Accessible PDFs: What a Tagged PDF Actually Has to Contain
A PDF that looks right can still be unreadable to a screen reader. Tags, reading order, and alternative text are what make the difference, and most store PDFs have none of them.
A store posts its returns policy as a PDF. It looks fine, it prints fine, and a screen reader user opens it to find a single block of text with no headings, no lists, and no order, because the PDF was exported from a design file with no tags at all. The document exists, the information is in it, and the customer who needs it cannot use it.
Tagged PDFs are the difference. A tag is a structural label, a heading, a paragraph, a list item, that tells assistive technology what each piece of text is. Without tags, a PDF is a picture of a document. With them, it is a document.
What tags actually do
Tags give a PDF a reading order and a structure. A screen reader follows the tag tree, not the visual layout, so a two-column document reads down the first column and then the second, instead of jumping across the page line by line. Headings in the tag tree let a user navigate by section, the way they would on a web page. Lists, tables, and figures all need tags too, or they collapse into unstructured text.
The standard behind it is PDF/UA, the ISO standard for universally accessible PDF, and WCAG 2.2 applies the same principles to PDF content as to web pages. The DOJ Title II rule, which adopts WCAG 2.1 Level AA for state and local government web content, covers documents posted on government websites, and the same expectation is spreading to private stores through demand letters and procurement contracts.
The failures that show up most
The most common failure is the untagged export. A designer builds a beautiful returns policy in a design tool, exports to PDF, and nobody turns tagging on. The file opens, the text is selectable, and the tag tree is empty. The second failure is the scanned image: a paper form photographed and saved as a PDF, with no text layer at all, so a screen reader finds nothing.
The third is a tag tree that exists but is wrong. Headings tagged as paragraphs, a table tagged as a list of unrelated lines, a figure with no alternative text, or a reading order that runs across columns. A wrong tag tree is harder to catch than a missing one, because the document looks tagged to anyone who does not check it with a screen reader.
Making a PDF accessible without a rebuild
Start in the source document, not in the PDF. Use real heading styles in the word processor, real lists, and real table headers, then export with tagging enabled. That single step fixes most of the structure. In Acrobat, the accessibility checker will list the remaining problems: missing document title, missing language, untagged content, figures without alternative text.
Set the document title and language in the file properties, because a screen reader announces the title when the document opens and uses the language to pronounce the text. Add alternative text to figures that carry meaning, and mark decorative images as artefacts so they are skipped. For scanned documents, run optical character recognition first, then tag, because a tag tree on an image with no text layer is decoration.
Related reading: Guarding Against the Instant Compliance Pitch, Captions on Product Videos: The Barrier Hiding in the Demo Reel, Accessible Store Locators and Pickup Selection: Maps Are Not the Only Way In.
Frequently asked questions
What is a tagged PDF?
A PDF with a structural tag tree that labels headings, paragraphs, lists, tables, and figures. A screen reader follows the tag tree to read the document in order. Without tags, a PDF is a picture of a document that assistive technology cannot navigate.
Is a scanned PDF accessible if the text looks clear?
No. A scanned page is an image with no text layer, so a screen reader finds nothing to read. Run optical character recognition to create the text layer, then add tags, before posting it.
What is PDF/UA?
PDF/UA is the ISO standard for universally accessible PDF. It sets the requirements for tags, reading order, alternative text, and document structure. WCAG 2.2 applies the same principles to PDF content as to web pages.
Do I need to remake old PDFs from scratch?
Usually not. Fix the source document with real heading styles and lists, re-export with tagging enabled, then run the accessibility checker and fix what it flags: document title, language, untagged content, and figures without alternative text.