Core requirements
Proposed requirements for the draft’s conformance level.
REFERENCE / THE NEXT GENERATION
The developing guidance, made easier to explore. Find a topic, understand the intent, and open the exact draft wording. Follow a practical review activity alongside each guideline.
W3C WORKING DRAFT
10 September 2026 snapshot
Source checked 15 September 2026.
WCAG 3 is incomplete and can change. Use this page for learning and draft review. Continue using WCAG 2.2 for current technical evaluations.
Read the dated W3C draft →This snapshot contains 12 topic categories, 46 guideline headings and 216 draft provisions. Every listed provision is at Developing status. Figure captions and Control text are headings without published provisions in this snapshot.
Proposed requirements for the draft’s conformance level.
Proposed requirements that extend the core.
Documented processes, such as review or usability testing.
An advisory provision: a prominent way to return to the start.
Guideline summaries and the “Explore this guideline” activities are original A11y.mv learning material. Expand an entry for a W3C draft excerpt, including W3C’s own notes and examples; follow its source for definitions and context. Guideline numbers belong to this draft and are not WCAG 2.2 success-criterion numbers. Activity completion does not establish conformance.
FIND / READ / EXPLORE
216 draft provisions across 46 guideline headings.
Try a shorter phrase or reset the search and filters. The two empty guideline headings appear only with All provisions selected.
GUIDELINE 2.1.1
An image can explain something, carry out an action or simply decorate a page. Its text alternative should serve that purpose, while decoration should not interrupt assistive technology users.
Choose an image in its real page context. Describe what a reader needs from it, then compare that purpose with the text exposed to assistive technology.
A link showing a printer has the name ‘Print this receipt’; the decorative border around it adds no announcement.
W3C draft excerpt · 10 September 2026
Non-decorative images are detectable.
Applies when content includes non-decorative images.
W3C draft excerpt · 10 September 2026
Text alternatives are available for non-decorative images.
W3C draft excerpt · 10 September 2026
Text alternatives for non-decorative images convey the equivalent purpose to the image.
GUIDELINE 2.1.2
This snapshot includes a Figure captions guideline heading but no requirements yet. The topic can support discussion, but there is no draft test here to apply.
Discuss what a reader learns from the caption below a photograph or chart, and how it relates to the image alternative. Treat the discussion as a design exercise.
A chart caption might explain the survey period while the text alternative identifies the chart and points to its detailed explanation.
No published provisions under this heading in the 10 September 2026 snapshot. This is an open area of work, not an empty requirement to pass.
GUIDELINE 2.1.3
Meaningful content can appear in forms other than ordinary text or images. An equivalent text description helps people understand it through the reading method they use.
List meaningful non-text content in one task, such as a canvas diagram or an audio signal. Identify what each communicates and where a person can get that information as text.
A canvas diagram of a delivery route is accompanied by a text list of stops in the same order.
W3C draft excerpt · 10 September 2026
All non-text content that is not decorative includes a programmatically determinable equivalent text alternative.
GUIDELINE 2.1.4
A transcript makes media available as readable text. Depending on the content, that includes dialogue, useful sounds, speaker changes and visual information needed to follow the meaning.
Read a transcript without playing the media, then compare it with the recording. Note missing information, unclear speaker changes and whether the transcript is easy to find beside the player.
A cooking lesson transcript includes the spoken instructions and the ingredient quantities displayed only on screen.
W3C draft excerpt · 10 September 2026
A text transcript is adjacent to audio and video content.
Except when
W3C draft excerpt · 10 September 2026
Dialogue transcripts are available for all prerecorded audio and video content.
Except when
W3C draft excerpt · 10 September 2026
Dialogue transcripts are available for all live audio and video content.
Except when
W3C draft excerpt · 10 September 2026
Equivalent transcripts are available for audio and video content.
Except when the audio or video has no spoken content and serves only as background.
W3C draft excerpt · 10 September 2026
Descriptive transcripts are provided for prerecorded audio and video content.
Except when
W3C draft excerpt · 10 September 2026
Speakers are identified understandably within all transcripts.
Applies when there are multiple speakers in the video.
Except when
Initially using “Maya Angelou” within the context of a story regarding poetry and then using “Angelou”.
W3C draft excerpt · 10 September 2026
When more than one language is spoken in audio content, the language spoken by each speaker is identified in all transcripts.
Except when words are used incidentally.
W3C draft excerpt · 10 September 2026
Sounds needed to understand the media are identified or described in transcripts.
Applies when there is sound.
Except when the video is decorative.
W3C draft excerpt · 10 September 2026
Visual information needed to understand the media is described in transcripts.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.1.5
Captions let people follow speech and meaningful sounds while media plays. Timing, speaker identification, placement and display choices all affect whether they are useful.
Review a representative clip with its captions, including a busy scene or speaker change. Check the information, timing and any important visuals the captions cover; explore the available settings.
During a recorded interview, captions identify an off-screen speaker and describe an alarm that explains why the conversation stops.
W3C draft excerpt · 10 September 2026
The appearance of captions, including associated visual indicators, is adaptable including font size, font weight, font style, font color, background color, background transparency, and placement.
W3C draft excerpt · 10 September 2026
W3C draft excerpt · 10 September 2026
W3C draft excerpt · 10 September 2026
W3C draft excerpt · 10 September 2026
W3C draft excerpt · 10 September 2026
W3C draft excerpt · 10 September 2026
W3C draft excerpt · 10 September 2026
Captions are synchronized with the audio content of synchronized media.
W3C draft excerpt · 10 September 2026
In 360-degree digital environments, captions remain directly in front of the user.
Applies when the position of the captions is controlled by the user.
W3C draft excerpt · 10 September 2026
In 360-degree digital environments, the direction of a sound or speech is indicated when audio is heard from outside the current view.
W3C draft excerpt · 10 September 2026
Speakers are identified understandably within all captions.
Applies when there are multiple speakers in the video.
Except when
Initially using “Maya Angelou” within the context of a story regarding poetry and then using “Angelou”.
W3C draft excerpt · 10 September 2026
W3C draft excerpt · 10 September 2026
Sounds needed to understand the media are identified or described in captions.
Applies when there is sound.
Except when the video is decorative.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.1.6
Audio description explains visual information that the soundtrack does not convey. Its wording and timing help a listener follow the story without losing dialogue or important sounds.
Compare the soundtrack with a visual review of the video. List meaningful information missing from the audio and discuss where description could fit or need a longer pause.
In a safety demonstration, a description explains which valve the presenter closes before the next spoken instruction.
W3C draft excerpt · 10 September 2026
Audio descriptions are available in prerecorded video for visual content needed to understand the media.
WCAG 3 needs to specify how to handle video content with audio that does not include gaps to insert audio descriptions. Two possible solutions are providing an exception that allows the content author(s) to use descriptive transcripts instead or requiring content authors to provide an extended audio description.
Except when the video content is an alternative for text and is clearly labeled as such.
W3C draft excerpt · 10 September 2026
The information conveyed by audio descriptions is equivalent to the visual content needed to understand the media.
W3C draft excerpt · 10 September 2026
Audio descriptions are synchronized with video content without overlapping dialogue and meaningful audio content.
Except when there are no audio descriptions.
W3C draft excerpt · 10 September 2026
Audio descriptions are available in live video for visual content needed to understand the media.
