Framework for Accessible Specification of Technologies (FAST) advises creators of technical specifications how to ensure their technology meets the needs of user with disabilities. It addresses primarily web content technologies but also relates to any technology that affects web content sent to users, including client-side APIs, transmission protocols, and interchange formats. Specifications that implement these guidelines make it possible for content authors and user agents to render the content in an accessible manner to people with a wide range of abilities.

This document is accompanied by a checklist which is more complete and usable for groups evaluating technologies at this time.



This document is accompanied by a checklist which is more complete and usable for groups evaluating technologies at this time.


Numerous guidelines exist for creating and supporting content that is accessible to people with disabilities, on and off the Web. When these guidelines are supported in the entire web ecosystem, content creators can author accessible content, and expect the accessibility features to be made available by user agents, including assistive technologies when needed. Authoring tools support creation of accessible content, and accessibility features survive transmission to different systems or conversion of content to different formats.

Nearly all of these accessibility features depend on support in some form from the technology in which content is encoded, transmitted, and sometimes transformed. But there is not yet a set of well-documented guidance for such technologies. Instead, requirements are inferred from authoring and user agent guidelines. This makes it complicated for technology creators to ensure they have met the full set of needs. Review from accessibility specialists is limited by bandwidth and expertise, so does not fully address that problem. As a result, varying technologies provide various levels of support with varying levels of compatibility with other technologies. These issues at the core layers of Web technology impact the progress that can be made from support of higher-level guidelines.

Framework for Accessible Specification of Technologies aims to fill this gap. It is intended to be a single, well-considered set of guidelines addressing specifically the features technologies need to provide to support accessible. These guidelines relate to the requirements of other guidelines but should not be confused with them. The goal of FAST is to provide a single source of guidelines for Web technology accessibility. They relate to other guidelines and documentation to provide additional information and rationale for the requirement, but are intended to be a self-sufficient set of guidelines that technology creators can follow.


The primary audience of FAST is creators of Web technologies. Most of the guidelines relate to content and presentation technologies like HTML, CSS, SVG, PDF, audio/video formats, etc. Some guidelines also address data formats, interchange formats, transmission protocols, etc., usually aimed at ensuring these technologies preserve the accessibility features of content impacted by these technologies. Because of this broad set of relevant technologies, all Web technology creators are considered part of the audience for FAST.

Secondary audience include creators of higher level accessibility guidelines and other advocates for web accessibility features. Because FAST has a strong grounding in user needs, researchers and advocates who identify the accessibility requirements of web users with disabilities are also an important audience.


Framework for Accessible Specification of Technologies is a product of the World Wide Web Consortium and as such targets only the accessibility requirements of web technologies. Many of the user needs are the same for web use and non-web use, so FAST will necessarily overlap and hopefully be compatible with similar guidelines addressing non-web space. Nonetheless, FAST is not designed to be used for non-web technologies and there could be key differences. Furthermore, there are user needs that exist outside the web that do not impact the web, and those needs are completely unaddressed by FAST.

In spite of these caveats about the scope of FAST, this scope will evolve as the Web does. More and more technologies are becoming part of the Web, and bringing user needs to the Web along with them. For instance, strictly hardware accessibility issues may be non-web requirements, but the Web of Things brings many of these issues closer to the Web than in the past. FAST will need to reflect this evolution, and future versions may be required to address user needs that are new to the Web.


The goal of FAST is to help ensure that web technologies meet the needs of users with disabilities. To do this, the work involves three stages:

Inventory user needs

The first step in the development of these guidelines is to inventory known user needs. Many user needs affecting web content accessibility are well known and documented in multiple places. These needs are collected and related to each other in order to arrive at a single set of known needs. Sources examined in the development of these guidelines include:

Note the goal of this exercise is not to supplant other good work in this field. The aim is to assemble disparate sources if knowledge about user needs in one place, to facilitate analysis. This work is likely to spin off from the core work of developing FAST. If another organization creates a sufficiently rich collection of documented user needs it will be possible to use that resource rather than reinvent the work in W3C.

