UX Accessibility and WCAG 2.2: What Product Teams Own and How to Check It

published on 20 November 2026

By Nick Babich

UX accessibility: WCAG 2.2 Level AA is the W3C target I recommend for product teams, and much of it is decided in design. Engineers build it, but many of the decisions are made long before code exists.

Automated tools caught 5 of the 12 problems I seeded in a test page, and that gap is why ownership and manual checks matter. If you are shortlisting UX teams that design B2B software to WCAG, ask how they test beyond a scan. The sections below cover which version to target, what changed in 2.2, who owns which criterion, and what my test showed.

Key Takeaways

  • Target: WCAG 2.2 is a W3C standard and is also approved as ISO/IEC 40500:2025, according to the W3C. Level AA is the level I recommend designing to.
  • Version: The W3C says WCAG 2.2 does not deprecate or supersede 2.1, but the legal baselines name different versions, so check which one applies to you.
  • Ownership: Contrast, focus states, target size, labels and error messages are design decisions, while engineering builds the markup and behavior underneath.
  • Automation: On my seeded test page (Own test, November 2026), axe-core and Lighthouse caught 5 of 12 issues, so a scan alone does not tell you whether the product meets WCAG.
  • Process: Put accessibility checks into discovery, design review, handoff, build, QA and periodic audits instead of saving them for the end.

What WCAG is, in product terms

WCAG 2.2 has 13 guidelines under 4 principles, with conformance levels A, AA and AAA, according to the W3C WCAG 2 Overview. Each guideline contains testable success criteria (SC), and every SC sits at one level.

The four principles are Perceivable, Operable, Understandable and Robust (POUR). Here is how I'd translate them for a product team:

  • Perceivable: can people see or otherwise perceive the content? Design checks: text contrast, text alternatives for images, color not the only cue, reflow on small widths.
  • Operable: can people use it with a keyboard, a pointer or touch? Design checks: visible focus, target size, a non-drag alternative for dragging.
  • Understandable: can people tell what to do and recover from mistakes? Design checks: labels, error text, consistent help, no repeated data entry.
  • Robust: does the interface expose its name, role and value to assistive technology? This is mostly engineering, but design decides which components exist and how they behave.
The four WCAG principles with design checks and the team that owns each
The four WCAG principles (Perceivable, Operable, Understandable, Robust) mapped to design checks and the team that usually owns them.

The versions matter because contracts and regulations name them. The W3C overview lists WCAG 2.0 as published on 11 December 2008, WCAG 2.1 on 5 June 2018, and WCAG 2.2 on 5 October 2023. It also says ISO/IEC 40500:2025 is the October 2023 version of WCAG 2.2, and that WCAG applies to dynamic content, multimedia, web on mobile and AI web interfaces.

If you are weighing outside help, a reviewed index of UX design agencies is a reasonable place to start your list. The questions near the end of this article tell you what to ask them.

Which WCAG version do you need: 2.1 or 2.2?

It depends on which rule applies to you: the ADA Title II web rule for State and local governments, adopted on April 24, 2024, requires WCAG 2.1 Level AA, and a Federal Register interim final rule of April 20, 2026, extended its compliance dates. Section 508 points to WCAG 2.0 Level AA. The W3C overview says "WCAG 2.2 does not deprecate or supersede WCAG 2.1", and it encourages people to use the latest version.

The table summarizes the baselines I read. This is a summary of public sources, not legal advice. Check your own obligations with counsel.

Rule WCAG version and level Date Source
ADA Title II web rule for State and local governments (2024 rule, dates extended by the April 2026 interim final rule) WCAG 2.1 Level AA State and local government entities with a total population of 50,000 or more, per Federal Register IFR 2026-07663: extended from April 24, 2026, to April 26, 2027. Public entities with a total population under 50,000, and special district governments: extended from April 26, 2027, to April 26, 2028. Federal Register, April 20, 2026
Revised 508 Standards WCAG 2.0 Level AA (incorporated by reference) Page shows "Reviewed/Updated: July 2026" Section508.gov
European Accessibility Act, Directive (EU) 2019/882 Not named in the sources I used Article 31(2): measures apply from 28 June 2025 EUR-Lex
EN 301 549, 2026 version WCAG 2.2 2026 version W3C WCAG 2 Overview

On the last row, the W3C overview says: "The 2026 version of EN 301 549 uses WCAG 2.2." I found no ETSI page to confirm the EN 301 549 version, so I rely on that one sentence. I also did not trace how the EAA and EN 301 549 connect, so verify that link before you build a compliance plan on it.

