A 20-minute self-assessment to check if your website meets EAA standards: automated scanning, keyboard, and content checks, including issues that automated tools miss.
In 20 minutes, you can find out if your website has an accessibility problem, not whether it is EAA-compliant. The test has three passes: an automated one with axe or WAVE, a manual one using the keyboard and zoom, and a content pass. Automated tools detect 57% of issues according to Deque; the rest can only be found manually.
We designed this test so that someone on your team without prior accessibility training can perform it. It does not replace an audit. It tells you if you have an emergency, not if you are in full compliance.
What this test does and what it does not
An automated accessibility scan detects 57% of issues, according to a Deque study covering over 13,000 pages and 300,000 issues across more than 2,000 audits (deque.com/automated-accessibility-testing-coverage, page last modified 07-10-2026, accessed 08-27-2026). Deque sells automated tools and still publishes that ceiling: the figure provided by the manufacturer is the optimistic maximum, not the floor.
The remaining 43% requires a person. And that 43% is where almost everything a user truly notices lives.
The industry norm: Many firms sell "accessibility tests" as a single button that returns a score out of 100. That score measures only what a machine knows how to measure, and a machine does not know if your alt text describes the image or just says "image1.jpg".
Our criteria: We separate the three passes and identify which is which. The automated pass runs in five minutes; the manual pass is where the work lies. If a report does not distinguish between the two, it is selling you coverage you do not actually have.
Minutes 0 to 2: check if the law applies to you
Before looking at your website, look at your business profile. Law 11/2023 of May 8 (BOE-A-2023-11022) transposes Directive (EU) 2019/882 and has applied accessibility requirements since June 28, 2025.
Three concrete checks, with your legal documentation at hand:
- Covered services. The law covers e-commerce, consumer banking services, passenger transport, electronic communications, e-books, and audiovisual services. A corporate website that only provides information, without sales or contracts, falls outside the scope of the norm.
- Micro-enterprises providing services. Fewer than 10 employees and an annual turnover or balance sheet not exceeding 2 million euros. Micro-enterprises that provide services are excluded from the obligations. The exception applies to services, not products.
- Prior contracts. The transitional provision allows service contracts signed before June 28, 2025, to remain unchanged until June 28, 2030, or until they expire.
The penalty regime is not derived from Law 11/2023 alone: infringements regarding equal opportunities and accessibility are sanctioned under Title III of Legislative Royal Decree 1/2013, with fines ranging from 301 to 1,000,000 euros depending on the severity (Legislative Royal Decree 1/2013, Title III, arts. 79 to 95 · BOE consolidated text · accessed 08-27-2026). The high end of this range corresponds to severe repeated offenses, not a poor contrast on a button. Anyone who shows you "fines of up to a million" without explaining that context is selling you fear, not compliance.
Minutes 2 to 7: the automated pass
Choose a different template page each time. Home, product or service page, contact form, and the checkout step if you sell online. Four pages, not forty: website failures are template failures.
| Tool | Cost | Technology | Best for |
|---|---|---|---|
| axe DevTools (extension) | Basic version free | axe-core | Technical list with the element highlighted |
| WAVE (WebAIM) | Free | Own engine | Seeing errors on the page, useful for non-technical users |
| Lighthouse (Chrome DevTools) | Included in browser | axe-core (subset) | A quick score, insufficient as a report |
| Pa11y CI | Free, command line | axe-core / HTML_CodeSniffer | Integrating into CI to break the build |
Lighthouse might score 100 on pages with serious keyboard failures. This is a consequence of its scope, not a defect: it only runs a subset of rules. Treat it as a thermometer, not a certificate.
The six failures you will find are almost always the same. In the WebAIM Million report from February 2026, covering the million most visited homepages, 95.9% had automatically detectable WCAG 2 failures, with an average of 56.1 errors per page. Breakdown: low text contrast 83.9%, missing alternative text 53.1%, missing form labels 51%, empty links 46.3%, empty buttons 30.6%, and undeclared document language 13.5%. Those six account for 96% of everything detected.
Practical translation: if your website fails, it is almost certainly failing in the same ways as everyone else's, and almost all of it is cheap to fix.
