Daniel Butnar Web accessibility

42 shop homepages, one scanner, one afternoon: 37 fail

On 15 September 2026 I ran the same automated WCAG 2.1 AA check on the homepages of 44 well-known online shops and service providers in Romania and Germany. Two blocked automated access. Of the 42 that answered, 37 have at least one detectable failure. This is roughly what a market surveillance scanner sees.

Key numbers

  • 37 of 42homepages fail at least one WCAG 2.1 A or AA rule, 88 %
  • 24 of 24Romanian sites fail, with 77 failing elements per page on average (median 68)
  • 13 of 18German sites fail, with 13 failing elements per page on average (median 6)
  • 21 of 42have at least one critical issue, the kind that blocks a screen-reader user outright

What fails most often

2,080 failing elements were found across the 42 pages. Four rule types account for 1,819 of them, 87 %. Each bar shows on how many of the 42 homepages the rule failed at least once.

These are the same failure types that WebAIM finds year after year on the top million homepages worldwide. None of them needs a redesign. All of them need someone to go through the templates.

  • Text contrast below 4.5:122 sites, 1,073 elements
  • Links without text (image and icon links)20 sites, 580 elements
  • Images without alternative text12 sites, 89 elements
  • Buttons without a name9 sites, 77 elements
  • List markup misused5 sites, 61 elements
  • Zoom disabled in the viewport tag5 sites
  • Hidden from screen readers but keyboard-focusable4 sites, 41 elements
  • Invalid ARIA attributes4 sites, 13 elements

Six observations

  1. Germany and Romania are a year apart. Five German sites, all large brands in fashion, telecom, rail, banking and a general retailer, pass the automated check completely. No Romanian site does. Mid-size shops in both countries look alike in the kinds of failures; the Romanian ones just have many more of them.
  2. Passing the scanner is not conformance. Automated tools detect roughly a third of WCAG failures. The five clean German sites still need a keyboard, screen-reader and zoom pass before anyone should publish a statement claiming conformance.
  3. Contrast is a design-token problem. 1,073 low-contrast elements sounds like a lot. On most sites they come from three to six colour pairs defined once in the theme. Fix the tokens and the count drops to zero.
  4. Links without text are banners. Almost every one of the 580 empty links is a promotional image link or an icon link without alternative text. That is a template fix, not a content fix.
  5. Five sites lock out zoom. A single meta tag with user-scalable=0 stops people from enlarging the page on their phone. It is a five-minute fix and a failure of WCAG 1.4.4 that affects everyone with reading glasses.
  6. Keyboard users land in the cookie banner. Only 11 of the 42 homepages offer a skip link. On several sites the first thing a keyboard user reaches is the consent dialog, and on some it is a link with no text at all.

Method

What was tested, and how
  • Sample: 44 homepages of well-known online shops and services, 24 in Romania and 20 in Germany, across retail sectors that the European Accessibility Act covers (electronics, fashion, DIY, books, pharmacy, furniture, toys, pets, beauty) plus telecom, banking and transport. Two German sites blocked automated access and are excluded.
  • Tool: axe-core 4.13 with the WCAG 2.0 and 2.1 level A and AA rule set only, no best-practice rules. Driven by Playwright in Google Chrome on a 1366 by 768 desktop viewport.
  • Cookie banners: where a “reject” or “accept” button could be found it was clicked so that the page behind the banner was scanned as well. Where it could not, the page was scanned as it landed.
  • Date: 15 September 2026, one run per site.
  • Counting: “failed rules” is the number of distinct axe rules with at least one failure; “failing elements” is the number of HTML elements that failed. Critical and serious follow axe's impact scale.
Limitations
  • Automated checks find only part of the barriers, commonly estimated at 30 to 40 %. Keyboard traps, focus order, screen-reader flows and understandable error messages need a person. All numbers here are lower bounds.
  • Only the homepage was scanned, on desktop. Product pages and checkouts usually contain more forms and more failures, not fewer.
  • A site with zero automated failures is not thereby conformant; a site with 300 failing elements may need only a handful of fixes if they come from one template.
Why the sites are anonymised

The point is the pattern, not the pillory. Sites are identified by an ID, country, sector and size. I keep the named results and share a site's own row, with the full list of findings, free of charge with the people who run it. If you think your shop is in the table, ask me for your row.

