested websites had 23 or fewer issues; the worst had 73. Detailed audits across 1,203 websites and 21 mobile apps found nearly 30,000 accessibility issues in total. The process itself produced real remediation, which is the genuinely encouraging part: 16,482 issues were fixed as a direct result of monitoring, 55.3% of identified issues were resolved during the monitoring period, and 67.9–68% of organisations either fixed issues or produced a credible short-term plan, avoiding further enforcement referral. But 69 websites failed to respond to multiple contact attempts and were referred directly to the equality bodies, and 6% of organisations never respond to monitoring correspondence at all — the genuine, GDS-sourced version of the escalation risk we flagged in Part I.

Part III — Sector Friction: What the Research Says

3.1 Government and Public Services

The clearest, best-evidenced friction point in government digital services is the collision between “digital by default” policy and the disproportionate exclusion of disabled and older users. UK research on the Universal Credit online benefit system documents claimants' lived frustration, anxiety and burden navigating a system explicitly designed as digital-first (Watson et al., Newcastle University Open Lab). Scottish government data makes the disparity concrete: only 49% of Pension Age Disability Payment applications are made online, compared with 91% of Scottish Child Payment applications — clear evidence that channel-shift policy lands very differently on different populations. DWP customer experience data shows internet access itself varies by claimant group, with Pension Credit recipients — the lowest-income group measured — reporting the lowest internet access (78%) of any benefit population.

Academic literature frames this using the concept of “accessibility regimes” and “access conditionality”: digital-by-default policy can function as a de facto eligibility filter, disproportionately screening out exactly the populations the service is meant to serve. One documented real-world case: a blind screen-reader user, Stephen Campbell, was unable to complete an online promotion application through Health and Social Care Northern Ireland's inaccessible recruitment portal, resulting in a settlement and a policy commitment from the Trust.

3.2 Banking and Financial Services

Banking is one of the most heavily documented friction sectors, and the research shows the problem isn't confined to public-facing marketing pages — it's concentrated in the authenticated, transactional experience, which is precisely the part disabled customers most need to use independently. Nearly 70% of bank websites were found to be inaccessible via keyboard-only navigation; almost 40% lack basic screen-reader accessibility features; more than three in five people report being unable to use financial websites or apps at all because of access barriers; and three-quarters of visually impaired online banking users report having to rely on a sighted person to complete financial tasks.

Specific, documented failure points include: bank statements delivered as untagged PDFs that screen readers read out of order; transaction-history tables that aren't navigable with assistive technology; charts and graphs with no text alternative; and identity-verification flows — a growing area given fraud controls — that break screen-reader compatibility entirely and can lock people out of their own accounts.

3.3 Insurance

Insurance-specific research is thinner than banking, but the pattern mirrors it closely: digital claims portals, policy documents and premium calculators are frequently built on the same complex form architectures and PDF-based documentation that drive banking's failures, without the volume of independent audit attention banking has received. Cross-sector eHealth and social-insurance research (Sweden's national eHealth portal 1177.se and its Social Insurance Agency website) found consistent accessibility and usability gaps reported directly by users with impairments — relevant here because claims, health and social-insurance portals share the same functional pattern: high-stakes, document-heavy, form-heavy journeys, exactly where WCAG's form-labelling and document-structure failures hit hardest. Worth flagging separately: legal and insurance commentary notes that standard business insurance — cyber, general liability, technology E&O — typically excludes or severely limits coverage for digital accessibility claims. Only some EPLI policies with third-party wrongful-acts coverage protect against them at all, meaning the sector best placed to price this risk isn't really underwriting it, for others or for itself.

3.4 Higher Education and Universities

Higher education sits inside the public sector accessibility regime in both the UK (PSBAR) and Ireland (WAD), but consistently self-reports weak delivery against strong stated intent. AbilityNet's Attitudes to Digital Accessibility survey (HE/FE cohort, n=201) found almost 80% consider accessibility a high organisational priority, but only 7% of senior management treat it as “very high priority” — the same resourcing gap that shows up sector-wide in Part II. In the same survey, 79% of HE/FE respondents had a published accessibility statement, but 13% acknowledged it was out of date and 18% didn't know its status. Comparable UK-wide EDUCAUSE research found 68% of institutions hadn't fully implemented web accessibility initiatives despite 71% of leaders calling it “very important,” with only 20% confident of meeting WCAG AA.

