WCAG criteria and accessibility rules, with real failure rates
See which WCAG criteria actually fail on real websites and how often, and which of them automated testing can detect at all.
This covers all 86 success criteria in WCAG 2.2, at Levels A, AA and AAA. (WCAG 2.2 lists 87, but 4.1.1 Parsing is obsolete and no longer tested.) We publish the limits of automation because they matter: across 609 website scans and 941 PDF documents, only 20 of the 86 criteria produced any automated signal. A site can pass every automated check we run and still fail criteria that need a person to evaluate.
Want these checks on your own site? A free scan reports every criterion automation can reach on a web page, with the exact elements that fail.
Run a free scanWhat can automation actually test?
Automated testing speeds up an accessibility review. It does not replace the person doing it.
measured20 of 86
Our scans detected failures of this criterion, on websites, in PDFs or both. The rate is a floor: it counts only the parts an engine can see.
proxy only8 of 86
Related checks can bound it but cannot confirm it, so we publish no failure rate.
needs a person58 of 86
No automated check can establish it, ours included. Only a human review can say whether a page passes.
Highest measured failure rates on websites
Share of the 609 scanned sites that fail each, across our five research cohorts. This is a convenience sample, not a representative estimate of the web. Rates for the 941 county government PDFs are a separate population, listed per criterion below and labelled as such.
- 4.1.2 Name, Role, ValueLevel A60.4% of sites
- 1.4.3 Contrast (Minimum)Level AA50.6% of sites
- 2.4.4 Link Purpose (In Context)Level A34% of sites
- 1.1.1 Non-text ContentLevel A32% of sites
- 1.4.10 ReflowLevel AA30% of sites
- 1.3.1 Info and RelationshipsLevel A28.4% of sites
- 2.1.1 KeyboardLevel A24.6% of sites
- 2.4.1 Bypass BlocksLevel A22.7% of sites
Explore all 86 criteria
Level
Principle
Testing
Showing 86 of 86 criteria
1.1 Text Alternatives
Perceivable
1.2 Time-based Media
Perceivable
- 1.2.1 Audio-only and Video-only (Prerecorded)Level Aneeds a person
- 1.2.2 Captions (Prerecorded)Level Aneeds a person
- 1.2.3 Audio Description or Media Alternative (Prerecorded)Level Aneeds a person
- 1.2.4 Captions (Live)Level AAneeds a person
- 1.2.5 Audio Description (Prerecorded)Level AAneeds a person
- 1.2.6 Sign Language (Prerecorded)Level AAAneeds a person
- 1.2.7 Extended Audio Description (Prerecorded)Level AAAneeds a person
- 1.2.8 Media Alternative (Prerecorded)Level AAAneeds a person
- 1.2.9 Audio-only (Live)Level AAAneeds a person
1.3 Adaptable
Perceivable
- 1.3.1 Info and RelationshipsLevel Ameasured28.4% of sites90.5% of PDFs
- 1.3.2 Meaningful SequenceLevel Aneeds a person
- 1.3.3 Sensory CharacteristicsLevel Aneeds a person
- 1.3.4 OrientationLevel AAneeds a person
- 1.3.5 Identify Input PurposeLevel AAmeasured0.7% of sites
- 1.3.6 Identify PurposeLevel AAAneeds a person
1.4 Distinguishable
Perceivable
- 1.4.1 Use of ColorLevel Ameasured7.9% of sites
- 1.4.2 Audio ControlLevel Aneeds a person
- 1.4.3 Contrast (Minimum)Level AAmeasured50.6% of sites
- 1.4.4 Resize TextLevel AAmeasured12.2% of sites
- 1.4.5 Images of TextLevel AAneeds a person
- 1.4.6 Contrast (Enhanced)Level AAAproxy only
- 1.4.7 Low or No Background AudioLevel AAAneeds a person
- 1.4.8 Visual PresentationLevel AAAneeds a person
- 1.4.9 Images of Text (No Exception)Level AAAneeds a person
- 1.4.10 ReflowLevel AAmeasured30% of sites
- 1.4.11 Non-text ContrastLevel AAmeasured4.9% of sites
- 1.4.12 Text SpacingLevel AAproxy only
- 1.4.13 Content on Hover or FocusLevel AAneeds a person
2.1 Keyboard Accessible
Operable
2.2 Enough Time
Operable
2.3 Seizures and Physical Reactions
Operable
2.4 Navigable
Operable
- 2.4.1 Bypass BlocksLevel Ameasured22.7% of sites
- 2.4.2 Page TitledLevel Ameasured1.8% of sites91.4% of PDFs
- 2.4.3 Focus OrderLevel Aproxy only
- 2.4.4 Link Purpose (In Context)Level Ameasured34% of sites
- 2.4.5 Multiple WaysLevel AAneeds a person
- 2.4.6 Headings and LabelsLevel AAneeds a person
- 2.4.7 Focus VisibleLevel AAmeasured18.7% of sites
- 2.4.8 LocationLevel AAAneeds a person
