WCAG 2.2 Is Final: Nine New Criteria, One Moving Bar

Five years after WCAG 2.1, the accessibility standard moved again. WCAG 2.2 reached W3C Recommendation on October 5. It adds nine new success criteria, weighted on purpose toward cognitive accessibility and mobile use. The people and modes the 2.1-era gaps most affected. It retires one legacy criterion, 4.1.1 Parsing, outdated now that browsers and assistive tech handle broken markup the same way. It otherwise preserves the conformance architecture the compliance world has built on. Every new criterion names a barrier real users already hit. The standard is just catching up to the complaints.

The timing gives the release more weight than a point-version usually carries. Regulators worldwide are writing WCAG versions into law by number. The European Accessibility Act’s 2025 deadline drawing near. The U.S. Department of Justice’s rulemaking citing 2.1. The procurement standards that name their versions.

That makes version literacy a compliance-planning skill now, not a specialist’s trivia. Know which version your duties cite. Know which version your last audit used. Know which version your new work should target. That is the difference between a coherent program and a costly muddle. The new criteria audited against the patterns we actually see, the guidance for programs at every stage, and the point the small delta is making.

The new AA criteria, audited against weekly patterns

The additions that matter most for institutional sites, each held against failures practitioners find weekly. Focus Not Obscured (2.4.11): the sticky header, persistent cookie banner, or floating toolbar that hides the keyboard user’s focus indicator as they tab through the page is now a citable failure. An extremely common pattern with a plain fix: scroll-padding, z-index discipline, focus management. This is the criterion we predict will drive the most fix tickets.

Dragging Movements (2.5.7): any interaction requiring drag (sliders, sortable lists, map pans, kanban boards) must offer a single-pointer alternative. That reaches deep into component libraries and argues for fixing at the design-system layer, where the fix spreads.

Redundant Entry (3.3.7): forms that make you retype what you already gave in the same process (confirm fields, multi-step flows that forget, the shipping-address re-type) now fail. A criterion that is also simply conversion work, and the cleanest example of the standard lining up with self-interest.

Accessible Authentication (3.3.8): login flows that demand mental tests (memorizing, retyping, solving) without alternatives now fail. That reaches CAPTCHA choices, two-factor setups, and password UX, and lands directly on the login surfaces every institutional site runs. Target Size Minimum (2.5.8) sets a floor for touch-target size with defined exceptions, codifying the mobile-tap-target advice we have given for a decade. The remaining additions, help placed in the same spot, focus-look refinements at AAA, round out a set whose common trait is plainness. Every criterion above is concrete, testable, common in the wild, and fixable at the component layer.

The guidance, by program state

• Mid-remediation programs: hold course, add the delta. Existing 2.1 AA obligations stand. The settlements, the policies, the regulatory citations do not retro-upgrade. 2.2 becomes the target for forward work: new builds, new components, the next audit cycle. The delta is genuinely modest for teams already practicing. Nine criteria, most fixable at the system layer, absorbed into the standing backlog rather than reconstituted as a crisis.

• Buying and policy owners: update the citations on purpose. The contract language, the vendor-conformance requests, and the internal standards that name ‘2.1 AA’ need a versioning decision. When new work targets 2.2, and what existing commitments remain 2.1. Made once, written down. Europe’s 2025 deadline and the pending U.S. rulemaking make this quarter the natural moment.

• Design-system teams: this is your release. Every criterion above lands at the component layer. The focus management and drag alternatives. The form memory, the target sizes. Fixing components fixes sites, and the 2.2 audit of the component library is the single highest-coverage action available this quarter.

The point the modesty is making

The delta’s size teaches the oldest lesson of this coverage. Nine criteria in five years is a standard refining, not upheaving. The experience of absorbing it divides this readership along exactly the line our accessibility coverage has drawn since 2010.

For the practice-mode teams, accessibility is a habit they keep. Owners at the system layer. Editing habits. A testing rhythm. Buying gates. For them, 2.2 is a sprint of component work and a citations update. Refinement absorbed in stride, exactly as 2.1 was for the same organizations five years ago.

For the project-mode teams, accessibility is one-off compliance. The audit bought under threat, the fixes shelved at sign-off. Each version bump re-runs the full crisis. The re-audit, the re-budget, the re-panic. At prices that compound precisely because nothing stuck between bumps.

The bar will keep stepping forward. 2.2 will not be the last number, and the W3C’s work on the next generation is public. The regulators will keep writing numbers into law. The mode question, practice or project, remains the highest-leverage accessibility call any institution makes. It sets the cost of every future criterion before any of them is written. Europe’s deadline and the U.S. rulemaking make this the quarter to choose deliberately, and thirteen years of evidence say practice mode is cheaper by every year that passes.

The quarter’s assignment

Compressed. Read the nine criteria against your site. The audit above covers the likely findings. Order the component-library pass, the action that covers the most ground. Make the versioning decision for policy and procurement citations. And, if your organization is still in project mode, use the modest delta as the on-ramp to practice mode that it genuinely is. Nine concrete criteria, absorbed properly, is how the ongoing practice starts.

For continuity: we have now covered the standard’s movement from 2.0’s testability revolution, through 2.1’s mobile-era additions, to 2.2’s cognitive and interaction refinements. The through-line has never varied. The standard tracks real barriers experienced by real people. Each version names failures that were always failures. And the teams that treat the numbered releases as write-ups of duties they already honored, rather than as new burdens, are the ones for whom the letters never arrive and the audits never surprise.

One moving bar, thirteen years of the same advice: walk with it. It remains the cheapest pace ever offered.