All 42 sites

Sorted by ID. “Most frequent issue” is the rule with the most failing elements on that page. “h1” counts level-one headings; a page should have exactly one.

Automated WCAG 2.1 AA results for 42 homepages
ID Country Sector Size Failed rules Failing elements Critical Serious Most frequent issue Skip link h1
DE-01 Germany general retail large 4 28 14 14 images without alt no 4
DE-02 Germany electronics large 6 66 15 51 list structure yes 1
DE-03 Germany musical instruments large 4 46 25 21 images without alt yes 1
DE-06 Germany DIY and garden large 1 10 0 10 hidden but focusable no 0
DE-07 Germany fashion large 0 0 0 0 none yes 1
DE-08 Germany IT and computers mid-size 4 13 8 5 images without alt no 1
DE-09 Germany IT and computers mid-size 1 1 1 0 buttons without name no 1
DE-10 Germany electronics large 2 20 1 19 text contrast no 1
DE-11 Germany car parts and tyres mid-size 2 4 0 4 icons without name no 1
DE-12 Germany outdoor mid-size 6 18 3 15 nested controls no 0
DE-13 Germany car parts and tyres mid-size 1 1 0 1 links without text no 1
DE-14 Germany fashion mid-size 0 0 0 0 none yes 1
DE-15 Germany toys mid-size 3 8 6 2 images without alt no 0
DE-16 Germany telecom large 0 0 0 0 none no 1
DE-17 Germany transport large 0 0 0 0 none no 1
DE-18 Germany transport large 4 13 3 10 images without alt yes 1
DE-19 Germany banking large 0 0 0 0 none yes 1
DE-20 Germany pharmacy large 1 1 0 1 frame without title no 1
RO-01 Romania marketplace large 1 139 0 139 text contrast no 0
RO-02 Romania electronics large 6 152 57 95 links without text no 1
RO-03 Romania fashion large 1 171 0 171 text contrast no 24
RO-04 Romania DIY and garden large 2 131 0 131 text contrast no 0
RO-05 Romania general retail mid-size 3 313 1 312 links without text no 1
RO-06 Romania electronics large 6 95 12 83 list structure no 0
RO-07 Romania electronics mid-size 3 85 5 80 text contrast no 0
RO-08 Romania electronics mid-size 5 79 10 69 text contrast no 0
RO-09 Romania books mid-size 3 8 0 8 text contrast no 0
RO-10 Romania IT and computers mid-size 1 84 0 84 text contrast no 0
RO-11 Romania books mid-size 7 176 19 156 text contrast no 0
RO-12 Romania fashion mid-size 1 3 0 3 list structure no 0
RO-13 Romania beauty large 4 76 0 76 links without text yes 0
RO-14 Romania pharmacy mid-size 1 2 0 2 invalid ARIA no 1
RO-15 Romania pharmacy mid-size 5 22 4 18 links without text no 0
RO-16 Romania furniture mid-size 3 39 10 29 text contrast yes 5
RO-17 Romania toys mid-size 4 21 1 19 text contrast no 0
RO-18 Romania pet supplies mid-size 3 131 1 130 text contrast yes 1
RO-19 Romania telecom large 2 26 7 19 links without text no 1
RO-20 Romania telecom large 4 10 0 9 list structure no 0
RO-21 Romania banking large 2 10 0 9 text contrast yes 1
RO-22 Romania transport mid-size 6 59 5 53 links without text no 1
RO-23 Romania fashion mid-size 2 3 0 3 hidden but focusable no 0
RO-24 Romania beauty large 1 16 0 16 links without text yes 1

Want the numbers for your own shop?

The quick check runs this scan on your home, product and checkout pages, adds a manual keyboard and screen-reader pass, and gives you a prioritised list within three working days.

Book a quick check, 490 € See a sample report