Accessibility
Every edit you save in the Visual Editor is scanned by axe-core. Issues surface as inline findings in the panel so you can fix them before publishing — but findings are warnings only and never block you from shipping.
Rule families
The editor runs a focused subset of axe rules tuned for the kinds of changes people make in a visual editor. Each family covers one common type of regression.
- color-contrast — Foreground and background colours fail to meet WCAG AA contrast ratios. Most common when editing text or background colours. Axe docs.
- image-alt — An
imgis missing analtattribute. Often triggered when swapping out a hero image without updating the description. Axe docs. - button-name — A
buttonhas no accessible name. Triggered by icon-only buttons that lackaria-labelor visually-hidden text. Axe docs. - link-name — An
ahas no accessible name. Same problem asbutton-namebut for links. Axe docs. - heading-order — Heading levels jump (e.g. an
h2directly under anh4). Easy to introduce when restructuring a page section. Axe docs.
Severity levels
Each finding is tagged with one of axe-core's four standard severities. Severity influences how prominently the finding is displayed but does not block publishing.
- critical — Blocks users from completing core tasks. Failing colour contrast on a primary CTA is the canonical example.
- serious— Significant impact on a user's ability to use the page. Missing form labels, unlabelled buttons.
- moderate — Causes confusion but not blocking. Misordered headings, low contrast on secondary text.
- minor — Best-practice violations that rarely affect day-to-day use. Surfaced for completeness.
Findings never block publish
A vs B treats a11y findings as guidance, not gates. Many tests intentionally use bold visual treatments that fail one rule or another for a short experimental run. You can publish a variation with open findings at any time — the dashboard simply surfaces a summary so your team has full context.
Org-wide summary
Findings roll up into the Accessibility tab of your organisation dashboard. There you can see the total open findings per experiment, severity breakdown, and which rule fires most often across the org. The summary is read-only and intended for accessibility leads or audit reviewers.
Related
- Troubleshooting — what to do if findings disagree between the editor and a third-party scanner.
- Review and publish — where the a11y summary appears in the publish flow.