W3C draft excerpt · 10 September 2026
The video pauses to extend the audio track and provides an extended audio description to describe visual information needed to understand the media.
Applies when the existing pauses in a soundtrack are not long enough.
W3C draft excerpt · 10 September 2026
The information conveyed by extended audio descriptions is equivalent to the visual content needed to understand the media.
W3C draft excerpt · 10 September 2026
A mechanism is available that allows users to change the audio description language if multiple languages are available.
W3C draft excerpt · 10 September 2026
A mechanism is available to turn audio descriptions on and off.
Except when audio descriptions are hard-coded into the audio or video content.
W3C draft excerpt · 10 September 2026
Speakers are identified understandably within all audio descriptions.
Applies when there are multiple speakers in the video.
Except when
Initially using “Maya Angelou” within the context of a story regarding poetry and then using “Angelou”.
W3C draft excerpt · 10 September 2026
When more than one language is spoken in audio content, the language spoken by each speaker is identified in all audio descriptions.
Except when words are used incidentally.
W3C draft excerpt · 10 September 2026
Sounds needed to understand the media are identified or described in captions and transcripts.
Applies when there is sound.
Except when the video is decorative.
W3C draft excerpt · 10 September 2026
Visual information needed to understand the media is described in audio descriptions.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.1.7
Sign language interpretation provides access for people who use the sign language of the intended audience. Recorded media and live events need planning for interpretation and how people reach it.
Identify the intended audiences and relevant sign languages with people from those communities. Review how interpretation would be commissioned, presented and found in a recording or live event.
An event registration page explains which sign language interpretation will be available and how to request further communication support.
W3C draft excerpt · 10 September 2026
Sign language interpretation is provided for all prerecorded audio content in the primary sign language that is most appropriate for each intended audience or region.
Except when
Audio description is an example of audio content that is an alternative for visual content.
W3C draft excerpt · 10 September 2026
A mechanism is available to show and hide sign language interpretation.
Except when
Hard-coded sign language interpretation would include video content where the interpreter is standing next to the speaker in the video.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.1.8
Information becomes harder to access when it depends on a single sensory cue. Consider meaning carried only by hue, sound, visual depth or the apparent direction of audio.
Find a status, instruction or data display and list the cues that explain it. Explore what information remains when someone cannot use one of those cues.
A delivery status uses the word ‘Delayed’ and a distinct symbol alongside its color.
W3C draft excerpt · 10 September 2026
Information is not conveyed by hue alone.
Information conveyed includes but is not limited to presenting data or meaning, indicating an action, prompting a response, distinguishing between items, conveying boundaries. Artistic expression is not part of information conveyed.
Except when
W3C draft excerpt · 10 September 2026
Parts of graphical objects required to understand the content meet a minimum contrast ratio test
Except when a particular presentation of graphical objects is essential to the information being conveyed.
W3C draft excerpt · 10 September 2026
Information is not conveyed through visual depth perception alone.
W3C draft excerpt · 10 September 2026
Information is not conveyed by sound alone.
Except when content is audio-based media.
Information conveyed includes but is not limited to presenting data or meaning, indicating an action, prompting a response, distinguishing between items, conveying boundaries. Artistic expression is not part of information conveyed.
W3C draft excerpt · 10 September 2026
Information is not conveyed by spatial audio alone.
Information conveyed includes but is not limited to presenting data or meaning, indicating an action, prompting a response, distinguishing between items, and conveying boundaries. Artistic expression is not part of information conveyed.
GUIDELINE 2.1.9
The choice of player affects whether people can use captions, descriptions and their available settings. This guideline currently contains assertions about selecting players that support appropriate media alternatives.
Create a player selection brief from the media alternatives your audience needs. Try those features in the proposed player and record which are supported, unavailable or still unverified.
Before publishing a course, the team checks that its embedded player exposes the supplied captions and described audio track.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Applies when a video is used that does not play in standard browsers.
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Applies when an audio format is used that does not play in standard browsers.
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.2.1
Readable text depends on its presentation and on people being able to adjust it. This draft explores spacing, type, size and color, with several values and the contrast method still undecided.
Try changing text size, spacing and colors in a sample page. Record where text disappears, overlaps or becomes difficult to follow; keep these observations separate from any WCAG 2.2 assessment.
A reader increases text size and changes the page colors, and the article and its controls remain usable.
W3C draft excerpt · 10 September 2026
The default/authored presentation of blocks of text meets the minimum @@ [values to be determined] for:
If you are aware of research in this area, especially involving non-Latin scripts, please email [email protected].
W3C draft excerpt · 10 September 2026
The default/authored presentation of text style property meets the minimum @@ [values to be determined] for:
If you are aware of research in this area, especially involving non-Latin scripts, please email [email protected].
W3C draft excerpt · 10 September 2026
The default visual presentation of text meets @@[contrast measure to be determined].
Applies when text is presented, including text embedded in an image format.
Except when the text is:
Transparency can cause testing issues, but should be tested as the rendered color.
The contrast algorithm used in WCAG 3 is yet to be determined. For this draft, the requirement assumes the algorithm will include a size/weight factor. If the algorithm does not include size/weight, it will need to be added to this requirement text.
A separate requirement may be needed if red/green color vision deficiency (CVD) is not accounted for within the contrast algorithm.
W3C draft excerpt · 10 September 2026
The presentation of blocks of text can be adjusted, without loss of content or functionality, to meet the @@ [values to be determined] for:
The requirement is that the text is manipulable and the style attributes can be overridden.
Except when the style attribute is hard-coded, such as raw text that is capitalized or hyphenated.
If you are aware of research in this area, especially involving non-Latin scripts, please email [email protected].
W3C draft excerpt · 10 September 2026
The presentation of text style properties can be adjusted, without loss of content or functionality, to meet the @@ [values to be determined] for:
The requirement is that the text is manipulable and the style attributes can be overridden.
Except when the text style property is hard-coded, such as raw text that is capitalized or hyphenated.
If you are aware of research in this area, especially involving non-Latin scripts, please email [email protected].
W3C draft excerpt · 10 September 2026
Text can be increased in size to at least 200% of the platform’s default body-text size.
Except when the same text is available elsewhere in the page/view which can be increased to at least 200% of the platform’s default body-text size.
W3C draft excerpt · 10 September 2026
The foreground and background color of text can be adjusted without losing content or functionality.
The requirement is that the text is manipulable and the colors can be overridden. That could be achieved by the user-agent (including operating system, browser, and assistive technology), or provided by the content author.
Applies when text is presented, including text embedded in an image format.
Except when the text is:
W3C draft excerpt · 10 September 2026
The default/authored presentation of blocks of text meets the enhanced @@ [values to be determined] for:
If you are aware of research in this area, especially involving non-Latin scripts, please email [email protected].
W3C draft excerpt · 10 September 2026
The default/authored presentation of text style properties meets the enhanced @@ [values to be determined] for:
If you are aware of research in this area, especially involving non-Latin scripts, please email [email protected].
W3C draft excerpt · 10 September 2026
The default visual presentation of text meets @@[contrast measure to be determined, at a higher level than the core requirement for Text contrast sufficient (minimum)].
