How to Make a PDF Accessible

A PDF describes where marks go on a page. A screen reader needs to know what those marks mean — this is a heading, this is a table cell, this image shows a chart. Bridging that gap is what accessibility work on a PDF consists of.

What an accessible PDF has

• Real text rather than images of text.
• Tags marking headings, paragraphs, lists and tables as what they are.
• A defined reading order, so content is announced in the right sequence.
• Alternative text on every meaningful image.
• Table headers identified so cells can be related to them.
• A document language set, so the correct pronunciation is used.
• A document title, which is what a screen reader announces first.
• Sufficient colour contrast, and meaning not carried by colour alone.

The single biggest problem

Scanned documents. A screen reader encountering a scan finds no text at all and announces nothing. The user has an empty document. Running OCR is the single highest-impact accessibility improvement available for most archives, and it takes one pass.

Fix it at the source

Tagging is far easier to get right in the authoring tool than to add afterwards. In Word, use the built-in heading styles rather than manually formatting large bold text, use real lists and real tables, add alt text to images, and export with "document structure tags for accessibility" enabled. The tags carry through into the PDF automatically.

Retro-fitting tags to an untagged PDF is genuinely laborious work, which is why doing it in the source is worth insisting on.

A practical order of work

1. 1OCR anything scanned.
2. 2Set the document title and language in the properties.
3. 3Re-export from the source with tagging enabled, where you have the source.
4. 4Add alternative text to images.
5. 5Check the reading order, particularly on multi-column pages.
6. 6Test with an actual screen reader rather than only a checker.

Consider whether it should be a PDF

For content that will be read on screen, an HTML page is more accessible by default: it reflows, it carries semantic structure natively, and assistive technology handles it far better. Publish the page and offer the PDF for people who want to print or file it. That single decision solves more accessibility problems than any amount of remediation.