The structural problem specific to universities is architectural: institutions run decentralised digital estates spanning admissions portals, learning management systems, library databases, departmental micro-sites and third-party vendor tools, each procured and maintained separately — precisely the procurement gap identified as a cross-sector root cause in Part II.

Part IV — Human Impact by Access Need: Distinct, Sometimes Conflicting Requirements

A central finding across the literature is that “accessibility” isn't one requirement — it's several, and design choices that help one group can actively harm another. Dense information layouts help low-vision users scan efficiently but overwhelm neurodivergent users; heavy motion and animation aids engagement for some cognitive profiles but triggers vestibular or sensory distress in others. We've separated the evidence by access need below for exactly that reason.

4.1 Blind Users — Screen Readers

Blind users primarily depend on screen readers (JAWS, NVDA, VoiceOver, TalkBack) that convert visual interfaces into synthesised speech or refreshable braille output, navigating via headings, landmarks, labels and keyboard commands rather than a visual scan. The WebAIM Million data maps directly onto this population's lived experience: missing alt text (53.1% of pages in 2026) makes image-based navigation completely inaccessible; missing form labels (51.0%) block completion of banking, benefits and insurance forms; empty links and buttons present as unlabelled, meaningless stops when tabbing through a page; and skipped heading levels (39% of pages) destroy the structural navigation model screen reader users rely on to jump between sections rather than reading linearly. Banking research finds this population disproportionately forced into “sighted assistance” dependency — three-quarters of visually impaired online banking users needing a sighted person to complete tasks — precisely the independence loss digital accessibility law exists to prevent.

4.2 Low Vision — Magnification, Contrast and Dark Mode

Low vision is a functionally distinct access need from blindness and needs different accommodation entirely. Research on residual-vision users documents heavy reliance on screen magnification software (used by roughly two-thirds of low-vision participants in observed studies), browser zoom, full-screen colour filters, high-contrast themes and customised pointers — not screen readers. Low contrast text is now the single most common accessibility failure measured anywhere in this report (83.9% of pages in 2026), with WCAG's minimum 4.5:1 ratio routinely unmet. Magnification research finds reading speed for low-vision users depends critically on the width of the visible “viewing window”: below roughly 10 characters of visible text, reading speed degrades sharply, which means responsive layouts that reflow poorly under 200%+ zoom are a direct, measurable barrier, not a cosmetic one.

Dark mode and adjustable colour themes are increasingly treated as a low-vision accommodation in their own right — RNIB's own digital transformation explicitly built an optional dark mode to help people with low vision, distinct from and complementary to the screen-reader support it built for blind users. NHS low-vision service research notes that access to low-vision rehabilitation support itself is highly inconsistent across the UK, compounding the digital access gap with a service-access gap: over four in ten low-vision clinic attendees report symptoms of clinical depression, and 31% report rarely or never feeling optimistic about the future — a reminder that digital exclusion for this group sits on top of, not separate from, broader wellbeing and service-access disadvantage.

4.3 Deaf and Hard of Hearing Users — Captions and Visual Alternatives

An estimated 430 million people globally have disabling hearing loss (WHO). For this population the core barrier is audio content without a synchronised, accurate visual equivalent — and the research is unusually clear that “having captions” and “having usable captions” are two different things. Studies on auto-generated (ASR-based) captions find persistent accuracy and context problems that distort meaning, particularly for technical or fast-paced content. Effectiveness research goes beyond raw accuracy: caption positioning, speaker identification and avoiding occlusion of on-screen content all measurably affect usability, not just correctness. The EAA now mandates synchronised, accurate captions across broadcast and streaming content, and requires speaker identification support for sign language users in video communication services (ETSI EN 301 549 §6.5.6) — a meaningfully higher bar than “captions exist.”

4.4 Neurodivergent Users — Cognitive Load and Sensory Design

Neurodivergent people — encompassing autism, ADHD, dyslexia and related profiles — are estimated at 15–20% of the population and represent, per recent large-scale UX research, one of the fastest-growing categories of documented digital accessibility need, yet cognitive accessibility receives comparatively little dedicated legislative or technical attention relative to sensory and motor accessibility. Documented barriers include complex, cluttered layouts causing cognitive overload; dense, unbroken text blocks causing fatigue, particularly for ADHD and autistic users; bright colours, high-motion graphics and flashing content causing sensory distress for autistic users; and time-pressured interactions — session timeouts, countdown-based forms — disproportionately disadvantaging users with slower processing or executive-function differences, a directly relevant finding for banking and insurance forms specifically, which routinely use session timeouts.

