WCAGvaktAccessibility you can document

WCAG 2.1 AA · legal requirements

Find the barriers. Show what is proven.

WCAGvakt tests your website in a real browser, explains what needs to be fixed and prepares the basis for the accessibility statement.

No polished vanity score and no guesswork presented as fact. Every result shows whether it is machine evidence, partial support or a manual check.

Free · no account · 3 tests per day, up to 10 pages.

Start with the website address

● Evidence level visible

Press Enter to start.

View example report
Advanced
What do you want to test?

Not published yet? Sign in and upload the files to test a redesign, components or drafts before launch.

Evidence levelAuto, partial or manual
Technical traceElement, rule and recommended fix
Expert reviewJudgement-based checks are never presented as final truth

You should be able to tell a finding from a suspicion.

That is why WCAGvakt shows the evidence level in the report itself. The matrix covers 48 of the 50 Level A and AA success criteria. Anything a machine cannot determine reliably is passed on as a clearly labelled professional review.

24requirements tested fully automatically
20partially machine-supported
4guided manual review
44/48have machine support before expert review

See the requirement matrix and the evidence level for every requirement →

From barrier to documented fix

See what actually stops people, how the test finds it, and what still needs human review.

    For professionals Method, evidence level and comparison

    This is why we find more than a pure code scanner

    • We test, we do not only read: keyboard, visible focus, text resize, reflow and target size are tested in a real browser – not guessed from the code.
    • The statement is included: the accessibility statement (Norwegian format) is generated from the findings and kept current – you do not have to write it.
    • We never guess: certain findings are kept apart from «needs review», so red findings are never mixed with professional judgement. Everything runs on our own servers – your page is never sent to a language model.
    Layer 1 · Rule engine

    Automated rules

    Contrast, missing alt text, form labels, ARIA validity — via axe-core.

    Layer 2 · Behaviour

    Keyboard & display

    Tabs through the page and checks visible focus, 200 % text, 320 px reflow, target size and hover content.

    Layer 3 · Judgement

    Measurable judgement

    Judgement-based requirements get partial support where facts can be measured: identical link names pointing to different targets, text inside images, instructions that rely on colour, error messages without suggestions.

    Layer 4 · Manual

    Guided check

    The few requirements nobody can automate (video captioning, legal forms) — with pre-filled answers.

    1. FetchDOM, CSS, links and media.
    2. TestRules, keyboard, zoom and reflow.
    3. AssessMachine evidence is clearly separated from «needs review».
    4. DocumentEvidence level, fix and remaining checks.

    Three evidence levels in the report

    Every finding is marked with how certain it is – so you know what is proven, and what a human still has to look at.

    Evidence levelUsed forWhat the report should say
    AutoObjective violations that can be reproduced in DOM, CSS, HTTP or browser behaviour.“Error found” with technical evidence and the exact element.
    PartialRequirements where the tool finds strong signals, but cannot settle the whole requirement on its own.“Should be checked” with the basis, a screenshot/context and a suggestion.
    ManualRequirements that need understanding of the content, media coverage or legal/domain-specific context.“Needs manual review” with a checklist and documentation.

    ✔ Passed ▲ Should be checked ✖ Error ● Needs a human

    Do you want to compare us with other tools?

    We have written an honest comparison against axe DevTools, WAVE and Lighthouse – including what they do better than us. See the comparison →

    ← Back
    LanguageNOEN