The EU Accessibility Baseline · Edition 1 · July 2026
A year after the European Accessibility Act took effect, 9 in 10 mid-market consumer sites still fail the automated floor.
We scanned 278 public pages across 187 European consumer websites against the WCAG 2.2 AA automated ruleset, the technical floor beneath the EAA's harmonised standard. This page publishes the results, the league tables and the full methodology. No company is named. Every number is reproducible.
- 91%of 160 mid-market homepages carry at least one WCAG 2.2 AA violation
- 97%of 99 shop-path pages (category, product, cart, booking) carry violations
- 90%of sites carry at least one violation classed critical or serious
- 11,219individual page elements affected across the dataset
The mid-market picture
The cohort: 160 scannable homepages of consumer-facing, mid-sized businesses trading in and into the EU, across ten countries and eight sectors. These are ordinary shops, brands and services, the companies the EAA actually regulates, not the giants with in-house accessibility teams.
146 of 160 homepages (91%) fail at least one automated WCAG 2.2 AA check. 144 (90%) carry at least one failure classed critical or serious, the classes most likely to block a real user. The average homepage carries 3.3 distinct violation types; the worst carries 13.
| Sector | Sites | With violations | Avg violations per homepage |
|---|---|---|---|
| General e-commerce | 27 | 96% | 4.0 |
| Fashion | 39 | 95% | 3.8 |
| Home and furniture | 32 | 91% | 3.5 |
| Beauty | 33 | 91% | 2.7 |
| Booking services * | 7 | 100% | 3.0 |
| Travel * | 7 | 86% | 3.4 |
| Consumer SaaS * | 8 | 75% | 2.1 |
| Ticketing * | 7 | 71% | 1.1 |
* Small sample (under 10 sites): treat as indicative, not representative. Larger sector samples arrive in Edition 2.
| Country | Sites | With violations | Avg violations per homepage |
|---|---|---|---|
| Netherlands | 20 | 100% | 5.0 |
| Norway * | 8 | 100% | 3.1 |
| France | 24 | 96% | 3.3 |
| United Kingdom | 22 | 95% | 3.4 |
| Germany | 37 | 92% | 2.9 |
| Denmark | 11 | 91% | 3.5 |
| Sweden * | 8 | 88% | 2.0 |
| Ireland | 16 | 81% | 3.1 |
| Finland * | 5 | 80% | 2.8 |
| Belgium * | 9 | 67% | 2.7 |
* Small sample (under 10 sites). UK and Norwegian sites are included because the EAA follows the customer, not the company register: a business outside the EU serving EU consumers is in scope for that service.
The closer to the money, the worse it gets
For 99 of the cohort sites we also scanned a transactional page: a category, product, cart or booking page. 97% carry violations, against 91% for homepages, and they average 4.2 distinct violation types against the homepage's 3.3. The pages where customers actually spend money are measurably less accessible than the shop window.
What actually fails
Six failure types dominate the mid-market cohort. Share of the 160 homepages affected:
| Failure | Sites affected | What it means for a real person |
|---|---|---|
| Insufficient colour contrast | 64% | Text is hard or impossible to read for low-vision users, and for anyone in sunlight |
| Links without accessible names | 49% | A screen reader announces "link" with no clue where it goes |
| Touch targets too small | 30% | Buttons and links too small or cramped to tap reliably, on precisely the devices most shopping happens on |
| Buttons without accessible names | 27% | Critical controls (menu, search, add to cart) are announced as "button", nothing more |
| Images without alternative text | 26% | Product photos and banners are silent to screen-reader users |
| Zoom disabled | 17% | The page actively blocks pinch-zoom, which many people need simply to read |
None of these is exotic. Most are days of development work, not months. Which is rather the point: the automated floor is the cheap part, and 9 in 10 sites have not cleared it.
Flagship sites do better, but not well
A comparison cohort of 19 scannable flagship European consumer sites (large fashion, electronics, marketplace, grocery and travel brands): 95% carry at least one AA violation, averaging 2.4 violation types per homepage against the mid-market's 3.3. Europe's best-resourced digital teams carry roughly a quarter fewer automated failures than the mid-market, and still almost none of them clears the floor.
The overlay finding. One flagship cohort site runs a third-party "accessibility overlay" widget while carrying six violations, two of them critical. The overlay's own interface element fails the WCAG 2.2 touch-target rule. Overlays do not produce conformance, and we do not sell them.
What this means if you sell to EU consumers
The European Accessibility Act has applied to consumer-facing digital services since 28 June 2025. Micro-enterprises (under 10 staff and under EUR 2 million turnover) are exempt for services; everyone else is in scope, wherever the company is registered. Enforcement is national: Germany's BFSG carries fines up to EUR 100,000 and the power to prohibit a non-compliant service; the Netherlands' regulator can fine a webshop up to EUR 900,000 or 1% of turnover; in Ireland non-compliance is a criminal offence. No headline EAA fine against an ordinary mid-market shop has been publicly confirmed yet, and we will not pretend otherwise. What is real: warning letters circulate in Germany, a French court has already ruled on digital accessibility this year, procurement teams increasingly demand accessibility statements, and remediation takes months. The floor measured on this page is the part you can find and fix cheaply, before any of that reaches you.
Methodology, in full
Engine and ruleset
axe-core (the industry-standard open-source rules engine, deterministic) via Playwright-driven Chromium, rule tags wcag2a, wcag2aa, wcag21aa, wcag22aa. One polite browser load per page, desktop viewport, July 2026. No logged-in areas, no interaction flows, robots directives respected.
Cohorts
Flagship: 30 large consumer brands (ES, DE, FR, IT, NL and pan-EU), homepage only; 19 scannable. Mid-market: 180 consumer-facing mid-sized businesses across DE, FR, UK, NL, IE, DK, BE, NO, SE and FI, sourced from public business directories for commercial prospecting and screened only for being consumer-facing, mid-sized and trading in or into the EU; scanning happened after selection, so results could not influence cohort membership; 160 homepages scannable, plus 99 scannable transactional pages. Statistics are computed over every scannable page, never over a filtered subset.
Exclusions
Pages are excluded, and removed from denominators entirely, when: the fetch failed (9 pages); the URL turned out to be a static asset picked up by our page-discovery tooling (9); a bot-protection interstitial was served instead of the real page (23); or the server returned an error page (4). We scan what a real browser was actually shown; when that was a Cloudflare wall, it does not count against anyone.
Honest limitations
Automated checks detect only a minority of real accessibility barriers; industry consensus puts it at roughly 30 to 40%. Manual testing finds the rest. Every number on this page is therefore a floor, not a ceiling: a site passing the automated ruleset is not thereby conformant, and full conformance is assessed against EN 301 549, not by any scanner. That includes ours.
Naming policy
No company is named, positively or negatively. Sectors, countries and aggregates only. Cohort site lists are retained internally for reproducibility and re-scanning.
Next edition
Edition 2 (quarterly) adds re-scan deltas against this baseline, larger sector samples, and anonymised aggregates from audited user journeys. Corrections, methodology challenges and requests to verify a number: hello@able.cat. If we got something wrong we will say so on this page.