Brackenmark

Website accessibility gap assessments · Ireland

The Irish E-commerce Accessibility Study

Nineteen Irish online shops. Not one of them passed.

Their median automated accessibility score was 92 out of 100 — a figure most people would read as a pass. Every single site still had failures against WCAG 2.1 Level AA, the standard the EU rules point at.

That gap is the reason this study exists. A score in the nineties tells you that most of what a machine knows how to check is fine. It is a different question from whether a customer using a screen reader can complete a purchase.

92/100

median automated accessibility score, across 19 sites

0

sites with no WCAG 2.1 AA failures

How this was done

The sites and the method were decided before anything was scanned, and written down, so the findings can't be shaped to suit a conclusion after the fact.

Who was included Irish-owned or Irish-operated shops selling to consumers online, large enough to sit outside the microenterprise exemption, spread across retail categories. Chosen without looking at their accessibility first.
Pages per site Six, chosen the same way every time — home, category, product, cart, contact and a content page. Identical for every site, which is what makes them comparable.
Standard WCAG 2.1 Level AA.
Tools axe-core and Lighthouse, running in a real browser. Automated and rule-based throughout — no judgement calls, no AI deciding what counts as a failure.
When August 2026.
Checkout Not scanned. On hosted platforms it runs on a separate domain the retailer can't edit, and we don't put automated traffic through live payment flows.

What these numbers are not. Automated testing reliably catches somewhere between 30% and 50% of WCAG issues. Everything below is therefore a floor, not a total. The real figure for every site in this study — including our own — is higher than what's published here.

What was found

Reported as prevalence — how many of the nineteen sites had each fault — rather than as a total count of problems. Raw totals scale with how large a site's pages are, not with how well it was built, so a shop with a big mega-menu would always look worse than a small one even when it's better made.

Only failures that map to a WCAG 2.1 A or AA success criterion are listed. Best-practice warnings, which the tools also produce, are excluded.

Text below minimum contrast 1.4.3
18 of 19
Links with no discernible name 2.4.4
15 of 19
Headings out of order 1.3.1
13 of 19
Images with no alt text 1.1.1
9 of 19
Visible label doesn't match the accessible name 2.5.3
9 of 19
Buttons with no accessible name 4.1.2
9 of 19
Broken list markup 1.3.1
7 of 19
Dropdowns with no label 4.1.2
6 of 19
Form fields with no label 4.1.2
6 of 19

The spread

Medians rather than averages throughout — a single outlier with hundreds of contrast failures would drag a mean into nonsense.

Automated score Median 92 out of 100. Lowest 74, highest 100.
Distinct fault types per site Median 10. Lowest 4, highest 22.
WCAG criteria affected per site Median 6. Lowest 4, highest 8.
Contrast failures Median 41 instances on the six pages checked. Highest 470.
Sites with no AA failures None.

Six of the nineteen had at least one page score a perfect 100 on the automated accessibility score while still failing Level AA elsewhere on the site. That is the most useful fact in this study, and the reason a score on its own should never be treated as an answer.

Seven shops we could not assess

Twenty-six businesses were selected. Nineteen produced a complete, comparable scan. The other seven are reported here rather than quietly dropped, because how a site fails to be assessed is itself a finding.

  1. Three blocked the scanner outright

    Their security layer returned a "checking your browser" page instead of the site. Nothing is wrong with those shops — it's a firewall doing its job — but it does mean no automated accessibility tool can see them, including the ones their own developers might use.

  2. Three exposed almost no navigable links

    On these, only one or two pages could be reached at all: the site returned hundreds of link elements, but almost none pointed anywhere a browser could follow without running scripts. That pattern is worth a closer look on its own merits, since anything that can't be followed by a crawler often can't be followed by a keyboard either.

  3. One could not be reached

    The site did not respond at the time of the study.

Our own site, scanned the same way

It would be poor form to run this and exempt ourselves. brackenmark.com was put through the same tool against the same standard: no WCAG 2.1 AA failures, and one best-practice warning from axe-core.

That result deserves an honest asterisk rather than a victory lap. This is a static one-page site with no product catalogue, no filters, no basket and no third-party apps. It is a far easier thing to get right than a shop with forty thousand products, and comparing the two would be dishonest. The reason we publish it is that a company selling accessibility assessments should be willing to be measured by its own instrument.

What we're not doing

No business is named, and none will be. There is no league table here and no scores out of a hundred attached to anyone. Every shop in this study is a normal, well-run Irish business, and almost none of these faults arrived through carelessness — they come in through themes, third-party apps and years of ordinary content edits. Most are a small number of changes at theme level that clear the problem across every page at once.

Each participating business is being sent its own results privately. Aggregate in public, specific in private. That was the rule before the first scan ran, and it doesn't bend.

If you want to check your own

You can get the same kind of score for free in about a minute: open Chrome DevTools, go to the Lighthouse tab, tick Accessibility and run it. That's worth doing.

Just be careful what you conclude from it. Every site in this study would have passed that test by most people's reading, and none of them actually did.

The complete sample report is published in full, so you can judge the quality before spending anything.