The contrast algorithm used in WCAG 3 is yet to be determined. For this draft, the requirement assumes the algorithm will include a size/weight factor. If the algorithm does not include size/weight, it will need to be added to this requirement text.
Applies when text is presented, including text embedded in an image format.
Except when the text is:
W3C draft excerpt · 10 September 2026
Content that is exported, saved, or printed retains user-applied text-appearance customizations.
Applies when the page/view can be exported, saved, or printed.
Examples of interoperable formats
Examples of non-interoperable formats
GUIDELINE 2.2.2
Reading software needs access to text and information about its language. Clear context for numbers also helps it convey dates, times and other quantities meaningfully.
Review a page containing another language, a date and a measurement. Inspect the available text and language information, then listen with reading software and note ambiguity.
An appointment shows ‘18 October 2026, 2 pm’ instead of leaving readers to interpret an ambiguous numeric date and time.
W3C draft excerpt · 10 September 2026
All visible text has a programmatically determinable equivalent.
Except when making visible text programmatically determinable would lead to duplication within the view.
W3C draft excerpt · 10 September 2026
The human language of all content within the view is programmatically determinable.
Except when
W3C draft excerpt · 10 September 2026
Numerical information includes sufficient context in written text and a programmatic equivalent to avoid confusion when presenting dates, temperatures, time, and Roman numerals.
Numerical metadata is information that provides context about the numbers presented. This context helps users understand what the numbers represent and how they should be read. Without these cues, numbers can be ambiguous or misleading, making it harder for users to understand the intended meaning—especially across different regions, disciplines, or assistive technologies.
GUIDELINE 2.2.3
Clear wording helps people understand information and decide what to do. The draft explores explanations, summaries and language-sensitive writing, supported by editorial review and useful visual aids.
Give a short section to someone unfamiliar with the topic. Ask them to explain its main point and next action, then review confusing terms, unexplained abbreviations and complex sentences.
A benefits page starts with a brief overview and explains a specialist term when it first appears.
W3C draft excerpt · 10 September 2026
Explanations of abbreviations are available when first used.
Except when the abbreviation is:
Abbreviation is introduced after first expanded use of the term.
“He has Avoidant/Restrictive Food Intake Disorder (ARFID).”
Abbreviation is linked to a glossary entry or tooltip.
“He has [ARFID].”
W3C draft excerpt · 10 September 2026
Explanations or unambiguous alternatives are available in text content for non-literal language, such as idioms and metaphors.
Except when text content is:
Translation software and other tools can aid content authors in identifying non-literal language.
W3C draft excerpt · 10 September 2026
A summary is available for long-form text content and:
Research is needed to determine the number of words that trigger the summary requirement and whether this threshold varies for different languages. If you are aware of research in this area, please email [email protected].
Applies when a page/view with continuous long-form text content that is organized in paragraphs and has 300 or more words.
Except when long-form text content continues on multiple pages/views, only the first page/view requires a summary.
W3C draft excerpt · 10 September 2026
Common words are used, and definitions are available for uncommon words.
Applies when human languages have more than 1,500 words.
Except when
This is not a core requirement because a list of common words would not cover terms that are known by specific audiences, such as accounting terms on an accounting site. However, in future guidance for policymakers, it is an example of a supplemental requirement that could be made mandatory for public service and education providers.
Lists of common words are called high-frequency corpora. They exist for many languages including Arabic, Hindi, Mandarin, and Russian as well as American English, British English, and Canadian English.
Research shows that using common words and defining uncommon words improves understanding. Making Content Usable for People with Cognitive and Learning Disabilities recommends using the 1,500 highest-frequency words or phrases because people with severe language impairments are most likely to know these terms. However, more research is needed to confirm if the same threshold applies to many languages for distinguishing common from uncommon words.
W3C draft excerpt · 10 September 2026
Diacritics required to identify the correct meaning of each word are available.
Applies when a human language has a version that removes diacritics for proficient readers.
A diacritic is a small mark that is added to a letter or character that changes how it is pronounced or what it means. Diacritics may appear above, below, within, or between letters or characters.
Hebrew and Arabic are examples of human languages that omit diacritics for proficient readers.
The Hebrew word אמר, with three consonants and no diacritics
The Hebrew word אָמַר, with diacritics under two of the three consonants
The Hebrew word יֹאמַר, with diacritics in front of the first consonant and under another
The Hebrew word אֹמֶר, with a different set of diacritics above one consonant and under another
W3C draft excerpt · 10 September 2026
Sentences do not include nested clauses.
Except when text content:
The sentence:
“The teacher was surprised that the student who did well on yesterday’s test was absent today.”
Includes the nested clause “who did well on yesterday’s test” of the clause “was surprised that the student … was absent today”.
This could be written without the nested clause as:
“The teacher was surprised that the student was absent today. This was unexpected because he did well on yesterday’s test.”
W3C draft excerpt · 10 September 2026
Sentences do not include unnecessary words.
Except when text content is:
Automated tools can help content authors identify unnecessary words in many languages, including Arabic, English, Hindi, Mandarin, and Russian.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.3.1
A visible focus indicator shows which control will respond to keyboard input. Custom indicators need careful consideration against the backgrounds and states where they appear.
Move through a page with the keyboard in its available themes. Note any point where you lose track of the active control, and compare custom styling with the browser default.
The focus outline on a checkout button remains easy to find against both the page background and the button fill.
W3C draft excerpt · 10 September 2026
The focusable item uses the user agent default focus indicator.
W3C draft excerpt · 10 September 2026
If a custom focus indicator is used, it has sufficient adjacent contrast and change of contrast.
Applies when the user agent’s default focus indicator is replaced by a custom focus indicator.
W3C draft excerpt · 10 September 2026
If a custom focus indicator is used, it has sufficient size and adjacency.
Applies when the user agent’s default focus indicator is replaced by a custom focus indicator.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.3.2
People need to locate a pointer or understand what their touch or gaze has selected. The draft considers visibility, activation feedback and control over custom pointer styling.
Review a pointer or touch interaction on its intended platform. Note whether people can follow the pointer or selection feedback, and whether a custom pointer overrides useful personal settings.
A touch kiosk visibly highlights the selected appointment before showing the next screen.
W3C draft excerpt · 10 September 2026
There is a visible indication of the activation of an interactive element when selected by the pointer.
Applies when the platform does not use a visible pointer indicator.
This is primarily aimed at touch-interfaces and VR where you don’t have a pointer indicator, but do need to know when something has been selected.
W3C draft excerpt · 10 September 2026
The user can choose to always have a visible pointer indicator.
Applies when the platform does not use a visible pointer indicator.
This is primarily aimed at eye-tracking and touch-screens, where it is useful for the user to be able to have a visible indicator, but it wouldn’t be universal. This aims to avoid it being in the way for some users.
W3C draft excerpt · 10 September 2026
The default pointer meets the @@[non-text-contrast] requirement, and is at least as large as the platform default.
Applies when the pointer indicator appearance can be adjusted from the platform default.
There can be multiple types of pointer indicator. For example, arrow, hand, caret. The size requirement applies to whichever type of indicator would be the default for that scenario.
W3C draft excerpt · 10 September 2026
The user can ensure that the appearance of the pointer is not overridden by the authored interface.
Except when changing the pointer appearance is essential.
W3C draft excerpt · 10 September 2026
There is a visible pointer indicator.
