How to tag a PDF for accessibility

Tags are the single most important thing about an accessible PDF, and the one almost every government document is missing. Here is what they are and how to get them.

What a tag actually is

Open a PDF and you see a page: a big bold line at the top, some paragraphs, a table. You know the bold line is a heading because it looks like one.

A screen reader can't see. It reads an invisible structure underneath the page that says "this is a heading, this is a paragraph, this is a table with these column headers." That structure is the tags.

Without tags, a screen reader gets one undifferentiated stream of words. No headings to jump between, no reliable reading order, no way to tell where a section begins. On a one-page notice that is annoying. On a 90-page budget it is unusable.

The way that works: start in Word

1. Use real headings, not big bold text

This is the step that matters most, and it happens before you export anything. In the Home ribbon there is a Styles gallery: Heading 1, Heading 2, Heading 3. Use them for your headings.

Selecting text and making it 16pt bold produces something that looks like a heading and carries no structure at all. Applying Heading 1 produces something that looks the same and becomes a real heading in the exported PDF. Same appearance, completely different document.

2. Add alternative text to every image

Right-click an image → View Alt Text → describe what it conveys. If the image is purely decorative, mark it as decorative instead. See our separate guide for what makes a good description.

3. Set the document title

File → Info → in the right panel, fill in the Title field. This is what a screen reader announces instead of the file name, and it is a requirement, not a nicety.

4. Run Word's own checker

Review → Check Accessibility. It catches missing alt text, blank table headers and low contrast before you export, when they are still easy to fix.

5. Export the right way — this is where most PDFs are ruined

If you only have the PDF

Sometimes the original is gone, or came from a consultant. Then you are working in Acrobat Pro:

1. Check what you have. All tools → Prepare for accessibility → Check for accessibility. The report tells you whether the document is tagged at all.
2. Auto-tag it. Prepare for accessibility → Autotag document. Acrobat makes a reasonable first pass at headings and paragraphs.
3. Fix what it got wrong. Auto-tagging is a starting point, not an ending one. It reliably struggles with complex tables, multi-column layouts, and reading order on anything visually elaborate. It also does not write alt text — you have to.
4. Set the title and language. File → Properties → Description for the title; Advanced → Reading Options for the language. Then in Initial View, set "Show: Document Title."

If the pages are scanned images

If someone scanned a paper document, there is no text in the file at all — a screen reader reads nothing. Tagging cannot help until the text exists. Run text recognition first (Acrobat: Scan & OCR → Recognize Text), verify the result is accurate, and only then tag it. Recognition on an old faxed document is often poor enough that retyping is faster.

Two things that trip people up after tagging

Tagged is not the same as correctly ordered

A document can be fully tagged and still read as nonsense. Tags say what each piece is; a separate setting controls the order a screen reader encounters them in. They are not automatically the same.

This bites hardest on anything with a visually interesting layout. A newsletter with two columns and a sidebar may be tagged perfectly and still be read left-to-right across both columns, producing half a sentence from column one followed by half a sentence from column two. Auto-tagging gets this wrong routinely.

Headings are a hierarchy, not a font size

Screen reader users navigate by pulling up a list of headings, the way you would skim a table of contents. That only works if the levels nest properly: Heading 1 for the document title, Heading 2 for major sections, Heading 3 for subsections inside them.

Do not skip levels. Going from a Heading 2 straight to a Heading 4 because Heading 4 happened to be the size you wanted leaves a gap that reads as a missing section. If you dislike how a level looks, change its formatting in the style — don't reach for a different level.

One Heading 1 per document is the norm: the title. Everything else nests underneath.

How to know it worked

Run Acrobat's accessibility check again, or drop the file into our free checker. What you want to see: a tag structure exists, headings are present, images have descriptions, and the document declares a title and a language.

A worthwhile habit: fix one document completely, then look at what changed. Once you have seen a tagged file next to an untagged one, the rest of the backlog stops being mysterious.