Identify ways to meet needs

The second stage in development of the guidelines is to identify ways these needs can be met. There are three high-level ways user needs can be met:

  • technology features;
  • author implementation;
  • user agent support.

These are not mutually exclusive categories. A given user need could be met by more than one of these categories, but the ability of a given category to meet a user need implies the need for guidelines targeting that category. In policy setting and evaluation there may be a preference hierarchy for how best to meet needs, e.g., user agent support of standard features is preferred, but author technical override is needed if user agent support is lacking.

Some needs can be met with present technology only via one of these routes. Other needs can be met by more than one route, and for content to be accessible it is only required that one of the available routes be implemented. Many needs, however, require more than one route to be implemented together for the need to be met. The most common example is that a technology provides a feature, the author uses that feature in the content, and the user agent makes the result available to the user.

All of these ways of meeting user needs are identfied, along with their relationships to each other. Once these approaches are identified, the result is separate lists of requirements for content technologies, authors, and user agents. The relationship among the routes may play a role in prioritization of guidelines, since needs that can only be met by one route may be more important to meet by that route, than needs that could be met by other means as well.

Develop technology guidelines

From the above analyses, it should also be easy to see where content technology features are required to make it possible to meet user needs. For example:

  • If the author must implement something, the technology must provide a feature for the author to implement.
  • If the user need is met by design, the technology must provide suitably rich design capabilities.
  • If the user need is met by user agents, the technology must provide a sufficiently rich definition of the object for user agents to implement.

Not all technologies will address all ways of meeting user needs. For instance, CSS is primarily design-oriented, and HTML is somewhat semantics-oriented. The technology requirements may need conformance profiles or some other way of guiding technology developers seeking to follow them. It may not be easy to state in a general prescriptive way whether a given technology should, for instance, provide a richer design capability to meet a user need or should instead rely on better semantics for assistive technology-oriented content alternatives. A good structure of the technology requirements should help make it clear that some method of meeting a given user need is important. Horizontal review may continue to be important in guiding technology developers through the possibilities.

The set of approaches to meeting user needs that affects technology features becomes the base information for the Framework for Accessible Specification of Technologies. (The other two routes, while important to the analysis, are not directly relevant to FAST but may inform other work.) These approaches are prioritized, organized, and translated into guidelines-type language to become the Framework for Accessible Specification of Technologies.

With the above analyses done, it should be easy to see how current guidelines address which user needs. In turn it should be easy to see where current guidelines to not meet user needs, that in theory should be able to be met by activities within the remit of that set of guidelines. This should be important input into WAI 3.0 / WAI 2020 planning.

Explanation of User Needs

Identify user needs that we plan to provide guidance on meeting. These should describe the needs of humans as they currently exist (i.e., without significant evolution or cyborgization from early 21st century norms), and therefore is as era-independent as possible with current knowledge. The focus is on the needs of people with disabilities, but because that is sometimes a relative / contextual condition, a significant proportion of mainstream needs will also be identified.

At least two levels of needs may be identified. The first is truly generic needs, requirements users have to access and use content such as perceive and understand it, and should be stable over time. The second layer is needs specific to technologies of the day, such as ability to understand and operate controls. This layer may be understood as an implementation of the generic needs, so may not be classed as user needs in the end. Regardless of its classification, it will be an important component of understanding the space. This level of needs evolves as technology and design patterns do, so needs to be maintainable separately from the generic needs.

Evolution of user needs

Where user needs are suspected but known, related work may expand the inventory through research when feasible. Therefore the set of documented user needs will evolve over time. A given set of guidelines including FAST, however, can only address needs that were known at the time of development of the guidelines.


  • User is accessing Web content on hardware / OS / AT combination that supports their needs. This may not always be true for shared device / public kiosk situations, but that issue is out of scope.

Applicability of Needs

  • Content
  • Controls (buttons, fields, etc.)
  • Input indicators (mouse, keyboard cursors)
  • Signals (non-actionable state indicators)
  • Alerts (dialogs, alarms, etc.)

