WCAG 2.2 explained: what changed and what it means for your site

The Web Content Accessibility Guidelines are written by the W3C, not by any legislator, which is why the same document ends up referenced by law in very different jurisdictions. Version 2.0 dates from 2008, 2.1 from 2018, and 2.2 became a W3C Recommendation in October 2023.

Each version is backwards compatible. A page that meets 2.2 also meets 2.1 and 2.0. That matters when a contract names an older version: meeting the newer one satisfies the older requirement automatically.

The structure nobody explains

WCAG is organised under four principles, usually abbreviated POUR: content must be perceivable, operable, understandable and robust. Under those sit thirteen guidelines, and under those sit the success criteria, which are the only testable part.

Each success criterion carries a level: A, AA or AAA. These are not a maturity ladder. Level AA is the target in essentially every legal reference, including EN 301 549. Level AAA was never intended as a general goal, and the W3C says so explicitly, because some AAA criteria cannot be met for all content types. Anyone selling you “AAA compliance” for a whole website is selling something that the standard itself calls unachievable.

What WCAG 2.2 added

Nine success criteria are new. Six are relevant to almost every site.

2.4.11 Focus Not Obscured (Minimum), AA. When an element receives keyboard focus, it must not be entirely hidden behind a sticky header, a cookie banner or a chat widget. This is the criterion that most sticky headers fail.

2.5.7 Dragging Movements, AA. Anything that works by dragging must also work with a single pointer without dragging. Sliders, sortable lists, map panning and range selectors are the usual offenders.

2.5.8 Target Size (Minimum), AA. Targets must be at least 24 by 24 CSS pixels, with defined exceptions for inline links in text and for spacing. Note that this is smaller than the 44 pixels often quoted from platform guidelines, which is a design recommendation, not this criterion.

3.2.6 Consistent Help, A. If a help mechanism exists on multiple pages, it has to appear in the same relative order on each of them.

3.3.7 Redundant Entry, A. Information the user already entered in the same process must not be asked for again, unless re-entry is essential, for example for a password confirmation.

3.3.8 Accessible Authentication (Minimum), AA. No step of a login may require a cognitive function test such as remembering a password or solving a puzzle, unless an alternative exists or the field supports password managers. In practice this means: do not block paste into password fields, and do not use image puzzles as the only route.

The remaining three, 2.4.12 Focus Not Obscured (Enhanced), 2.4.13 Focus Appearance and 3.3.9 Accessible Authentication (Enhanced), are AAA.

What was removed

Success criterion 4.1.1 Parsing was removed. It required valid markup in a narrow sense: no duplicate IDs, correctly nested elements. Browsers and assistive technology now handle the errors it described, so the criterion tested something that no longer caused a barrier. Automated tools were producing large numbers of 4.1.1 failures that had no effect on any user, which is a good illustration of why an automated score is not an accessibility result.

What automated testing can and cannot see

This is worth being blunt about, because it determines how much a testing tool is worth to you. Automated checks reliably find missing alternative text, insufficient colour contrast in plain cases, missing form labels, missing page language and broken ARIA references. Depending on the study you read, that is somewhere between a quarter and a third of the criteria.

They cannot tell you whether the alternative text is a useful description or the word “image”. They cannot tell you whether the heading structure reflects the actual outline of the page. They cannot tell you whether a keyboard user can complete your checkout. Those require a person, and the last one requires a person who tries.

Our testing walkthrough covers the manual checks in the order that finds the most problems fastest.