Except when
Examples of pointers which do not always show the pointer indicator:
W3C draft excerpt · 10 September 2026
The pointer indicator is always visible.
Except when
W3C draft excerpt · 10 September 2026
Provide a more visible pointer indicator than the platform default. The enhanced pointer indicator can be enabled in a setting, and be visible temporarily (for a few seconds) or permanently.
GUIDELINE 2.3.3
Keyboard navigation should follow a meaningful path through relevant controls. People also need to enter a focused area, such as a dialog, and return to the wider page without getting lost.
Follow the keyboard path through a task and describe the sequence aloud. Investigate unexpected stops, hidden controls and what happens when you leave a popup or dialog.
After editing delivery details in a dialog, the user returns to the checkout section where they started.
W3C draft excerpt · 10 September 2026
The focus order does not include hidden, static, or groups of repeated interactive elements.
W3C draft excerpt · 10 September 2026
A user can focus on a content “area”, such as a modal or popup, then resume their view of all content using a limited number of steps.
W3C draft excerpt · 10 September 2026
The keyboard focus moves sequentially through content in an order and way that preserves meaning and operability.
GUIDELINE 2.3.4
Familiar controls are easier to use when their behavior matches what people expect. Repeated functions, labels and locations benefit from a consistent approach and review of platform conventions.
Compare several instances of the same control across a product. Record differences in labels, placement, activation and feedback, then decide which differences have a clear user benefit.
Every expandable section uses the same control pattern and announces whether its content is open or closed.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
Where an interactive element with the same purpose is used across pages/views, its visual position in the layout is maintained.
Except when
An e-commerce checkout is an example of where the layout might change within a longer process.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
A component library specifies a set of standardized components used in a product or products. It can include code to use, but at a minimum would define how the component is used and define pointer, keyboard, and assistive technology interactions.
Using a component library can help larger teams and organizations provide a consistent experience.
Examples of component libraries include GDS, Carbon Design System, and Lion.
Examples of platform patterns are ARIA Platform Authoring Guide (web), Apple human interface guidelines, and Android Material Design.
GUIDELINE 2.3.5
A control needs an understandable purpose, current state and any instructions needed to use it. That information should be available visually and to assistive technology, with visible labels reflected in accessible names.
Choose a form control and compare what you can see with its accessibility information. Review its name, role, value, state and any format or required-field guidance.
A ‘Delivery date’ field keeps its label visible and associates the permitted date format with the input.
W3C draft excerpt · 10 September 2026
Visual information required to identify interactive elements and states meet a minimum contrast ratio test.
Except when
W3C draft excerpt · 10 September 2026
Persistent names (including labels) that identify the purpose of the interactive element are visually and programmatically available.
Visible labels can be text or non-text, for instance icons.
W3C draft excerpt · 10 September 2026
Changes to interactive element names, roles, values, or states are visually and programmatically indicated.
W3C draft excerpt · 10 September 2026
Field constraints and conditions are available.
W3C draft excerpt · 10 September 2026
The programmatic name includes the visual label.
W3C draft excerpt · 10 September 2026
Accurate names, roles, values, and states are available for interactive elements.
GUIDELINE 2.4.1
Keyboard users need access to both controls and the content those controls reveal. Navigation, shortcuts and focus movement should let people complete a task and leave each component.
Complete a representative task using the keyboard alone, moving forward and backward. Include opening, reading and closing any menus or dialogs, and record where progress depends on another input method.
A date picker can be opened, used and closed from the keyboard, with focus returning to a useful place.
W3C draft excerpt · 10 September 2026
All components on the page/view that can be operated by pointer, audio (voice or other), gesture, camera, or other means can be operated using keyboard interface only.
W3C draft excerpt · 10 September 2026
All content that can be accessed by other input modalities can be accessed using keyboard interface only.
All content includes content made available via mechanisms including but not limited to hovers, right clicks.
Other input modalities include pointing devices, voice and speech recognition, gesture, camera, and any other means of input or control.
The “Keyboard operable” requirement allows you to navigate to all actionable elements, but if the next element is 5 screens down, you also need to be able to access all the content. Also, if the content is in expanding sections, you need to not only open them but also access all of the content, not just its actionable elements.
W3C draft excerpt · 10 September 2026
Documentation for each custom keyboard command is actively available on the page/view to which it applies, or within the applicable process.
W3C draft excerpt · 10 September 2026
Custom keyboard commands do not conflict with standard platform keyboard commands or they can be remapped.
W3C draft excerpt · 10 September 2026
When keyboard focus moves from one context to another within a page/view, whether automatically or by user request, the keyboard focus is preserved so that, when the user returns to the previous context, the keyboard focus is restored to its previous location unless that location no longer exists.
When a user closes a modal dialog or other popup, keyboard focus is returned to the element that caused the dialog or popup to open.
When there are tags, with a delete button on each tag, the keyboard focus needs to be moved to a meaningful location after a tag is deleted.
W3C draft excerpt · 10 September 2026
Components that can be activated or entered using the keyboard interface, can be deactivated or exited using a standard keyboard navigation-operation technique, standard platform keyboard commands.
Except when the non-standard keyboard navigation technique is described in the page/view or earlier in the process.
W3C draft excerpt · 10 September 2026
The keyboard focus only moves as a result of user interaction.
Except when
W3C draft excerpt · 10 September 2026
Except for skip links and other elements that are hidden but specifically added to aid keyboard navigation, tabbing does not move the keyboard focus onto content that was not visible before the tab action.
Accordions, dropdown menus, and ARIA tab panels are examples of expandable content. According to this requirement, these would not expand simply because they include an element in the tab-order contained in them. They would either not expand or would not have any tab-order elements in them.
A menu that expands when you tab to it, but then uses arrow keys to navigate in it would pass. But a menu that expands and then requires you to tab through all the newly-visible elements to navigate past it would fail.
GUIDELINE 2.4.2
Keyboard access can still be tiring when a task adds repeated stops or unfamiliar commands. This topic explores reducing that effort and explaining any special navigation.
Compare the steps needed to complete the same task with keyboard and pointer input. Look for duplicate links, avoidable repetition and commands users would have to guess.
A product card combines its image and title into one link instead of requiring two consecutive stops for the same destination.
W3C draft excerpt · 10 September 2026
Adjacent interactive elements that achieve the same outcome are not included in the page/view.
Except when that outcome is a form of dismissal.
A common pattern is having a component that includes a linked image and some linked text, where both links go to the same content. Someone using screen reading software can be disoriented from the unnecessary chatter, and a keyboard user has to navigate through more tab stops than should be necessary. Combining adjacent links that go to the same content improves the user experience.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Other input modalities include pointing devices, voice and speech recognition, gesture, camera, and any other means of input or control.
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.4.3
Pointer actions are more usable when they can be completed without precise speed, pressure or complex gestures. People also need a way to avoid or reverse an unintended activation.
Find a feature that uses dragging, repeated tapping or a timed gesture. Explore a simple input alternative and investigate what happens if the user changes their mind before activation.
A list supports moving an item with ‘Move up’ and ‘Move down’ buttons as well as dragging it.
W3C draft excerpt · 10 September 2026
At least one of the following is true for functionality that can be activated using a simple pointer input:
Except when completing the function on the down event is essential.