My recommendation is to design to WCAG 2.2 Level AA even when your contract names 2.1 or 2.0. WCAG 2.2 adds 9 success criteria and treats 4.1.1 Parsing as obsolete, and the W3C overview says: "Content that meets WCAG 2.2 also meets WCAG 2.1 and WCAG 2.0." Choosing 2.2 AA is my recommendation, not a legal conclusion. If you design to 2.2 AA once, you avoid a redesign each time a baseline moves.

What changed in WCAG 2.2

WCAG 2.2, published on 5 October 2023, adds 9 success criteria and marks 4.1.1 Parsing obsolete, according to the W3C overview and the What's New in WCAG 2.2 page. Six of the nine are Level A or AA.

Here are the 9, with the one-line design meaning from the W3C "What to do" lines:

  • 2.4.11 Focus Not Obscured (Minimum), AA: make sure an item that gets keyboard focus is at least partially visible. The W3C example is a sticky footer banner, and I would check sticky headers and cookie banners as well.
  • 2.4.12 Focus Not Obscured (Enhanced), AAA: make sure the focused item is fully visible.
  • 2.4.13 Focus Appearance, AAA: use a focus indicator of sufficient size and contrast.
  • 2.5.7 Dragging Movements, AA: for any action that involves dragging, provide a simple pointer alternative.
  • 2.5.8 Target Size (Minimum), AA: make targets at least 24 by 24 CSS pixels or give them enough spacing.
  • 3.2.6 Consistent Help, A: put help in the same place when it appears on multiple pages.
  • 3.3.7 Redundant Entry, A: don't ask for the same information twice in the same session.
  • 3.3.8 Accessible Authentication (Minimum), AA: don't make people solve, recall or transcribe something to log in.
  • 3.3.9 Accessible Authentication (Enhanced), AAA: don't make people recognize objects or user-supplied images and media to log in.
The nine new WCAG 2.2 success criteria with their levels and the design change each asks for
The 9 success criteria added in WCAG 2.2, with level and the design change each one asks for.

Two of these are design-heavy and easy to miss. Focus not obscured is a layout problem, because a sticky banner you approved in a mockup can cover the very thing a keyboard user just reached. Redundant entry is a flow problem, because a form that asks for the same email twice has to be rethought, not patched.

What design owns and what code owns

Many Level AA criteria are visual or about interaction design. For example, SC 1.4.3 in the WCAG 2.2 Recommendation sets a contrast ratio of at least 4.5:1 for text, with 3:1 for large-scale text. No code fix helps if the approved color pair fails.

The table below is my suggested split of ownership, not something the W3C prescribes. Adjust it to your team.

Criterion Design Content Engineering QA
Text and non-text contrast (1.4.3, 1.4.11) Owns None Builds Tests
Use of color (1.4.1) Owns Reviews Builds Tests
Labels and error text (3.3.1, 3.3.2) Owns Writes Builds Tests
Focus visible and not obscured (2.4.7, 2.4.11) Owns None Builds Tests
Target size and dragging alternative (2.5.8, 2.5.7) Owns None Builds Tests
Redundant entry (3.3.7) Owns Reviews Builds Tests
Text alternatives (1.1.1) Reviews Owns Builds Tests
Keyboard operation, name, role, value (2.1.1, 4.1.2) Reviews None Owns Tests

Here is the designer checklist, with the numbers taken from the success criteria:

