WCAG and the four POUR principles

Overview

What WCAG is

WCAG stands for the Web Content Accessibility Guidelines. It is published by the W3C, the World Wide Web Consortium, the group that sets standards for the web. WCAG describes what digital content must achieve so that people with disabilities can use it. It started with web pages, but it now applies to almost any digital content, including your PDF.

WCAG does not tell you which buttons to press in a particular program. It describes the result a reader must be able to get. For example, it says a reader must be able to tell what an image shows, not that you must use a specific menu in a specific tool. This keeps the standard useful no matter what software made the file.

The four principles, in plain words

WCAG is built on four ideas. The first letters spell POUR: Perceivable, Operable, Understandable, and Robust.

Perceivable means a person can take the content in. If they cannot see the screen, the content must reach them another way, through a screen reader or braille display. An image that has no text description is not perceivable to a reader who cannot see it.

Operable means a person can move through the content and use it. They must be able to reach every part, whether they use a mouse, a keyboard only, or voice control. A link they cannot reach with the keyboard is not operable.

Understandable means the content reads and behaves in a way the reader can follow. The words make sense, the language is declared so a screen reader pronounces it correctly, and parts of the document appear in a sensible order.

Robust means the content works with assistive technology, the software and devices people with disabilities rely on, such as screen readers. A file that is built correctly will still work as those tools change and improve over time.

These four principles cover the whole experience, not just one part. A document can be easy to see and still be impossible to navigate. All four have to hold for the document to work for everyone.

In depth

Each principle, and what breaks it

Perceivable is about getting the content to the reader's senses. Most people read a document with their eyes. A reader with a print disability may use a screen reader that speaks the text aloud, or a braille display that raises the text under their fingers. For that to work, every piece of meaning has to exist as something other than a picture on a screen. A photograph needs a text description. A scanned page needs real text behind the image, not just a flat picture of text. Colour contrast also lives here: if your text is light grey on a white background, a reader with low vision cannot make it out, even though the words are technically present.

Operable is about reaching and using the content. Many people never touch a mouse. Someone with a motor disability may use only the keyboard, or a switch, or their voice. A screen reader user moves through a document by jumping from heading to heading, or from link to link. If your document has no headings, that reader has no way to skim and must listen to every word from the top. The headings you can see on the page are not enough. They have to be marked as headings in the file so the assistive technology can find them.

Understandable is about meaning and predictability. The reader should be able to follow the content and not be surprised by it. Part of this is plain writing. Part of it is technical. Your document has to declare its language, for example English or Hindi, so a screen reader uses the right pronunciation. A French phrase inside an English document should be marked as French, or the screen reader will read it with an English accent that may be impossible to understand. Reading order also lives here: the order in which content is read aloud must match the order a reader expects, not jump around the page.

Robust is about working with the tools people actually use. A document is robust when it is built to the rules of the format so that any conforming assistive technology can interpret it. The structure that a screen reader follows, the tags, has to be valid and complete. If the tags are broken or missing, the screen reader cannot present the content correctly, no matter how good the writing is. Building to the standard also means the file keeps working as screen readers and other tools are updated.

How the common PDF problems map to the principles

The accessibility problems found in PDFs are not random. Each one sits under one of the four principles, and seeing that mapping helps you understand why a problem matters.

Missing alternative text is a Perceivable problem. A reader who cannot see the image gets nothing. If a chart in your report shows that enrolment doubled, and the chart has no text description, a screen reader user simply does not learn that fact. Low colour contrast is also a Perceivable problem, for a reader with low vision who can see the page but cannot separate faint text from the background.

A missing or wrong document language is an Understandable problem. If your PDF does not declare that it is written in English, the screen reader may guess wrong and read the whole thing in the wrong language's pronunciation rules. Reading order is also Understandable. If a floating text box was added last, a screen reader may read it last, so the reader hears a caption before the paragraph it belongs to, and the document stops making sense.

Broken or missing tags are a Robust problem. The tags are the structure the screen reader follows. If a heading is not tagged as a heading, the reader cannot jump to it. If a table is not tagged as a table, the reader cannot tell which cell belongs to which column. The content may look perfect on screen and still be unusable, because the layer the assistive technology reads is broken.

Keyboard traps and unreachable links are Operable problems, though these appear more often in interactive forms than in plain documents. If a reader can tab into a form field but cannot tab out of it, they are stuck.

Success criteria and the three levels

Each principle is broken down into specific, testable requirements. WCAG calls each one a success criterion. A success criterion is a single rule that your content either meets or does not meet. "Images that carry meaning have a text alternative" is the kind of thing a success criterion states. The principles tell you the goal; the success criteria tell you exactly what to check.

Every success criterion sits at one of three levels: A, AA, or AAA. The levels describe how far you have gone, not how hard each item is.

Level A is the minimum. These are the most basic requirements, the ones that, if missed, leave some people completely shut out.

Level AA is the level most laws and policies require. When an organisation says a document must "meet WCAG," they almost always mean Level AA. Meeting AA includes meeting everything in A as well. AA is the practical target for your document, and it is the level our own checks are built around.

Level AAA is the highest. It adds the strictest requirements. The W3C itself does not expect every item of content to meet every AAA criterion, because some of them cannot be satisfied for all types of content. AAA is something to reach for in specific places, not a blanket requirement.

Where WCAG and PDF/UA meet, and where machine checking stops

WCAG is not the only standard behind an accessible PDF. PDF/UA, the accessibility profile for PDF, says how to build the file using the PDF format correctly. The two overlap. PDF/UA conformance covers WCAG for the PDF's own page content. But WCAG adds a few things PDF/UA does not test, and colour contrast is the main one. A PDF can follow every PDF/UA rule and still have text too faint to read, because the contrast check is a WCAG matter, not part of the PDF/UA test set. For the relationship between these standards, see the topic on PDF/UA and the Matterhorn Protocol.

Automated tools can check some success criteria and not others. A tool can confirm that an image has alternative text attached. It cannot judge whether that text actually describes the image. A chart whose alternative text reads "image" passes the machine check and fails the reader. A tool can confirm that the document declares a language. It cannot confirm that the language is correct, or that a quoted French sentence inside it is marked as French. A tool can confirm that headings exist in the structure. It cannot confirm that they are on real heading text rather than on a sentence that merely happens to be bold.

This is why a checker's green light covers only part of the picture. The machine handles the parts that can be measured. A person still has to judge meaning. For more on this split, see the topic on what automated checking can and cannot find.

Reference detail

WCAG identifiers and versions

WCAG 2.0 was adopted as an international standard, ISO/IEC 40500:2012. The guidelines have been updated since then. The current version is WCAG 2.2, published in 2023. The ISO mapping covers version 2.0, not the later updates.

The four principles

The three conformance levels

Common PDF problems mapped to the principles

Authoritative sources