WCAG 2.2 explained: the nine new success criteria and what changed from 2.1
WCAG 2.2 is the version most accessibility work is now measured against, and it is the version AdaChecked scans for. If you last looked at this when WCAG 2.1 came out, the delta is small but pointed: nine success criteria added, one removed, and the additions are noticeably more concerned with motor ability, memory, and the parts of a site people actually have trouble operating.
This is a walkthrough of what changed and what each new criterion means in practice, aimed at the person who has to decide what to fix rather than the person writing the specification.
What WCAG 2.2 is
The Web Content Accessibility Guidelines are a W3C standard that breaks accessibility into testable success criteria at three conformance levels: A, AA, and AAA. Level AA is the practical benchmark — it is the level cited in accessibility statements, referenced by procurement, and used as the yardstick in web-accessibility claims. WCAG 2.2 became a W3C Recommendation on 5 October 2023 and was republished on 12 December 2024.
WCAG 2.2 is backwards compatible. A site that conforms to 2.2 also conforms to 2.1 and 2.0, with the single exception noted below. That means moving your target from 2.1 AA to 2.2 AA is an addition of six checks, not a re-audit from zero.
What changed from WCAG 2.1
Nine success criteria were added and one was removed. Six of the nine additions sit at Level A or AA, which is the range that matters if AA is your target: two at Level A (Consistent Help, Redundant Entry) and four at Level AA (Focus Not Obscured (Minimum), Dragging Movements, Target Size (Minimum), Accessible Authentication (Minimum)). The other three are Level AAA.
The removal is 4.1.1 Parsing, which W3C marked obsolete. Counting forward from the 38 Level A and AA criteria in WCAG 2.0 and the 50 in WCAG 2.1, the A and AA set in WCAG 2.2 comes to 55.
2.4.11 Focus Not Obscured (Minimum) — Level AA
When a keyboard user tabs to an element, that element must not be entirely hidden by other content the page author put there. The usual offenders are sticky headers, cookie banners, chat bubbles, and promo bars: the element receives focus, the browser scrolls it into view, and a fixed bar sits right on top of it. The user is now operating a control they cannot see.
Test it by tabbing slowly down a long page with a sticky header present. The fix is normally scroll-margin on focusable elements, or reducing the sticky element's height, not removing the header.
2.5.7 Dragging Movements — Level AA
Anything operated by dragging must also be operable with a single pointer action that does not require dragging — unless dragging is essential to the function. Sliders, drag-to-reorder lists, map panning, carousels that only respond to swipe, and signature pads are the common cases.
This is a motor-ability criterion. People using a head pointer, a trackball, or a tremor-affected hand can click but cannot reliably hold and drag. The fix is usually an alternative rather than a replacement: arrow buttons beside the slider, up and down controls on the reorderable list, previous and next buttons on the carousel.
2.5.8 Target Size (Minimum) — Level AA
The W3C wording is specific: the size of the target for pointer inputs is at least 24 by 24 CSS pixels. There are five exceptions, and two of them cover most real designs. The spacing exception allows a smaller target if a 24-pixel circle centred on it would not overlap another target's circle. The inline exception covers links inside a sentence, where the line-height constrains the size.
In practice this bites on icon-only buttons, close buttons on modals, pagination numbers, star ratings, social icons in footers, and quantity steppers. The cheapest fix is padding rather than a bigger icon: keep the glyph, grow the hit area.
3.2.6 Consistent Help — Level A
If you offer a help mechanism — a contact link, a phone number, a chat widget, a help page link — and it appears on multiple pages, it has to appear in the same relative order on each of those pages. It does not require you to offer help at all. It requires that where you do, it stays where people learned to find it.
The failure is usually accidental: help sits in the header on the marketing pages and moves into a footer accordion in checkout, which is precisely where someone needs it.
3.3.7 Redundant Entry — Level A
Information the user already entered in the same process must be either auto-populated or available to select, rather than asked for again. The classic case is a checkout that asks for the shipping address, then asks for the billing address with no copy option, then asks for the email a third time on the confirmation step.
Exceptions exist where re-entry is essential — confirming a password, or a memory test that is the point of the exercise. Everywhere else, a checkbox that says the billing address is the same as shipping satisfies this.
3.3.8 Accessible Authentication (Minimum) — Level AA
A cognitive function test — remembering a password, solving a puzzle, transcribing characters — must not be the only way to log in, unless there is an alternative, a mechanism to assist, or the test is object recognition or identifying non-text content the user provided.
The practical reading: allow password managers to work. Do not block paste into the password field, do not split a code across several inputs in a way that defeats autofill, and make sure email-link or passkey sign-in exists where you use image-based CAPTCHA. A surprising number of stores fail this by disabling paste on the discount-code and password fields for anti-fraud reasons.
The three additions at Level AAA
- 2.4.12 Focus Not Obscured (Enhanced) — the focused element must not be obscured at all, not merely not entirely obscured.
- 2.4.13 Focus Appearance — sets minimum size and contrast requirements for the focus indicator itself.
- 3.3.9 Accessible Authentication (Enhanced) — removes the object-recognition and user-provided-content exceptions from 3.3.8.
4.1.1 Parsing was removed
4.1.1 Parsing required markup to have complete start and end tags, correct nesting, no duplicate attributes, and unique IDs. W3C removed it from WCAG 2.2 on the basis that the problems it was written to prevent are now handled by browsers and assistive technology, so failing it no longer corresponded to a real barrier.
This matters if you are reading an older audit. Duplicate IDs and malformed markup can still cause genuine failures, but they now surface under the criteria they actually break — 1.3.1 Info and Relationships, or 4.1.2 Name, Role, Value — rather than as a parsing violation in their own right.
Which version applies to you
- United States, federal agencies and their vendors: the Revised Section 508 Standards incorporate WCAG 2.0 Level A and AA by reference, and have applied since 18 January 2018.
- United States, private sector: ADA Title III does not name a technical standard. WCAG is what plaintiffs, settlements, and consent decrees have used in practice, which is why most private-sector remediation targets it.
- European Union: the European Accessibility Act points to the harmonised European standard EN 301 549. The version cited in the Official Journal of the European Union tracks WCAG 2.1 Level AA, and a revision aligning the standard to WCAG 2.2 is in progress. Check the current citation before relying on a presumption of conformity.
- Everywhere: targeting WCAG 2.2 AA satisfies the older references too, because 2.2 is a superset of 2.1 and 2.0 apart from the removed parsing criterion.
Where to start
If you already conform to 2.1 AA, the six new A and AA criteria are a short, concrete list and most of them are design fixes rather than engineering projects. Target size and focus obscuring are the two that tend to affect every template at once, so they are usually the best first pass.
If you have never audited at all, start with a scan rather than a checklist. The AdaChecked Site Scanner tests against WCAG 2.2 A, AA and AAA in a real browser and returns issues mapped to the specific criterion they fail, which is the only way to know which of these actually apply to your templates.