Definition: Manual accessibility testing means people testing a website against WCAG with a keyboard, a screen reader, magnification and their own judgement. It is needed for everything software cannot decide: whether alt text fits, whether an error message helps, whether the reading order makes sense.
German: Manuelle Prüfung · Last reviewed: 28 September 2026
How automated and manual testing fit together at Reviseberg is set out in our methodology.
What people check
- Meaning: does the alt text describe the image? Does the heading match its section? Is the link text clear in context?
- Operation: can every function be reached by keyboard? Does the focus order make sense?
- Perception: does the page work at 200% text size and 400% zoom? Does a screen reader announce a status message?
- Media: are captions complete and in sync? Does the video need audio description?
- Forms: do error messages say what is wrong and how to put it right?
How a manual test works
Usually a representative sample of pages and journeys is chosen – home page, search, product page, basket, contact form. Each page is checked criterion by criterion, with the method recorded: "keyboard only", "NVDA + Firefox", "magnifier at 400%". In Germany, a recognised procedure is the BITV-Test from the BIK project; internationally, the W3C's WCAG-EM describes the steps [2][3].
For a first impression, the W3C's Easy Checks are often enough: a handful of simple checks anyone can do without specialist knowledge [1]. They do not replace a full test, but they quickly show where things go wrong.
A manual result is a snapshot. After a redesign or new content, it has to be repeated.
How Reviseberg handles it
Reviseberg takes the mechanical part off your hands: every axe-core rule on every page and a keyboard agent on your most important journey. For the criteria no automated check reaches, guided tests give step-by-step instructions. A result needs a method, is never overwritten and expires after twelve months – after which the criterion is "untested" again. If a later crawl finds a failure, the criterion counts as failed, even if somebody ticked it before.
Related terms
Automated accessibility testing · Screen reader · Keyboard accessibility · Accessibility statement · Assistive technology