An example of Cancel would be dragging where there is a pickup action on button down, but it can be canceled by dropping anywhere other than the drop area.
Examples of places where action on down event may be essential include a descending-price auction or a game trigger.
W3C draft excerpt · 10 September 2026
All functionality and content available using complex pointer inputs is also available using a simple pointer input, or a sequence of simple pointer inputs that do not require timing.
Complex pointer inputs include but are not limited to:
Complex pointer inputs are not banned, but they cannot be the only way to accomplish an action.
Simple pointer input is different than single pointer input and is more restrictive than simply using a single pointer.
W3C draft excerpt · 10 September 2026
W3C draft excerpt · 10 September 2026
W3C draft excerpt · 10 September 2026
GUIDELINE 2.4.4
Speech cannot be the only practical route through a service for everyone. This topic explores alternative input, real-time text communication and compatibility with generated speech.
Map the tasks in a voice service and identify how someone could complete them without speaking. For supported voice input, plan research that includes people who communicate using generated speech.
A live customer-support service offers a real-time text conversation alongside its voice call option.
W3C draft excerpt · 10 September 2026
Content or functionality does not rely on speech alone.
Except when speech input is essential to the functionality.
W3C draft excerpt · 10 September 2026
A real-time text option is available for real-time bidirectional voice communication.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.4.5
People may combine input methods or need an alternative to a body movement, gaze or gesture. Content revealed by hover or focus also needs to remain usable and dismissible.
Try switching between the input methods a product supports during one task. Review any hover panel or gesture-only feature for lost content, unexpected focus changes or missing alternatives.
A map offers zoom buttons alongside gestures, and its information popups remain available when the pointer moves into them.
W3C draft excerpt · 10 September 2026
A mechanism is available to dismiss content that appears on pointer hover or keyboard focus without moving pointer hover or keyboard focus, unless the additional content does not obscure or replace other content.
Applies when receiving and then removing pointer hover or keyboard focus triggers additional content to become visible and then hidden, and the visual presentation of the additional content is controlled by the author and not by the user agent.
This applies to content that appears in addition to the triggering of the interactive element itself. Since hidden interactive elements that are made visible on keyboard focus (such as links used to skip to another part of a page/view) do not present additional content, they are not covered by this requirement.
Additional content controlled by the user agent includes browser tooltips created through use of the HTML title attribute.
Custom tooltips, sub-menus, and other non-modal popups that display on hover and keyboard focus.
W3C draft excerpt · 10 September 2026
If pointer hover can trigger content, then the pointer can be moved over the additional content without the additional content disappearing.
Applies when receiving and then removing pointer hover or keyboard focus triggers additional content to become visible and then hidden, and the visual presentation of the additional content is controlled by the author and not by the user agent.
This applies to content that appears in addition to the triggering of the interactive element itself. Since hidden interactive elements that are made visible on keyboard focus (such as links used to skip to another part of a page/view) do not present additional content, they are not covered by this requirement.
Additional content controlled by the user agent includes browser tooltips created through use of the HTML title attribute.
Custom tooltips, sub-menus, and other non-modal popups that display on hover and keyboard focus.
W3C draft excerpt · 10 September 2026
Content that appears on pointer hover or keyboard focus remains visible until the hover or keyboard focus trigger is removed, the user dismisses it, or its information is no longer valid.
Applies when receiving and then removing pointer hover or keyboard focus triggers additional content to become visible and then hidden, and the visual presentation of the additional content is controlled by the author and not by the user agent.
This applies to content that appears in addition to the triggering of the interactive element itself. Since hidden interactive elements that are made visible on keyboard focus (such as links used to skip to another part of a page/view) do not present additional content, they are not covered by this requirement.
Additional content controlled by the user agent includes browser tooltips created through use of the HTML title attribute.
Custom tooltips, sub-menus, and other non-modal popups that display on hover and keyboard focus.
W3C draft excerpt · 10 September 2026
Path-based gestures are not the only way of achieving any functionality.
Except when a path-based gesture is essential to the functionality.
W3C draft excerpt · 10 September 2026
The ability to switch between input methods is available at any time.
W3C draft excerpt · 10 September 2026
Functionality does not rely solely on full or gross body movement.
Except when full or gross body movement is essential to the functionality.
This includes both detection of body movement and actions to the device, such as shaking, that require body movement.
W3C draft excerpt · 10 September 2026
Content and functionality does not rely solely on eye tracking.
Except when eye tracking is essential.
This is primarily aimed at ensuring there is an alternative for people who cannot use eye tracking (but do have sight) due to eye conditions.
Some platforms may only allow eye tracking. Ideally the platforms allow additional mechanisms for control.
W3C draft excerpt · 10 September 2026
Pointer selection of elements moves the keyboard focus to that element, even if the user selects an interactive element and drags away from the element without activation.
Applies when content can interfere with pointer or keyboard focus behavior.
A user scrolls a document down six screens, then clicks on a paragraph with their pointer. The user then presses the Tab key, which moves the focus to the first interactive component after the position on the screen that was clicked, rather than from the previous position, six screens up the document.
GUIDELINE 2.4.6
Biometric authentication can exclude people who cannot provide the expected physical or voice characteristic. The draft explores another way to identify or authenticate the same person.
Review an account journey that offers face, fingerprint or voice recognition. Find the alternative route and consider whether it gives people access to the same task without the unavailable biometric.
A banking app offers an account sign-in method that does not require fingerprint recognition.
W3C draft excerpt · 10 September 2026
Biometric identification is not the only way to identify or authenticate.
Biometrics includes facial recognition software, fingerprinting, vocal patterns and other voice characteristics.
W3C draft excerpt · 10 September 2026
Voice identification is not the only way to identify or authenticate.
GUIDELINE 2.5.1
An error message should explain the problem and help the person locate what needs attention. Its placement, persistence and connection to the relevant control affect whether recovery is possible.
Trigger a realistic error in a sample form. Review how it is announced, where it appears, whether it stays available and how the user reaches the field to correct it.
A booking form keeps ‘Enter a departure date’ beside the empty field and associates the message with that input.
W3C draft excerpt · 10 September 2026
Errors that are programmatically determined are identified and the problem is described to the user in text.
W3C draft excerpt · 10 September 2026
Error messages include suggestions for corrections.
Applies when errors require corrections by the user.
Except when including suggestions would jeopardize the security or purpose of the content.
W3C draft excerpt · 10 September 2026
Error messages are visually indicated using at least two of the following:
An example of a “textual indication” would be to add “Error: ” before the error description.
W3C draft excerpt · 10 September 2026
Error messages persist at least until the error is resolved or the user dismisses them.
W3C draft excerpt · 10 September 2026
When input validation fails, the errors are visually and programmatically associated with the element that caused the error or that can resolve it.
Failing validation includes but is not limited to:
W3C draft excerpt · 10 September 2026
Error messages are visually collocated with the error source or the focus is moved to the error message and a mechanism is available to move to the input that is in error.
Applies when error messages relate to user input.
GUIDELINE 2.5.2
Good form design reduces avoidable mistakes and gives people a chance to review or correct information. Clear submission feedback helps them know whether their action succeeded.
Walk through a form with a plausible mistake. Explore the opportunity to review or correct it and confirm how the final submission status is communicated.
Before sending an application, the user can review entered details, change an answer and receive a clear submission confirmation.