Collected User Needs

This is a draft user needs collection. It may or may not be prove desirable to prioritize needs in order to yield manageable sets. These needs will be cross-referenced to authoritative sources that also express the same requirement.

In the list below, there may be overlap between user needs, and known ways to meet user needs. These will need to be teased out over the course of the project.

Structure: Top level is user need, 2nd level refines to user groups, 3rd level is ways to meet the need, below that is details.

Meeting User Needs

This is a preliminary draft to document how user needs are met in various ways.

For each user need, ways to meet it are proposed for:

Other categories may be included later. Many user needs can be met in more than one way. The mechanism to meet user needs in one of the above areas may require support from one or more of the other areas.

Ways to Meet User Needs

User needs need to be analyzed for how they can be met. The following ways of meeting needs are currently understood:

Author Implementation

User Agent Features

Accessibility Support in Mainstream User Agents
Assistive Technology
Accessibility APIs

Meeting User Needs Table

This version of the resource is primarily to show the structure, not yet a comprehensive documentation of how user needs can be met.

The table below shows how two of the user needs identified above might be met by technology features, author implementation, and user agent support. Each row of the table shows a related set of approaches, in which the approach in column depends on successful implementation of the approaches in the other columns for that row. For instance, many author features depend on support from the technology as well as exposure from the user agents. Some approaches to meeting user needs do not require support from others, which is reflected by rows with blank columns. For instance, it is possible for a user agent to meet certain needs with no particular support provided by the technology or author. This layout is preliminary and a more expressive layout is sought.

User Need Technology Content Author User Agent
Text Alternatives Provide a mechanism for author to create text alternatives and associate with content Create text alternative content and associate with primary content using features of the content technology Expose text alternatives provided by the author
Define parseable and semantically rich content encoding that supports automated creation of text alternatives Encode content using a content technology that is sufficiently rich that machines can create useful automated text alternatives Create automated text alternative content based on the semantics of the primary content
Color Contrast Provide color definition features that allow authors to set colors to meet requirements Use only colors that meet luminosity contrast guidelines
Provide color definition features that allow users to override author-set colors Provide a feature for users to override author colors
Provide color definition semantics that allow colors of common object types to be globally remapped easily Use semantically defined color mappings to allow user global preferences to be easily applied Support semantically defined color mappings to allow users to define global preferences that are easily applied across a range of content
Provide a feature to allow users to define their own color preferences
Provide a feature to allow users to request "high contrast" mode
Provide a "high contrast" mode that overrides author colors

Framework for Accessible Specification of Technologies

The content below is merely initial draft content intended to show how guidelines aimed at web technology developers might look. It has not yet been related to the user needs and ways of meeting them outlined above. It serves as initial brainstorming to help demonstrate viability of this set of guidelines.

Alternative Content

Text Alternatives

  • Provide a way to define short text alternatives / labels for non-text content.
  • Provide a way to define long text alternatives for non-text content.
  • Allow text alternatives to be semantically "rich" e.g., with page structure, text style, hyperlinks, etc.

Rich Alternatives

  • Provide a way to define non-text alternatives for text content.
  • Provide a way to define non-text alternatives for non-text content .


Device Independence

Universal meaning support

  • Provide declarative mechanisms (that have accessibility semantics pre-defined in the spec) to implement technology features whenever possible.
  • Define unambiguous ways to express relationships between units of content, such as object nesting, ID referencing, etc.
  • Prefer structural semantics to presentational semantics.
  • When providing presentational semantics, define ways they can be easily mapped to structural semantics, e.g., to support restyling or meaningful exposure to AAPIs.
  • Minimize the need for alternative content by supporting a comprehensive set of authoring use cases. (e.g., don't make authors resort to text in images to get the style they want)