- 2.4.9 Link Purpose (Link Only)Level AAAproxy only
- 2.4.10 Section HeadingsLevel AAAproxy only
- 2.4.11 Focus Not Obscured (Minimum)Level AAproxy only
- 2.4.12 Focus Not Obscured (Enhanced)Level AAAneeds a person
- 2.4.13 Focus AppearanceLevel AAAproxy only
2.5 Input Modalities
Operable
- 2.5.1 Pointer GesturesLevel Aneeds a person
- 2.5.2 Pointer CancellationLevel Aneeds a person
- 2.5.3 Label in NameLevel Aproxy only
- 2.5.4 Motion ActuationLevel Aneeds a person
- 2.5.5 Target Size (Enhanced)Level AAAneeds a person
- 2.5.6 Concurrent Input MechanismsLevel AAAneeds a person
- 2.5.7 Dragging MovementsLevel AAneeds a person
- 2.5.8 Target Size (Minimum)Level AAneeds a person
3.1 Readable
Understandable
3.2 Predictable
Understandable
3.3 Input Assistance
Understandable
- 3.3.1 Error IdentificationLevel Aneeds a person
- 3.3.2 Labels or InstructionsLevel Aneeds a person
- 3.3.3 Error SuggestionLevel AAneeds a person
- 3.3.4 Error Prevention (Legal, Financial, Data)Level AAneeds a person
- 3.3.5 HelpLevel AAAneeds a person
- 3.3.6 Error Prevention (All)Level AAAneeds a person
- 3.3.7 Redundant EntryLevel Aneeds a person
- 3.3.8 Accessible Authentication (Minimum)Level AAneeds a person
- 3.3.9 Accessible Authentication (Enhanced)Level AAAneeds a person
4.1 Compatible
Robust
Accessibility checks beyond WCAG
Not every accessibility problem maps to a WCAG success criterion. These 22 checks find structural problems that decide whether assistive technology can navigate a page efficiently rather than merely read it. Failing them is not failing WCAG, but they still matter to the people using the page.
- Some content sits outside page landmarksregion63.1% of sites
- Headings are out of orderheading-order49.3% of sites
- Page regions share the same namelandmark-unique41.5% of sites
- The page is missing a main headingpage-has-heading-one36.6% of sites
- The page has no clearly marked main contentlandmark-one-main24.5% of sites
- An element uses a role it isn't allowed to havearia-allowed-role23.6% of sites
- There's an empty headingempty-heading18.4% of sites
- An image repeats the text next to itimage-redundant-alt9.5% of sites
- The page has more than one banner landmarklandmark-no-duplicate-banner7.7% of sites
- A form field is labelled only by a tooltiplabel-title-only6.1% of sites
- The banner landmark sits inside another landmarklandmark-banner-is-top-level4.8% of sites
- The main landmark sits inside another landmarklandmark-main-is-top-level4.8% of sites
- An element has conflicting accessibility rolespresentation-role-conflict4.8% of sites
- The page has more than one main landmarklandmark-no-duplicate-main4.1% of sites
- Keyboard navigation order is disruptedtabindex4.1% of sites
- A dialog has no readable namearia-dialog-name3.9% of sites
- An aside sits inside another landmarklandmark-complementary-is-top-level3.9% of sites
- The footer landmark sits inside another landmarklandmark-contentinfo-is-top-level3.4% of sites
- The page has more than one footer landmarklandmark-no-duplicate-contentinfo3.3% of sites
- The page limits how far it can be zoomedmeta-viewport-large2.5% of sites
- A skip link points to a target that doesn't existskip-link1.3% of sites
- A table header cell is emptyempty-table-header0.8% of sites
What this does not measureLimits of these figures, and where the corpus came from
- These are five targeted cohorts, not a random sample of the web. Per-cohort rates are the figure to quote; the pooled rate is a convenience sample.
- Denominators count sites that produced a result. Sites that were blocked or errored are excluded rather than counted as passing.
- Occurrence counts are floors: the engine records at most 25 occurrences of a rule per page.
- The EU and Indian cohorts retained only the primary success criterion per finding, so a rule mapping to several criteria contributes to one of them in those two cohorts and to all of them in the other three.
| Cohort | Scanned | Attempted |
|---|---|---|
| Shopify DTC stores | 75 | 75 |
| US local government | 158 | 221 |
| US healthcare providers | 208 | 218 |
| EU e-commerce | 107 | 170 |
| Indian financial services | 61 | 72 |
Corpus generated 2026-09-18.