  • Text contrast (1.4.3): at least 4.5:1, or 3:1 for large-scale text.
  • Non-text contrast (1.4.11): at least 3:1 against adjacent colors for visual information needed to identify UI components and states, and for graphics needed to understand the content.
  • Color (1.4.1): color is not the only visual means of conveying information, indicating an action, prompting a response or distinguishing an element.
  • Labels (3.3.2): labels or instructions when content requires input. A placeholder that disappears while typing is a weak label, so I'd always add a visible one.
  • Errors (3.3.1): identify the item in error and describe the error in text.
  • Focus (2.4.7, 2.4.11): a visible keyboard focus indicator, and the focused component not entirely hidden by author-created content.
  • Target size (2.5.8): at least 24 by 24 CSS pixels, with exceptions such as spacing, an equivalent control or inline targets.
  • Reflow (1.4.10): no scrolling in two dimensions for vertical-scrolling content at a width equivalent to 320 CSS pixels. The SC notes that is a 1280 CSS pixel viewport at 400% zoom (WCAG 2.2, 2023).
Before and after versions of a sign-up form: placeholder-only labels and red-border error versus visible labels and error text
Illustrative example: the same sign-up form for a fictional product, Hollowbirch Payroll, before and after. Before: placeholder-only labels, a red-border-only error and gray hint text. After: visible labels, error text with an icon and 4.5:1 text.

I computed these contrast ratios with the WCAG relative-luminance formula (Own calculation, November 2026). Gray #999999 on white comes out at 2.85:1 (axe reports 2.84:1), so it fails 4.5:1 for normal text. #767676 on white is 4.54:1 and #757575 is 4.61:1, both passing. #949494 is 3.03:1, which clears 3:1 but not 4.5:1.

For borders, #dddddd on white is only 1.36:1, well short of 3:1, so it may fail 1.4.11 if the border is what identifies the field. The red #e53935 is 4.23:1 on white, which is fine as a border contrast number but still fails as the only error cue. If you need a button color, #1565c0 with white text is 5.75:1.

The cheapest way to hold these values is to build states and accessibility into shared components, so a focus ring, an error state and a minimum target size ship with every button and field. Then pass them on through a handoff checklist that includes focus order and contrast notes. For further reading aimed at designers, see Digital.gov's accessibility guide for UX designers.

What automated checkers catch and miss: my test

On my seeded test page, axe-core 4.14.1 and Lighthouse 13.6.0 caught 5 of 12 issues, and Lighthouse scored the page 54 (Own test, November 2026). This is one small page I built, not a study, so read it as a demonstration and not as a rate.

I wrote a one-page sign-up form with 12 deliberate WCAG problems, plus two more I counted separately. I ran axe-core 4.14.1 in headless Chrome with the wcag2a, wcag2aa, wcag21a, wcag21aa, wcag22aa and best-practice rule tags, and Lighthouse 13.6.0 on the accessibility category only. Both ran locally. Lighthouse runs axe-core under the hood (its report lists axe-core 4.14.0), so the WCAG findings matched.

Both tools caught 5 of the 12 seeded issues:

  1. Gray #999999 helper text on white, which axe reported at 2.84:1 (Own test, November 2026).
  2. An image with no alt attribute.
  3. An icon-only button with no name.
  4. No lang attribute on the html element.
  5. Three 16x16 px social links side by side, which axe flagged as below 24 by 24 px.

They also flagged one problem I did not plan: the default blue #0000ee "Privacy" link inside the dark #222222 cookie banner, at 1.69:1 against the 4.5:1 requirement (Own test, November 2026). axe also raised the best-practice rules heading-order, landmark-one-main and region.

Both tools missed the other 7:

  1. An image with alt="image". The attribute exists, so the check passes, but the text does not serve the "equivalent purpose" that SC 1.1.1 asks for.
  2. Form fields labeled only by placeholder text. axe accepts a placeholder as an accessible name. I would not call it a definite failure, since 3.3.2 asks for labels or instructions when content requires input, but it is a weak label.
  3. An error shown only by a red border on the email field (1.4.1 Use of Color).
  4. The focus outline removed with outline:none on every control (2.4.7).
  5. A "Create account" button built as a span with an onclick handler (2.1.1).
  6. A fixed 110 px cookie banner covering focused items (2.4.11).
  7. A "Billing email (same as above)" field asking for the email again (3.3.7).

The tools also did not flag the drag-only list of dashboard widgets (2.5.7) or the #dddddd input borders at 1.36:1 (1.4.11).

Twelve seeded accessibility issues: five caught by axe-core and Lighthouse, seven found only by manual checks
My 12 seeded issues: 5 caught by axe-core and Lighthouse, 7 found only by manual checks.

Then I ran a scripted keyboard pass at a 1280x600 viewport. Tab went through Full name, Email, Confirm email, Billing email, the icon button, a "click here" link, the three social links and Privacy, then back to the page. The "Create account" span never received focus. All three social links were fully hidden behind the fixed banner when focused, and no control showed a focus outline.

