Accessibility

// Definition

The practice of designing and testing software so it is usable by people with a wide range of abilities — including users who rely on screen readers, keyboard navigation, voice control, switch access, or high-contrast display modes. In QA, accessibility testing involves both automated scanning and manual verification. Automated tools (axe, Lighthouse, Accessibility Scanner for Android, Accessibility Inspector for iOS) catch structural issues such as missing labels, insufficient colour contrast, and incorrect ARIA roles — typically around a third of all accessibility issues. The remaining two-thirds require testing with actual assistive technologies: VoiceOver on iOS, TalkBack on Android, NVDA or JAWS on Windows. WCAG 2.1 AA is the most widely referenced standard; Level AA compliance is required by law in many jurisdictions (ADA, EN 301 549, AODA). Integrating accessibility checks into CI — for example, running axe as part of a Playwright or Selenium test suite — prevents regressions from being merged undetected.

// Related terms

• WCAG

  Web Content Accessibility Guidelines — the international standard for accessible web content, organised into four principles (Perceivable, Operable, Understandable, Robust). Levels A, AA, and AAA define increasing conformance. Most regulations target WCAG 2.1 or 2.2 Level AA.

• Screen Reader Testing

  Verifying an interface works with assistive technology that reads content aloud — VoiceOver, NVDA, JAWS, TalkBack. Catches missing labels, broken focus order, and content visible only to sighted users.

• ARIA Attributes

  HTML attributes (role, aria-label, aria-describedby, aria-live) that expose semantics and state to assistive technologies. The first rule of ARIA is: don't use ARIA — prefer native HTML elements when they convey the same meaning.

• Accessibility Audit

  A structured review of a product against accessibility standards, combining automated scans (axe, Lighthouse), manual checks, and assistive-technology testing. Automated tools catch roughly a third of issues; the rest need human judgement.