Minutes 7 to 14: keyboard, focus, and zoom
Here begins what no tool does for you. Put your mouse in a drawer.
- Tab from the start: check if a "skip to main content" link exists and appears upon the first Tab.
- Focus is always visible: a visible outline on any background, not a blue halo that disappears in a dark section.
- Tab order follows the visual order of the page, without jumping from the footer to the menu.
- Dropdown menus open with Enter or the Space bar and can be navigated with arrow keys.
- Cookie banners can be accepted and, most importantly, rejected using only the keyboard.
- You can exit a modal using Escape, and the focus returns to the element that opened it, not the top of the document.
- Carousels or videos can be paused; nothing moves for more than five seconds without control.
- Browser zoom at 200% and window width at 320px: no horizontal scrollbar appears and text does not overlap.
- With zoom enabled, the "Buy" button is still reachable without side-scrolling.
If the focus is not visible, the website is not accessible, regardless of how many alt tags you have. It is the most frequent failure we find and the most invisible to those who have never stopped using a mouse.
Minutes 14 to 18: content and semantics
This pass is done by reading, not by executing.
- A single H1 per page and headings in order, without jumping from H2 to H4 due to font size.
- Alt tags describe the function of the image, not the filename; decorative images have empty alt attributes.
- Links indicate where they go: "View price breakdown" instead of "Read more" repeated nine times.
- Every form field has a visible label, not just placeholder text.
- Form errors are explained with text: "The ID number must have 8 digits and one letter" rather than just a red border.
- Videos with audio have actual subtitles, not unreviewed auto-generated ones.
- Language is declared in the HTML and changes in sections with different languages.
- No instruction relies solely on color: "fields in red" provides no guidance to someone who cannot distinguish red.
The list of what no automated tool detects
This is the part that other sites positioning themselves on this topic do not write. Each row is a standard criterion that a scanner either executes or flags as "manual review" and never resolves.
| What the machine cannot judge | Criterion | What fails in practice |
|---|---|---|
| Whether the alt text describes the image | 1.1.1 | The tool sees an alt tag; it does not see that it says "banner-home-v3" |
| Whether the tab order makes sense | 2.4.3 | Grid-based CSS reorders visually, causing focus to jump randomly |
| Whether the visible focus is sufficient | 2.4.7 and 2.4.11 | Browser outline against a background of the same tone |
| Whether the error message is helpful | 3.3.3 | "Invalid field" passes technical validation but helps no one |
| Whether subtitles are accurate | 1.2.2 | The system transcribes "Bizum" as "visum" |
| Whether the heading describes the section | 2.4.6 | Perfect structure, but titles that say nothing |
| Whether an equivalent alternative exists | 1.1.1 | A scanned PDF uploaded as "accessible conditions" |
| Whether dynamic content is announced | 4.1.3 | The "product added" notice exists but does not reach the screen reader |
| Whether gestures have alternatives | 2.5.1 | A map that only responds to pinch-zoom |
| Whether the entire process works | 3.2.2 and 2.5.8 | Each step works individually, but the purchase cannot be finished via keyboard |
That last one is the deciding factor. No one uses a page: people complete a process. And the process is exactly what no tool traverses.
Minutes 18 to 20: the results sheet
Note down four numbers and date them. Without a date, it is useless, as the website changes.
| Item | Figure | Note |
|---|---|---|
| Total automated errors (axe or WAVE) | — | — |
| Keyboard blocks found | — | — |
| Processes that cannot be completed without a mouse | — | — |
| Actual test minutes | — | — |
— = data being measured; we will publish it once we have our own sample.
Decision rule we use: a keyboard block in a purchase or contracting process is a stop-work, not an improvement point. It gets fixed this week. A low-contrast footer enters the normal cycle.