Compatibility with AAPIs

  • For every user interface object type, define the "type" of object as a role to AAPIs.
  • For every user interface object type, define how authors provide or user agent determines the "accessible name" for AAPIs.
  • For user interface objects that can have states, properties, or values, define how authors can set these and how these are exposed to AAPIs.
  • When providing imperative mechanisms to implement technology features (e.g., scripts), provide a way for authors to expose accessibility information to AAPIs.
  • Provide a way to title Web pages and sections of content.
  • Provide a way to clearly indicate the target of a hyperlink and function of a control.
  • Provide a way to indicate content language, for the page as a whole and for blocks of content.
  • Provide a way for authors to support understanding of abbreviations / acronyms / initialisms, idioms, jargon, etc.
  • Provide a mechanism to support correct machine pronunciation of ambiguously spelled terms (e.g., in the phrase "I am content with this content" there are different correct pronunciations of the lexeme "content").

Hardware Interfaces

  • Abstract hardware interfaces so various device types can simulate each others' functions.
  • Always provide a keyboard interface to interact with content.

User Customization

User Control

Blinking / Flashing

  • Define a mechanism to warn users of flashing content.
  • Define a mechanism to identify potentially flashing content to user agents so they can suppress it on user preference.
  • Define declarative mechanisms for features that could cause blinking or flashing so their parameters can be more easily controlled by user agents than imperative mechanisms.

Automatic actions

  • Provide a way for users to prevent time-based content from playing automatically, e.g., as a user agent preference that implementations must provide.
  • Provide a way for users to request no interruptions. Note, bona fide emergency interruptions should still be allowed.
  • Define that there will be not automatic action leading to a change of context when focus lands on a given element.

Time-based content

  • Provide a way for users to pause and stop time-based content.
  • Provide a way for users to start or restart time-based content.


Audiovisual Content


Alternate view modalities


User input


The following people contributed to the development of this document.

Participants in the PFWG at the time of publication

  1. David Bolter (Mozilla)
  2. Sally Cain (Royal National Institute of Blind People)
  3. Michael Cooper (W3C/MIT)
  4. James Craig (Apple Inc.)
  5. Steve Faulkner (Invited Expert, The Paciello Group)
  6. Geoff Freed (Invited Expert, NCAM)
  7. Jon Gunderson (Invited Expert, UIUC)
  8. Markus Gylling (DAISY Consortium)
  9. Sean Hayes (Microsoft Corporation)
  10. Kenny Johar (Vision Australia)
  11. Matthew King (IBM Corporation)
  12. Gez Lemon (International Webmasters Association / HTML Writers Guild (IWA-HWG))
  13. Thomas Logan (HiSoftware Inc.)
  14. William Loughborough (Invited Expert)
  15. Shane McCarron (Invited Expert, Aptest)
  16. Charles McCathieNevile (Opera Software)
  17. Mary Jo Mueller (IBM Corporation)
  18. James Nurthen (Oracle Corporation)
  19. Joshue O'Connor (Invited Expert)
  20. Artur Ortega (Yahoo!, Inc.)
  21. Sarah Pulis (Media Access Australia)
  22. Gregory Rosmaita (Invited Expert)
  23. Janina Sajka (Invited Expert, The Linux Foundation)
  24. Joseph Scheuhammer (Invited Expert, Inclusive Design Research Centre, OCAD University)
  25. Stefan Schnabel (SAP AG)
  26. Richard Schwerdtfeger (IBM Corporation)
  27. Lisa Seeman (Invited Expert, Aqueous)
  28. Cynthia Shelly (Microsoft Corporation)
  29. Andi Snow-Weaver (IBM Corporation)
  30. Gregg Vanderheiden (Invited Expert, Trace)
  31. Léonie Watson (Invited Expert, Nomensa)
  32. Gottfried Zimmermann (Invited Expert, Access Technologies Group)

Enabling funders

This publication has been funded in part with Federal funds from the U.S. Department of Education, National Institute on Disability and Rehabilitation Research (NIDRR) under contract number ED-OSE-10-C-0067. The content of this publication does not necessarily reflect the views or policies of the U.S. Department of Education, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.