WCAG 2.2 Demo Suite

Cognitive and content testing

These criteria have no automated rules, need no assistive technology, and are skipped by most audits — which is unfortunate, because cognitive and learning disabilities are the largest category of disability there is, and these failures affect every user under stress.

Why this is the neglected half

Nothing here can be checked by a scanner, and none of it shows up in a screen reader pass. It requires someone to read the content and use the process with attention. That makes it easy to leave out of a schedule — and it is where a great deal of real exclusion lives.

Language and content

  1. Read the page aloud. Sentences you stumble over are sentences your readers will too. 3.1.5
  2. List every abbreviation and piece of jargon. Is each one expanded or defined at or near first use? 3.1.3 3.1.4
  3. Search for idioms. "Ballpark figure", "beat around the bush", "on the fly" — all read literally by some users. 3.1.3
  4. Check instructions for sensory-only references. Search the text for "click the green button", "on the right", "the box below", "the large text". Each needs a name. 1.3.3
  5. Read the headings alone as an outline. Does it work as a table of contents? 2.4.6 2.4.10
  6. Check the language attributes. Is lang on <html> correct, and does every foreign passage carry its own? 3.1.1 3.1.2

Forms and error recovery

  1. Submit the form empty. Is every error described in text, tied to its field, and announced? 3.3.1
  2. Submit deliberately wrong values. Does the message say how to fix it, or only that it is wrong? 3.3.3
  3. Look for format rules stated only in errors. The expected format must appear before submission. 3.3.2
  4. Walk a multi-step process and note anything asked twice. 3.3.7
  5. Try to do something irreversible. Is there a confirmation, a review step, or an undo? 3.3.4 3.3.6
  6. Check the authentication flow. Can you paste a password? Do the fields carry autocomplete tokens? Is there a CAPTCHA? 3.3.8

Timing and interruptions

  1. Start a form and walk away for the length of the timeout. Are you warned? Can you extend? Is your work kept? 2.2.1 2.2.5
  2. Is the timeout duration disclosed before you start? 2.2.6
  3. Sit on the page and do nothing for two minutes. Note every pop-up, chat prompt, banner and auto-update that appears uninvited. 2.2.4 3.2.5
  4. Find moving content. Can it be paused, stopped or hidden? 2.2.2

Consistency across pages

These three cannot be evaluated on a single page. You need a sample — which is one of the reasons choosing a sample matters so much.

  1. Compare the navigation across five pages. Same items, same relative order? 3.2.3
  2. List the accessible name of every control that does the same job across pages. Same function, same name? 3.2.4
  3. Find every help mechanism. Is it in the same relative place wherever it appears? 3.2.6

The five-minute sanity check

Give the page to someone who has never seen it and ask them to complete the main task while thinking aloud. Do not help. Every hesitation is data, and most of what you learn will not correspond to any success criterion — which is the point. Conformance is the floor.