EN 301 549 testing works best as two layers: automated tools catch machine-detectable failures such as missing alternative text and low contrast, while manual checks confirm keyboard operation, screen reader output and whether content makes sense. Neither layer alone shows conformance, so a reliable process uses both.
Key takeaways
- Automated tools find machine-detectable failures quickly, but a clean scan does not prove conformance.
- Keyboard operation, focus order and screen reader output need manual testing by a person.
- Run automated scans on every build and reserve manual passes for each release and major change.
- Treat a passing scan as the start of testing rather than the end of it.
What does EN 301 549 require for web applications?
Clause 9 of EN 301 549 sets the accessibility requirements for web content, and it is aligned with the success criteria of WCAG 2.1. The standard is published by ETSI, the European standards body, and it is the reference most European accessibility rules point to for digital products.
The European Accessibility Act sets accessibility requirements for certain products and services placed on the EU market, and it sets 28 June 2025 as its application date. You can read the official text on EUR-Lex, under Directive (EU) 2019/882. This article is general information and not legal or tax advice.
For web applications, the practical meaning is that every screen a user reaches must be perceivable, operable, understandable and robust. The WCAG criteria describe those outcomes in detail, and the WCAG 2.1 Quick Reference lists each success criterion with its technique notes.
What can automated tools detect against EN 301 549?
Automated tools answer the machine-checkable part of the question quickly and consistently. They are good at finding missing text alternatives, form fields without labels, empty buttons, duplicate element IDs and colour contrast below the WCAG thresholds. Engines such as axe-core run these rules inside an automated test suite, and browser-based checkers such as WAVE let a tester inspect the same issues visually.
Where automated scans stop
A scanner can see that an image has alternative text, but it cannot judge whether that text is accurate. The same limit applies to headings, link wording and reading order. Automated tools work on the markup and the rendered page at one moment, so they miss problems that appear only after a sequence of interactions.
- Alternative text that is present but meaningless, such as a file name
- Link text such as 'here' that makes sense only in context
- Focus order that is technically possible but confusing
- Error messages that exist in the code but do not explain the fix
- Modal dialogues and menus that appear only after user input
Which checks need a person?
Manual checks cover the requirements that depend on meaning, context and behaviour. They are slower than scans, but they are the only way to confirm that a user can finish a task. A useful manual pass focuses on the flows that matter most to the business rather than every page.
Keyboard and focus testing
Use only the keyboard to complete the main tasks. Tab through every control, confirm that focus is always visible, and check that no element traps focus. Open and close each dialogue, menu and date picker without a mouse.
Screen reader testing
Test the main flows with at least one common screen reader and browser combination. Listen for the name, role and state of each control, and check that dynamic messages such as validation errors are announced. A scan cannot tell you whether the spoken output makes sense.
Content and meaning
Read the labels, instructions and error text as a non-expert user would. Check that colour is not the only way information is conveyed, and that zoomed layouts do not hide content or controls.
- Every interactive element is reachable and operable by keyboard
- Focus is visible and moves in a logical order
- Form errors are identified in text and linked to the field
- Status messages are announced without moving focus
- Content remains usable when zoomed and when text spacing changes
How should a test workflow combine both?
A workflow should run automated checks continuously and manual checks at defined points. The automated layer catches regressions early, while the manual layer protects the flows that users depend on. Neither layer replaces the other, and the boundary between them should be written down.
- Run automated scans in the continuous integration pipeline on every build and fail the build on new violations.
- Triage the scanner output by user impact, and record the rules that were not covered by the tool.
- Run a manual keyboard and screen reader pass on the core flows before each release.
- Retest the affected flows after any change to navigation, forms, dialogues or custom components.
- Keep a short statement of the scope, tools, browsers and assistive technologies used in each test.
Technical teams that build or maintain these applications often need help wiring such checks into existing delivery pipelines. Rui Codex offers technical implementation support for web development projects, including the engineering work of making accessibility testing part of a build process. It does not provide accessibility audits or certification, and the scope of any test should be agreed with the organisation responsible for compliance.
Hypothetical examples of testing a booking flow
Here are two hypothetical scenarios:
In the first scenario, a booking form passes an automated scan with no reported violations. A keyboard tester then finds that a custom date picker opens but never returns focus to the input after a date is chosen, so the user must reload the page to continue. The scanner had no rule for that behaviour, and only the manual pass exposed it.
In the second scenario, a dashboard uses red and green status markers that pass the contrast rule. A screen reader user hears only the element name because the status is conveyed by colour alone and never written in the markup. The automated check missed it, and a manual review of the status labels corrected it.
When do automated results mislead?
Automated results mislead when a pass is read as proof of conformance. A clean report means that the rules the tool implements found no problems on the pages it visited. It says nothing about pages it did not reach, states it did not simulate, or meanings it cannot evaluate.
- Single-page scans that never reach logged-in or multi-step views
- Rules that were disabled in the configuration without a recorded reason
- Components rendered only after a user action, such as menus or modals
- Content that changes with time, location or user role
When you plan the next testing cycle, a conversation about your application's architecture is often the quickest route to a clear scope. You can request a project consultation or review our web development services to see how implementation work is organised.
Conclusion
EN 301 549 testing is a combination of two kinds of evidence. Automated tools give fast and repeatable coverage of the rules that machines can check, and manual checks confirm the behaviour and meaning that scanners cannot judge. Teams that record what each layer covers, and what it leaves out, can make honest claims about their applications. If your team needs help building that process into the delivery pipeline, you can email info@ruicodex.com to discuss the scope of your project.
Frequently Asked Questions
Can an automated tool make a web application compliant with EN 301 549?
No, an automated tool cannot make a web application compliant on its own. It can find many machine-detectable failures, but it cannot evaluate meaning, behaviour across a full task or every state of a dynamic interface. Compliance depends on a combined testing approach and on the judgement of the organisation responsible for the product.
Which manual checks are most important for EN 301 549?
Keyboard operation and screen reader testing of the core user flows are the most important manual checks. They reveal focus traps, unlabelled controls and error messages that users cannot understand. Testing should cover forms, navigation, dialogues and any custom components that the scanner does not fully inspect.
How often should EN 301 549 tests be run?
Automated scans should run on every build, and manual checks should run before each release and after significant changes to navigation, forms or interactive components. The right frequency depends on how often the application changes. A written schedule makes it clear which flows are retested and by whom.
Does Rui Codex carry out accessibility audits?
Rui Codex offers technical implementation support for web development and related engineering work. It does not provide accessibility audits, certification or legal compliance advice, so organisations should agree the scope of formal testing with the party responsible for compliance. Contact details are available at https://ruicodex.com/Contact.
Sources
- Directive (EU) 2019/882 on the accessibility requirements for products and services - EUR-Lex
- ETSI home page - ETSI, publisher of EN 301 549
- WCAG 2.1 Quick Reference - W3C Web Accessibility Initiative
- axe-core repository - Deque Systems
- WAVE web accessibility evaluation tool - WebAIM