Worth flagging: WCAG's own cognitive accessibility guidance — the W3C's “Making Content Usable for People with Cognitive and Learning Disabilities” — is itself criticised in the design literature as dense and difficult for the very population it's meant to serve. An unusually self-aware failure mode, and worth raising with any organisation treating WCAG compliance as sufficient for cognitive accessibility on its own.

4.5 Older Adults — Compounding Exclusion

Older adults are simultaneously the largest population experiencing age-related sensory and motor decline — vision, hearing, fine motor control, processing speed — and the largest population still offline altogether, which makes them a genuinely compounding case study rather than a single-axis one. In the UK, 21% of over-65s have no home internet access at all (Ofcom, 2025), compared with 2% of higher-income households. Age UK estimates 2.4 million older people are digitally excluded, with 33% not using a smartphone and 31% citing poor IT skills as a barrier to going online at all. Crucially, this isn't static: 920,000 older people (7%) reported reducing their internet use over the past year — evidence that even those who do get online don't reliably stay online as age-related decline progresses, a “leaky pipeline” the sector rarely designs for.

Eurostat's 2024 EU-wide data shows the same pattern by disability severity rather than age alone: 93.5% regular internet use among people without disability, falling to 86.0% for those with moderate activity limitation and 78.2% for those with severe limitation — with the largest relative gaps concentrated in learning and education-related internet use. Qualitative accounts collected by Age UK are strikingly consistent with the WCAG failure data in Part II: users report being unable to complete online GP appointment bookings, being asked to submit personal information through channels they neither trust nor can navigate, and — in the words of one respondent — feeling “rejected and excluded” by services that offer no non-digital alternative.

Part V — Corporate and Institutional Risk

5.1 Legal and Regulatory Risk — Including the Two Court Outcomes Most Briefings Are Missing

This is the section we most significantly rewrote, and it's worth being direct about why: an earlier version of this briefing stopped the enforcement timeline at “first lawsuits filed, November 2025” and left it there. By the time a July 2026 briefing is actually being read, that's not the current state of play — there have already been two court outcomes, and they point in opposite directions.

Auchan — dismissed, but not vindicated. The Auchan case was heard in the Tribunal judiciaire de Lille in May 2026 and dismissed. Critically, this wasn't a finding that Auchan's site was accessible — the court itself acknowledged that Auchan's e-commerce website doesn't conform to digital accessibility standards, which Auchan didn't dispute. The dismissal turned on procedural grounds specific to the emergency (référé) summary proceedings the case was brought under: the court found it couldn't establish the “obvious unlawful breach” needed for that fast-track relief, even while noting Auchan's “lack of interest” in accessibility as a major online retailer. The disability organisations behind the case — apiDV and Droit Pluriel — are appealing to the Court of Appeal of Douai.

Carrefour — the other outcome, and the one with teeth. On 4 June 2026, the Tribunal judiciaire de Caen ordered Carrefour France to make its e-commerce site and mobile app fully accessible, with six months to comply and daily fines accruing for delay. Carrefour hadn't disputed being subject to the obligation, and argued it already met 71% of the applicable criteria under France's RGAA standard. The court's response is the single most quotable line in either ruling: partial accessibility isn't enough — where a company measures itself against the RGAA, 100% of the applicable criteria must be met, “just as 100% of the steps on a staircase must be covered by a handrail.”

Read together, these two rulings tell a more interesting story than either one alone: courts are willing to acknowledge non-compliance even where they won't grant emergency relief on a technicality, and where a full hearing on the merits does happen, the standard being applied is full conformance, not “reasonable effort.” For any organisation weighing up whether EAA enforcement is “still mostly warning letters,” Carrefour is the answer: it isn't, any more.