Deque says in the axe-core README (read in November 2026): "With axe-core, you can find on average 57% of WCAG issues automatically." That is Deque's own claim about its engine, and my single page is not comparable to it. The lesson I take from the test is practical: run an automated scan, then a keyboard pass, then a screen reader check. I did not run a screen reader in this test, so I can't tell you what it would have added, but I would put it in your process.

How to build WCAG into your product process

Put WCAG 2.2 Level AA checks into six points of your process: discovery, design review, handoff, build, QA and a periodic audit. The later a problem is found, the more it costs to change, so most of the work belongs at the start.

  • Discovery: write down the target (for example, WCAG 2.2 Level AA) and which legal baseline applies to you.
  • Design review: check contrast values, target sizes, focus states, error text and reflow on the actual designs, not on the shipped build.
  • Handoff: note focus order, keyboard behavior, error messages and contrast values next to the visual specs.
  • Build: use shared components that already carry visible focus, error states and minimum target sizes.
  • QA: run the automated scan, then a keyboard pass, then a screen reader check on the key flows.
  • Periodic audit: schedule one after major releases. See what a full UX audit covers to decide how deep to go.

Questions to ask an outside UX team

If you hire outside help, these five questions show how they work in practice:

  1. Which WCAG version and level do you design to, and how do you decide?
  2. Can you show where focus states, error messages and target sizes appear in your design files?
  3. How do you check a design before code exists: contrast values, 24 by 24 CSS pixel targets, reflow at 320 CSS pixels?
  4. What do you test by hand besides an automated scan?
  5. How do you document accessibility decisions so engineers and QA can verify them?

If the only answer is that they run a contrast plugin, you have learned something useful.

What about WCAG 3 and the new WCAG standards for 2026?

WCAG 2.2 is the latest published version of WCAG 2 on the W3C pages I read in November 2026, and the W3C WCAG 3 Introduction describes WCAG 3 as an "incomplete draft that will change". The W3C published updated drafts in September 2026.

The W3C says WCAG 3 is intended to become a W3C Standard "in a few years". The drafts use conformance tags (Physical harm, Risk, Barrier and Friction), which differ from the A, AA and AAA levels you know from WCAG 2.

For product teams the advice is simple. The W3C says: "The best way to prepare for WCAG 3 in the future, is to meet WCAG 2.2 success criteria now." I would not plan a redesign around a draft. Meet 2.2 AA, and revisit when the W3C publishes a finished standard.

Frequently Asked Questions

What are the WCAG guidelines for UI/UX accessibility?

WCAG 2.2 has 13 guidelines under 4 principles: Perceivable, Operable, Understandable and Robust. Each guideline has testable success criteria at Level A, AA or AAA. For UI and UX work, the ones you touch most are contrast, focus, labels, error messages and target size.

Is WCAG 2.1 or 2.2 required?

It depends on the rule that applies to you. The ADA Title II web rule for State and local governments requires WCAG 2.1 Level AA, and Section 508 incorporates WCAG 2.0 Level AA. The W3C says WCAG 2.2 does not deprecate or supersede 2.1 and encourages using the latest version, so I'd design to 2.2 AA. This is not legal advice.

What are the new WCAG standards for 2026?

WCAG 2.2 is the latest published version of WCAG 2 on the W3C pages I read in November 2026. WCAG 3 is an incomplete draft that will change, and the W3C published updated drafts in September 2026. The W3C says the way to prepare is to meet WCAG 2.2 success criteria now.

What are the key guidelines for accessibility in UX design?

Check text contrast (4.5:1, or 3:1 for large-scale text) and non-text contrast (3:1). Do not use color as the only cue, give fields labels, and describe errors in text. Keep keyboard focus visible and not hidden, make targets at least 24 by 24 CSS pixels, and make layouts reflow at 320 CSS pixels width.

Conclusion

WCAG 2.2 Level AA gives product teams clear, testable limits for contrast, focus, labels, errors and target size, and design owns many of them. A scan is a start, not a verdict: in my test (Own test, November 2026) it caught 5 of 12.

Your next step is small. Pick one live flow, such as sign-up, and run a scan, a keyboard pass and a screen reader check this week. Then write down who owns each fix.

Read more