Verified 08-27-2026. There is a real contradiction in the market worth noting. A large portion of Spanish websites claim that the EAA mandates WCAG 2.2 Level AA. The harmonized standard cited for ICT accessibility in the EU remains EN 301 549 V3.2.1 (March 2021), which incorporates WCAG 2.1 AA. The next version, V4.1.0, with WCAG 2.2 AA, had a draft published on 11-12-2025 and as of 06-24-2026 was in the EN approval procedure; as of 08-27-2026, there is still no final publication (list of versions at etsi.org/deliver/etsi_en/301500_301599/301549/, accessed 08-27-2026). Design against WCAG 2.2 AA because it is a superset of 2.1 and saves you from redoing work, but do not accept a report that labels a new 2.2 criterion as a legal violation without stating which version of the standard it applies.
A serious report works against the published version of the standard, not against the version the report assumes.
Accessibility is not a layer added at the end: it is a consequence of how the grid, focus, and document order are built. That is why this test detects symptoms and does not cure anything on its own.
Run it this week with your team. If it comes out clean, you do not need us, and we will tell you so. If a keyboard block appears in your checkout process, write to us that same day; the rest can wait for the normal cycle.
Frequently asked questions
If I run axe and get zero errors, am I EAA compliant?
No. Zero automated errors means you have solved the mechanical part of the 57% that the tool measures. Compliance is declared against the complete criteria, and half of them can only be checked by hand.
Does an accessibility overlay or plugin help pass this test?
You will pass some automated rules and still be non-compliant. An overlay does not fix tab order or write meaningful alt text, and it adds a layer that often breaks screen readers that were already working.
I am a freelancer with an informative website, no store. Does the law force me?
If you do not provide any of the services covered by Law 11/2023, the obligations of that law do not apply to you. Do it anyway for the low-hanging fruit: contrast, labels, and keyboard support take a few hours and affect customers who are currently leaving without telling you.
How often should the test be repeated?
Every time a template changes and, at a minimum, once per quarter. We set up Pa11y in our CI, so a regression breaks the build rather than being discovered through a complaint.
What should I ask the person who built my website?
The automated report for the main templates, the result of the keyboard pass, and the version of the standard they worked against. If those three things do not exist in writing, no accessibility work was performed.
Keep reading
- Web accessibility in Spain after the European Accessibility Act: who it covers, what it requires, and how to check if your website complies
- Accessibility overlays do not make you EAA-compliant: the evidence
- Downloadable EAA checklist to run the test with your team
- 47-point checklist before putting a website into production
- 18 questions to ask an agency before signing
Transparency note. Some of our own data in this guide is still being measured on real projects, with a single method. We will publish it with its sample size and date range once the sample is large enough; until then, every figure you see carries its source and date of consultation.
Sources
- https://www.boe.es/buscar/act.php?id=BOE-A-2023-11022 — Law 11/2023, of May 8: transposition of Directive (EU) 2019/882 (EAA); application date 06-28-2025, micro-enterprise service exclusion (<10 employees and ≤2M €), transitional contract provision until 06-28-2030, covered services
- https://www.dsca.gob.es/es/derechos-sociales/discapacidad/derechos-personas-discapacidad/infracciones-y-sanciones — Ministry of Social Rights: Title III infringement and sanction regime of RDL 1/2013, range from €301 to €1,000,000, limitation periods
- https://www.deque.com/blog/automated-testing-study-identifies-57-percent-of-digital-accessibility-issues/ — Deque, 03-10-2021: automated testing identifies 57% of issues across over 13,000 pages, ~300,000 issues, and 2,000+ audits; remaining 43% requires manual testing
- https://webaim.org/projects/million/ — WebAIM Million, February 2026: 95.9% of homepages with detectable WCAG 2 failures, average of 56.1 errors per page, and the six dominant failures with their percentages (contrast 83.9%, alt 53.1%, form labels 51%, empty links 46.3%, empty buttons 30.6%, language 13.5%)
- https://digital-strategy.ec.europa.eu/en/policies/latest-changes-accessibility-standard — European Commission: EN 301 549 V3.2.1 is the harmonized version cited in the OJEU since August 2021, based on WCAG 2.1
- https://en.wikipedia.org/wiki/EN_301_549 — EN 301 549 version history: V2.1.2 (2018), V3.2.1 (2021, WCAG 2.1), V4.1.1 planned with WCAG 2.2 AA; basis of the contradiction noted in the verified status block
Want to see this thinking applied to a real project?