Beyond France: under the EAA, fines now range from roughly €60,000 (Ireland) to approximately €900,000 (Sweden), with some national regimes empowered to fine up to 4% of annual revenue; national authorities can also order product withdrawal, ban non-compliant products from sale, mandate independent audits, and publicly name non-compliant organisations. In the UK, while there's no EAA-equivalent for domestic transactions, the genuine (GDS-sourced) escalation pathway set out in Part I still applies, and legal commentators continue to regard a landmark Equality Act website-accessibility ruling as a matter of timing rather than possibility. One further structural risk worth flagging directly: standard commercial insurance — cyber, general liability, technology E&O — typically doesn't cover digital accessibility claims, meaning organisations often discover this only after a claim has already been filed.

5.2 Commercial Risk: The Cost of Getting It Wrong

UK research quantifies lost revenue directly, and the numbers are large enough that “accessibility as compliance cost” starts to look like the wrong frame entirely. The Click-Away Pound surveys found that 69–71% of disabled online shoppers abandon a website they find difficult to use, representing an estimated £17.1 billion in lost UK revenue in 2019 (up from £11.75 billion in 2016). 83–86% of disabled consumers restrict their shopping entirely to sites they already know are accessible, and a similar majority say they'll pay more for a product on an accessible site rather than buy the same product cheaper on an inaccessible one. Only around 7% of disabled customers who hit a barrier ever contact the business to report it — meaning the overwhelming majority of this lost revenue is invisible in ordinary customer feedback channels and won't surface without proactive testing.

The combined “purple pound” — spending power of UK households with at least one disabled member — is estimated at approximately £249–274 billion annually, with banking and building societies alone estimated to represent roughly £935 million of at-risk annual spend from accessibility-driven customer churn. Accenture's Disability Equality Index analysis found that companies leading on disability inclusion generate materially higher shareholder returns and economic profit than peers who don't prioritise it — which reframes accessibility from a compliance cost to a demonstrable commercial variable, not just an ethical one.

5.3 Procurement and Vendor Risk

A recurring theme across UK and EU research is that accessibility failures are frequently inherited rather than created in-house: third-party platforms, learning management systems, CRM and vendor tools get procured without accessibility warranties or VPAT/ACR review, then become permanent structural liabilities that internal teams can't fully remediate. This is particularly acute in higher education (LMS platforms, research databases) and in regulated sectors relying on vendor-supplied claims, identity-verification or payment components. As public sector and large enterprise procurement increasingly writes WCAG 2.2 AA directly into supplier contracts, private-sector vendors — including UK vendors serving public-sector or EU clients — face indirect but material commercial pressure to reach the same bar even where the legal obligation itself is ambiguous.

Where This Actually Stands, in Plain Terms

  • Global conformance: 97.8% of homepages had a WCAG failure in 2019; six years of gradual improvement brought that to 94.8% in 2025; 2026 reversed it back up to 95.9%.

  • Page complexity: roughly doubled over seven years, and it's the primary correlate of the 2026 reversal.

  • Regulatory scope: public-sector-only in 2016/2018 (WAD/PSBAR) to full private-sector coverage under the EAA from June 2025.

  • Enforcement posture: warning letters and voluntary remediation, then the first court cases in 2025–2026 — one dismissed on a technicality, one resulting in a binding compliance order.

  • EU disability internet-use gap: a persistent ~15–17 percentage point gap, largest in Eastern Europe, smallest in Ireland, the Netherlands and Finland.

  • UK older-adult digital exclusion: essentially flat since 2022, at around 21% of over-65s still without home internet in 2025.

  • Corporate risk framing: shifting from a pure legal-compliance cost to a quantified commercial risk, via the Click-Away Pound and the Disability Equality Index.

Conclusion

The evidence base here is unusually mature and consistent: independent monitoring across the EU, UK and Ireland, academic research spanning every major disability category, and hard commercial data all point the same direction from different angles. Digital accessibility law has expanded in scope and enforcement intensity faster than actual conformance has improved. The failures are concentrated in a small, well-documented set of technical issues — contrast, labelling, alt text, form structure, document tagging — that are inexpensive to fix individually but pervasive because of governance, procurement and prioritisation gaps rather than genuine technical difficulty. And the human cost isn't one uniform experience: it's at least five distinct sets of requirements — blind/screen-reader, low vision/magnification-contrast, Deaf/HoH/captions, neurodivergent/cognitive load, older-adult/compounding exclusion — that any credible organisational accessibility strategy needs to treat as separate design problems, not one checklist.

And on the point we corrected most substantially here: the enforcement landscape isn't static, and neither is this briefing's job. Auchan and Carrefour won't be the last EU rulings on this, and we'll keep this updated as the picture develops rather than let it calcify around whatever the state of play happened to be on the day it was first written.