W3C draft excerpt · 10 September 2026
Data entry interfaces allow for users to do at least one of the following before submission:
Except when entered data is auto-saved and/or reversible.
Editors are looking at removing the grey area that may exist in this requirement due to interpretations of the word “submission“ (pressing “Submit” in the UI or receiving information server-side).
W3C draft excerpt · 10 September 2026
Data entry interfaces notify users of submission status at the time of submission.
Applies when data submission has succeeded or failed.
W3C draft excerpt · 10 September 2026
Data entered is validated after the user enters data, either:
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
This review includes checking to make sure that we:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.6.1
Flashing, visual motion, vibration and shifting audio can cause harm or discomfort. This draft explores reducing exposure, providing control and offering access to information without triggering effects.
Inventory these effects through content and configuration review before playback. Investigate controls and equivalent alternatives; do not ask a person to expose themselves to a known trigger as a test.
A moving product demonstration also has a still-image explanation, and optional vibration feedback can be turned off.
W3C draft excerpt · 10 September 2026
Flashes are below the general flash and red flash thresholds.
Applies when content includes flashes.
Except when the flashing is essential.
If there is an accessibility supported method of setting a user-preference to prevent flashing, the content can be considered to avoid flashing if that preference is respected.
There is research into the size and frequency metrics underway. The exact values are likely to change before WCAG3 is published.
W3C draft excerpt · 10 September 2026
Flashes are below the general flash and red flash thresholds without a minimum size.
Applies when content includes flashes.
Avoiding flashing entirely is the safest option because some people who may not know they are susceptible to photosensitive seizures, and some people may have a much larger view of content than the default (zoomed in or with the display close to the eye).
W3C draft excerpt · 10 September 2026
Content does not include pseudo-motion or visual motion lasting longer than 5 seconds.
Except when the motion or pseudo-motion is essential.
W3C draft excerpt · 10 September 2026
Content does not include pseudo-motion or visual motion lasting longer than 5 seconds.
W3C draft excerpt · 10 September 2026
A warning is provided before users encounter triggers and a mechanism is available to access the same information without the triggering content.
Applies when triggers are present.
Triggers are flashing, motion lasting more than 5 seconds, and pseudo-motion.
W3C draft excerpt · 10 September 2026
Haptic feedback can be reduced or turned off.
Applies when content triggers haptic feedback.
Except when the operating system or user agent converts non-haptic feedback to haptics at user request.
W3C draft excerpt · 10 September 2026
Audio shifting designed to create a perception of motion can be paused or turned off.
Applies when content includes audio shifting.
Except when operating system or user agent triggers audio shifting.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Troubling content will vary based on culture or individual situations, and reviews should take target audiences into consideration.
GUIDELINE 2.7.1
A familiar layout helps people predict where information and actions will be. This guideline currently proposes an assertion about reviewing conventions or testing a different layout.
Compare the layout with products used for a similar task. Identify any unfamiliar arrangement and plan research to understand whether people can find their way through it.
A service team tests whether visitors can locate appointment details and the booking action in a redesigned clinic page.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.7.2
People need to understand where they are and where a process will take them. Descriptive titles, current-step information and understandable page changes can support that orientation.
Enter a service at an interior page and explain its purpose, location and next step. Follow a multi-step journey and note where the current position or return route becomes unclear.
An application page is titled ‘Employment details’ and marks that section as the current step in the process.
W3C draft excerpt · 10 September 2026
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
It is often helpful for users to understand where within a digital product they are. There are many ways to achieve this, for example, a breadcrumb. Ideally this is consistently presented throughout the conformance scope but for some pages/views it may make less sense to include. For example, including a breadcrumb trail on the homepage or on pages that sit outside the hierarchy, for example a shopping cart.
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
A list of all steps in a multi-step process is visually and programmatically available at each step.
Except when the total number of steps is unknown, or the sequence of steps depends on user actions.
W3C draft excerpt · 10 September 2026
The current step within a multi-step process is visually and programmatically indicated.
W3C draft excerpt · 10 September 2026
When content triggers a change of page/view there is a visual change within the view and programmatic notification of the change.
W3C draft excerpt · 10 September 2026
A visual and programmatically available mechanism exists that allows users to return to the starting point of the conformance scope.
Except when
Where the conformance scope is a sub-part of larger digital product, then the starting point should be the conformance scope starting point. For example, an organization’s careers website that is separate from the main website.
W3C draft excerpt · 10 September 2026
Mechanisms that return the user to the starting point of the conformance scope are available in prominent positions both programmatically and visually.
For HTML, a good programmatic positioning of such a mechanism would be early in the DOM.
GUIDELINE 2.7.3
Meaningful sections, relationships and ordering help people scan and understand content. Visual grouping and the structure exposed to assistive technology should tell a coherent story.
Review the heading outline and reading order of a long page, then compare them with its visual sections. Look for missing relationships, unclear labels and steps that lose their order.
A preparation guide uses titled sections and an ordered list for instructions that must happen in sequence.
W3C draft excerpt · 10 September 2026
Relationships of meaning between elements are conveyed programmatically.
W3C draft excerpt · 10 September 2026
Meaningful blocks of content are programmatically determinable and visually presented with sufficient surrounding space.
W3C draft excerpt · 10 September 2026
Meaningful blocks of content have a semantically appropriate label that defines their purpose.
Except when a label is not needed to understand the purpose of the content within the context of use.
W3C draft excerpt · 10 September 2026
Meaningful blocks of content are organized with a logical hierarchy of headings.
Except when the technology does not support heading levels.
W3C draft excerpt · 10 September 2026
Ordered content includes programmatically determinable markers that indicate the position of each item.
Except when the nature of the ordering of the content is presented immediately prior.
This includes lists and processes.
W3C draft excerpt · 10 September 2026
Styling is used to enhance the visual separation between meaningful blocks of content.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.7.4
New content that covers the main page can prevent people from reading or continuing a task. The draft explores providing a way to dismiss that obstruction.
Identify overlays that can appear during a visit. Review how someone dismisses each one and whether they can then reach the content it covered.
A promotional panel includes a discoverable close control that restores access to the article beneath it.
W3C draft excerpt · 10 September 2026
GUIDELINE 2.8.1
Stable structure and navigation reduce the need to relearn a site on every page. Repeated elements can stay in a recognizable relative order even when the amount of content changes.
Compare several pages in the same journey. Track repeated navigation and structural elements, looking for unexpected changes in order or labels.
A service website keeps ‘Appointments’, ‘Locations’ and ‘Help’ in the same relative order across its main pages.
W3C draft excerpt · 10 September 2026
The relative order of structural components remains consistent throughout each variation of pages/views in the conformance scope.
Applies when in a set of pages/views.
Relative order means that content can be added or removed, but repeated items are in the same order relative to each other.
GUIDELINE 2.9.1
A process can exclude people when it requires remembering, transcribing or solving a cognitive test. Allowing input assistance and another route through the process can remove that barrier.
Review a sign-in or verification journey with paste and a password manager. Identify any step that requires a cognitive test and explore whether a usable alternative is available.
A sign-in form accepts a saved password and allows a verification code to be pasted from another application.
W3C draft excerpt · 10 September 2026
Automated input of personal information from user agents, third-party tools, or paste is not prevented.
