Accessibility

05 / Checking & Testing

|

Key question

How do I know if it works?

Accessibility can be tested, but not with a single tool alone. Automated tests only find some of the issues. The rest are discovered using a keyboard, screen reader, and zoom, and above all, with people who use these assistive technologies every day.

Learning objective: You can describe controls with their role, name, and state, write suitable alternative text, and check a page in three stages.

What is not tested is not accessible.

5.1 / Role, Name, Status

A screen reader reads what the code says.

For assistive technologies, every control must answer three questions:

01

Role

What is it?

02

Name

What is it called?

03

Condition

What's the score at the moment?

The first rule is: native HTML elements over ARIA. A real button comes with its own role, keyboard support, and focus. ARIA only complements what HTML cannot do, such as aria-expanded for a drop-down menu.

No ARIA is better than bad ARIA.

5.2 / Describing pictures

The alternative text replaces the image.

An alternative text explains in a single sentence what the image means in context: for a chart, the key takeaway; for a photo, the essentials; for a button, its destination. 'Image of' is redundant, as the screen reader already announces that.

Purely decorative images are given an empty alternative attribute (alt="") so that they are skipped. Complex graphics also require a detailed description or a data table.

Describe the meaning, not the pixels.

Try it out
Alternative text

Three pictures, three contexts: Choose the matching text for each.

Bild 1

Auf einer Event-Seite, neben dem Anmeldeknopf. Welcher Alternativtext passt?

Bild 2

Als Trennlinie zwischen zwei Abschnitten. Welcher Alternativtext passt?

Bild 3

In einem Bericht, Text: «Der Umsatz stieg 2026 deutlich». Welcher Alternativtext passt?

0 von 3 passend gewählt.

5.3 / Testing and documentation

Automatically, by hand, and with people.

A good test has three stages: automated tools such as WAVE or axe find missing labels and poor contrast. Manual testing using a keyboard, zoom, and a screen reader identifies workflow issues. User testing with disabled people reveals whether it truly works.

Whatever is identified is prioritised in the planning. Many organisations publish an accessibility statement: what is compliant, what is not yet compliant, and how to report any issues.

Testing is a process, not a destination.

Exercise

A minor audit

Task: Test a page of your project in three stages.

STEP 1

Run an automated testing tool and note down the errors.

STEP 2

Test manually: keyboard, 200% zoom, screen reader on mobile.

STEP 3

Check all images and icons for suitable alternative text.

STEP 4

Write a brief explanation: fulfilled, open, next step.

Evaluation: Which errors did the tool find, which ones did only you find, and what surprised you the most about the screen reader?