References and Sources

We've grouped these by type rather than alphabetically, because where a claim comes from matters as much as what it says. Primary regulatory and monitoring sources are listed first; independent academic and NGO research second; industry and vendor commentary — used sparingly, and never as the sole support for a figure that has a primary source — last, and labelled as such.

Primary regulatory and monitoring sources

  • European Commission — European Accessibility Act (Directive (EU) 2019/882) and Web Accessibility Directive (Directive (EU) 2016/2102), commission.europa.eu

  • GOV.UK / Government Digital Service — Accessibility monitoring of public sector websites and mobile apps from 2022 to 2024 (December 2024)

  • WebAIM — The WebAIM Million: 2024, 2025 and 2026 reports on the accessibility of the top 1,000,000 home pages, webaim.org/projects/million

  • National Disability Authority (Ireland) — Annual Monitoring Reports, EU Web Accessibility Directive (2021–2025)

  • Central Statistics Office (Ireland) — Census of Population 2022, Profile 4: Disability, Health and Carers

  • Eurostat — Disability statistics: access to information and communication technologies (2024/2025 releases)

  • Ofcom — Adults' Media Use and Attitudes Report (2025); digital exclusion research

  • World Health Organization — Hearing loss global prevalence estimates (2025)

  • Government Analysis Function (Civil Service) — Accessibility legislation: what you need to know; Stephen Campbell / HSCNI case reference

  • Tribunal judiciaire de Lille (May 2026) and Tribunal judiciaire de Caen (June 2026) — EAA-linked rulings, Auchan and Carrefour, as reported by Droit Pluriel and apiDV

Independent academic and NGO research

  • AbilityNet — Attitudes to Digital Accessibility survey; Higher Education/FE PSBAR insights (2023–2025)

  • Watson, Parnaby & Kharrufa, Open Lab, Newcastle University — Precarious Experiences: Citizens' Frustrations, Anxieties and Burdens of an Online Welfare Benefit System

  • Scope / Business Disability Forum / Extra Costs Commission — Click-Away Pound Survey (2016, 2019); Walk-Away Pound report

  • UK Parliament, Women and Equalities Committee — Accessibility of products and services to disabled people (report)

  • ACM Digital Government: Research and Practice — Accessibility Analysis of Educational Websites Using WCAG 2.0

  • EDUCAUSE Center for Analysis and Research (ECAR) — Web Accessibility in Higher Education statistics

  • Oxford Academic, Interacting with Computers — Designing assistive technologies for and with neurodivergent users (2025)

  • Fathallah et al. — Empowering the Deaf and Hard of Hearing Community: Enhancing Video Captions Using Large Language Models (ACM/arXiv, 2024–2025)

  • RNIB — Research and data hub; computer accessibility guidance; Torchbox case study on RNIB digital transformation (dark mode)

  • PMC / NCBI — Low Vision Services Provision throughout NHS Trusts in the UK; disability digital divide research (Sweden, SSIA/1177.se)

  • Age UK — Facts and figures about digital inclusion and older people (July 2025)

  • Centre for Ageing Better — The State of Ageing 2025

  • Scottish Parliament SPICe — Pensioner Poverty and Digital Exclusion (2025)

Industry and vendor commentary (context only, not used as sole evidence for any figure with a primary source)

  • Deque — France's major court decision supporting digital accessibility under the EAA (2026) — used for the Auchan/Carrefour case details, cross-checked against Level Access and primary French-court reporting

  • Level Access — EAA Compliance in 2026: How Enforcement Has Evolved (2026) — cross-checked against Deque and primary sources for the same case details

  • Datos Insights — The Accessibility Challenge in Digital Banking

  • UsableNet / Recite Me — banking accessibility case studies, cited only where the underlying statistic is also independently attributable

  • Accessibility.Works — higher education compliance commentary; insurance coverage commentary

A closing note on sourcing: we've deliberately thinned the industry-commentary section compared to an earlier draft, because several of those sources are themselves accessibility vendors with a commercial interest in the scale of non-conformance and enforcement urgency looking as large as possible. Where a figure in this briefing has a primary source, that's what's cited; vendor commentary appears only where it added case-level detail (like the French court rulings) that we then verified against more than one independent account.