Personal information includes but is not limited to names and passwords.
W3C draft excerpt · 10 September 2026
Processes can be completed without a cognitive function test.
User authentication or login.
GUIDELINE 2.9.2
People need different amounts of time to read, decide and act. The draft explores avoiding unnecessary deadlines and explaining or adjusting time limits when they exist.
List time limits in a task and ask what makes each necessary. Review when users learn about the limit and how they can request more time or disable it where available.
A form explains its session limit before data entry and offers an accessible way to extend the session.
W3C draft excerpt · 10 September 2026
W3C draft excerpt · 10 September 2026
The completion of a process does not include time limits.
Except when the time limit is essential.
Essential time limits would include but are not limited to auctions, ticket sales, or timed exams.
Implying to a user that they will lose a benefit if they don’t act immediately is not an essential time limit.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
W3C draft excerpt · 10 September 2026
Users are informed at the start of the process or session that a time limit exists, its length, and that it can be adjusted.
Applies when pages/views within the conformance scope have a time limit.
Except when hiding the existence of the time limit is essential.
GUIDELINE 2.9.3
Misleading wording, hidden choices and artificial pressure can undermine informed decisions. The draft explores visible preselected options and reviews that include people affected by deceptive designs.
Review a signup or purchase journey for preselected extras, urgency messages and unclear choices. Ask participants to explain what they think they are agreeing to before they proceed.
A checkout clearly presents an optional recurring service and its current selection state before the customer places an order.
W3C draft excerpt · 10 September 2026
During the completion of a process, preselected options that impact finance, privacy or safety are visibly and programmatically available to the user by default.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.9.4
Returning to an earlier step or taking a break should not needlessly erase a person’s work. The draft explores preserving information, reusing earlier answers and saving progress.
Enter sample information, move backward and forward, then explore any save-and-resume feature. Record lost answers, repeated entry and what the interface tells users about retained data.
An application lets a user correct an earlier address and return to the current step with the other answers intact.
W3C draft excerpt · 10 September 2026
In a multi-step process, the interface supports stepping backwards in a process and returning to the current point without data loss.
Except when it is essential that the user cannot step back in a process.
Certain tests in education may require that the student cannot go back through previously-submitted responses and change them.
W3C draft excerpt · 10 September 2026
Information previously entered by or provided to the user that is required to be entered again in the same process is either auto-populated, or available for the user to select.
Except when
W3C draft excerpt · 10 September 2026
Data entry and other task completion processes allow saving and resuming from the current step in the task.
Except when the task completion is part of a real-time event, and no alternative to the time limit is possible.
Real-time events would include but are not limited to auctions or ticket sales.
GUIDELINE 2.9.5
A process is easier to complete when people know what to prepare and what action comes next. Instructions and required actions should be available when they help someone move forward.
Read the start of a multi-step process before filling anything in. Identify the materials and steps you expect, then compare those expectations with the actual journey.
A permit application explains the documents to prepare and identifies the action needed to continue at each stage.
W3C draft excerpt · 10 September 2026
The interface indicates when user input or action is required in order to proceed to the next step.
Applies when the user needs to complete an action in order to proceed to the next step.
This would apply when the user needs to agree to the terms and conditions before they can submit the form.
W3C draft excerpt · 10 September 2026
Information and resources that are needed to complete a multi-step process are provided at the start of the process, including the:
Applies when the user needs to complete a multi-step process.
W3C draft excerpt · 10 September 2026
The instructions needed to complete a multi-step process are available.
Applies when the user needs to complete a multi-step process.
GUIDELINE 2.9.6
Extra actions and unnecessary questions can make a process harder to finish. This guideline currently proposes an assertion about investigating that burden through inclusive usability testing.
Map every question and action in a task, then explain why each is needed. Plan sessions with people with cognitive or mental health disabilities to investigate steps that add avoidable effort.
A booking service removes a repeated account-details screen after research shows it adds work without helping users confirm their booking.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.10.1
People need understandable information about the consequences of important choices. This topic also explores how a service’s safety decisions and algorithms affect people with different disabilities.
Choose a decision involving money, privacy or security. Review what its explanation covers, when it appears and whether disability-related risks have been investigated with relevant people.
Before sharing an account’s information with another service, the page explains what will be shared and the consequences of allowing it.
W3C draft excerpt · 10 September 2026
Choices with legal, financial, privacy, or security consequences are accompanied by a description of the benefits, risks, and potential consequences when users make the choice.
W3C draft excerpt · 10 September 2026
Legal, financial, privacy, and security consequences are provided before finalizing an agreement.
Applies when entering an agreement is required.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
GUIDELINE 2.10.2
Automated decisions can disadvantage people with disabilities through their data or behavior. This guideline currently contains assertions about disability representation, usability testing and ethics review.
Map an automated decision that affects access to a service. Ask what disability perspectives informed its data and evaluation, and record the evidence still needed to investigate unequal effects.
A team reviews whether an automated interview tool disadvantages applicants who communicate differently and documents changes informed by that review.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
GUIDELINE 2.11.1
Useful help is easy to find and relates to the action a person is trying to complete. Clear instructions, explanations for disabled controls and support for difficult decisions can prevent dead ends.
Pause at a confusing step or disabled control and look for help without leaving the task. Review whether the instructions identify the action by name and explain how to proceed.
A disabled ‘Continue’ button has an explanation nearby: ‘Choose an appointment time to continue.’
W3C draft excerpt · 10 September 2026
Help is labeled consistently and is available in a consistent location relative to other content.
Applies when human contact information, a human contact mechanism, a self-help option, or a fully automated contact mechanism is available.
W3C draft excerpt · 10 September 2026
Context-sensitive help is available.
W3C draft excerpt · 10 September 2026
Information explaining why a visible interactive element is disabled is available and, if the user can take action(s) to enable the element, those action(s) are described.
When a submit button is disabled until all required fields are filled in, explain that this is the case.
W3C draft excerpt · 10 September 2026
Instructions and help do not rely on sensory characteristics.
Sensory characteristics include but are not limited to shape, color, size, visual location, orientation, or sound.
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
W3C draft excerpt · 10 September 2026
[Title, role, or organization] asserts that:
Information that needs to be included publicly:
Recommended internal documentation (Informative):
Solutions include help with completing functionality such as data entry, task completion, search, and understanding complex ideas and visualizations.
GUIDELINE 2.11.2
People need a way to tell content authors about a problem or improvement. A reachable feedback route helps a service learn about barriers that its own reviews may miss.
From a representative page, look for a way to send feedback about its content. Review whether the route itself is understandable and usable with the input methods you are evaluating.
A help article links to an accessible form where readers can report unclear or inaccessible information.
W3C draft excerpt · 10 September 2026
A mechanism is available to provide feedback to authors.
GUIDELINE 2.12.1
People bring their own reading, navigation and input tools to a product. The draft explores supporting those tools, respecting accessibility settings and allowing control over interruptions.
Review a task with relevant assistive technology and personal settings. Include reading outside keyboard focus, changing supported appearance or motion settings, and managing notifications.
A reader uses a screen reader’s reading cursor to explore an expanded section without having to focus every paragraph.
W3C draft excerpt · 10 September 2026
Content can be controlled using assistive and adaptive technology.
W3C draft excerpt · 10 September 2026
Content responds to users’ platform and user agent accessibility-related settings.
Accessibility-related user settings include font size, icon size, color scheme, magnification, and motion.
W3C draft excerpt · 10 September 2026
Assistive technologies can access content and interactions when using mechanisms that convey alternative points of regard or focus.
A virtual cursor is an example of a mechanism that conveys alternative points or regard or focus.
W3C draft excerpt · 10 September 2026
The timing or positioning of notifications can be changed, suppressed, or saved.
Applies when notifications or other interruptions are present.
Except when the notification involves an emergency or is essential.
GUIDELINE 2.12.2
This snapshot includes a Control text guideline heading but no requirements yet. Discuss the topic as an area for exploration; there is no draft test under this heading.
Ask readers which text adjustments help them and where those preferences come from. Use the separate Text appearance guideline when exploring the provisions already present in this snapshot.
A design discussion records a reader’s need to enlarge text and change spacing, without treating those notes as a Control text conformance result.
No published provisions under this heading in the 10 September 2026 snapshot. This is an open area of work, not an empty requirement to pass.
GUIDELINE 2.12.3
Content needs to remain understandable when the viewing area or device orientation changes. The draft explores reflow and acknowledges content whose meaning depends on a two-dimensional arrangement.
Review an ordinary page at a narrow viewport and in different supported orientations. Record lost controls, hard-to-follow text and areas that genuinely need a two-dimensional layout.
An article forms a readable single column on a narrow screen while a wide data table has a clearly contained scrolling area.
W3C draft excerpt · 10 September 2026
If the platform has a default orientation, content supports that orientation. If the platform does not have a default orientation, content supports both portrait and landscape orientations.
Except when
For extended reality, the platform default orientation aligns with the real world orientation.
Content does not have to re-layout or change aspect ratio in a different orientation, it just needs to display in the device orientation.
W3C draft excerpt · 10 September 2026
W3C draft excerpt · 10 September 2026
Blocks of text are legible at 320 CSS pixels in the orientation of text, without the need to scroll in the orientation of text.
Except when the meaning of text relies on a two-dimensional structure.
Two-dimensional structure might be relied upon in preformatted text such as code, poems, maps, or comics.
Other languages may have other rules around line breaking: https://r12a.github.io/scripts/script-features/index.html
W3C draft excerpt · 10 September 2026
All content fits within 320 CSS pixels in the default orientation of text without requiring scrolling in more than one direction. Sections of content within the page/view that scroll in a different direction to the page/view fit within 320 CSS pixels of the page-scrolling direction.
Except when
All block-level elements fit within a 320px inline-size without requiring scrolling in more than one direction.
GUIDELINE 2.12.4
People need control over media playback and the alternatives that support it. Searchable alternatives and chapter navigation can also help them find or revisit information.
Try stopping unexpected audio, adjusting its volume and switching available alternatives. In a longer recording, explore how a user would locate a particular topic again.
A training recording includes named chapters and a searchable transcript, with controls for its caption and description tracks.
W3C draft excerpt · 10 September 2026
A mechanism is available to pause, stop, and adjust the volume independently of the overall system volume level, of any automatically playing audio in a page/view.
Applies when notifications or other interruptions are present.
Mechanisms include controls on each instance of content, or a single app-wide control that disables audio, for example: app-wide earcons.
W3C draft excerpt · 10 September 2026
Alternatives for media can be searched and queried.
W3C draft excerpt · 10 September 2026
Closed captions and audio descriptions can be turned on and off.
W3C draft excerpt · 10 September 2026
Audio or video that lasts five minutes or longer can be navigated by chapters.
Except when the media is a piece of music that the composer has not divided into movements.
Research is needed to determine the length of time that triggers the chapter requirement and whether this threshold varies for different languages. If you are aware of research in this area, please email [email protected].
Gustav Mahler’s Symphony Number 5 is approximately 70 minutes long and is divided into five movements. Each movement would be a chapter.
Iron Maiden’s Rime Of The Ancient Mariner is approximately 14 minutes long and isn’t divided into movements. This meets the exception and would not need to be divided into chapters.
GUIDELINE 2.12.5
An interface can change without a full page load. People need to discover meaningful updates, understand focus changes and know in advance when an action will switch their device or application context.
Change a filter or complete an action that updates the page. Compare the visual feedback with the information available to assistive technology, and observe any unexpected focus or application change.
After a visitor filters a directory, an accessible status message announces the new result count while focus stays on the filter.
W3C draft excerpt · 10 September 2026
Meaningful changes in visual content are conveyed programmatically.
Applies when
Except when changes are continuous, without pause.
W3C draft excerpt · 10 September 2026
When the focus changes on-focus or automatically, the user is notified visually and programmatically.
W3C draft excerpt · 10 September 2026
When content triggers a device change or an automatic user agent change, the user is notified before the change occurs.
WHAT CHANGES / WHAT TO USE TODAY
| Question | WCAG 2.2 | WCAG 3 draft |
|---|---|---|
| What is its status? | Published W3C Recommendation. | Incomplete Working Draft; wording and model can change. |
| How is it organized? | Four principles, guidelines and success criteria. | Topic categories, guidelines, requirements, assertions and supporting methods. |
| How is conformance described? | Levels A, AA and AAA. | One proposed conformance level, with separate reporting tiers. The model remains in development. |
| Can results transfer directly? | Keep the criterion, evidence, scope and evaluation context. | No automatic conversion or one-to-one grade. Changes and additional work are expected. |
| What should I do now? | Meet applicable WCAG 2.2 criteria and test real tasks. | Explore emerging needs and contribute specific draft feedback. |
Read WAI’s WCAG 3 introduction and the dated explainer for the current direction. WAI recommends meeting WCAG 2.2 now as preparation for WCAG 3.
The September draft separates reporting from conformance. Its proposed sequence is Avoid physical harm, Foundational access, Conformance, then Bronze, Silver and Gold. The three upper tiers describe progress above conformance; their numerical thresholds remain undecided.
These labels are not the old idea of three conformance levels. The explainer also discusses an alternative scoring approach. This reference does not calculate a tier, score or compliance grade.
It does not describe your product’s accessibility. W3C uses Placeholder, Exploratory, Developing, Refining and Mature to communicate how settled a section is and the feedback it needs.
The published snapshot lists Developing provisions. The changing Editor’s Draft contains additional exploratory material. Contrast methods and some values remain unsettled; do not treat a particular contrast algorithm as an adopted WCAG 3 rule.
This independent learning application presents the W3C Accessibility Guidelines (WCAG) 3.0, Working Draft 10 September 2026. Source checked 15 September 2026. Counts describe this snapshot, not the final standard. For subsequent changes, consult the latest published draft and WAI introduction.
Copyright © 2021–2026 World Wide Web Consortium. Copyright © 2023 W3C®. This software includes material copied from or derived from W3C Accessibility Guidelines (WCAG) 3.0, Working Draft 10 September 2026, https://www.w3.org/TR/2026/WD-wcag-3.0-20260910/, and its Explainer, Group Note Draft 10 September 2026, https://www.w3.org/TR/2026/DNOTE-wcag-3.0-explainer-20260910/. W3C Document License. W3C excerpts are identified separately from original A11y.mv explanations and activities. This application is not an official W3C technical specification.