Skip to content
SilktideHelp

PDF tagging

Silktide checks whether each PDF document on your website carries accessibility tags. Tags are a hidden layer describing the document's structure - which text is a heading, which is a paragraph, where a list starts and stops - and software that reads a document aloud depends on them entirely.

A stores that structure in a , much as a web page stores it in . An untagged PDF has none: it is ink positioned on a page, and every question a needs to answer has to be guessed from where things happen to sit.

Why this matters

This is the single largest accessibility barrier a document can have, and it is the one everything else rests on. Without tags there is no reliable , so a two-column newsletter is read straight across, mixing two unrelated sentences together. Tables collapse into a stream of numbers with nothing to say which column they came from. Headings are indistinguishable from bold text, so there is no way to skim, and a reader must listen to a fifty-page report from the top to find one section. Images have no descriptions at all.

The practical result is that many untagged PDFs are not merely awkward without sight, they are unusable. Because needs tags to do anything, no other PDF fix helps until this one is in place: alternative text, headings, and table headers are all stored in the tags.

Untagged documents are also the most common reason an organisation fails a document accessibility audit, and PDFs published by public sector bodies are explicitly in scope of most accessibility regulations.

How to fix it

  1. Fix it at the source wherever you still have it. Re-exporting from the original file is faster and produces better structure than repairing the PDF afterwards.
    • In Microsoft Word, apply real heading and list styles rather than manually enlarging text, then export to PDF with "Document structure tags for accessibility" enabled.
    • In Adobe InDesign, set the article order and paragraph style export tags, then export as a tagged PDF.
    • In LibreOffice, export with "Universal Accessibility (PDF/UA)" selected.
  2. Or let Silktide remediate it. Silktide can rebuild a full tag structure for a document without changing how it looks. See How PDF remediation works.
  3. Or add tags by hand in a PDF editor such as Adobe Acrobat Pro. Acrobat can auto-tag a document in one step, but auto-tagging is a starting point rather than a fix: expect to correct the reading order, heading levels, and table headers yourself afterwards.
  4. If the document is a scan, recognise the text first. A scanned page is a photograph, and there is nothing for tags to describe until has produced real text. See Scanned documents.

If a document is only published as a PDF out of habit, the best fix is often not to fix the PDF. A web page is easier to make accessible, easier to keep current, and easier to read on a phone.

How Silktide tests this

  1. Find the PDF documents linked from your website, and analyse each unique file once however many pages link to it.
  2. Read the document to see whether it declares a structure tree, and whether it is flagged as tagged.
  3. Report the document when either is absent.
  4. Report each document once, no matter how many pages it contains.

This check tests only whether tags exist, not whether they are correct. A tagged document can still misrepresent its own content, which is what the other document checks and AI review examine.

Troubleshooting

The document looks fine when I open it

Tags are invisible. They change nothing about how a document appears on screen or in print, so a document that looks perfect can be completely untagged. What tags change is what software can tell about the content.

I can select and copy the text, so isn't it tagged?

No. Selectable text and tags are separate things. Text is extractable whenever the words are stored as characters rather than as a picture, and that happens with or without tags. Tags add the structure around the text: what is a heading, what order to read in, what the images mean.

The document is not listed

Silktide reports the PDFs it can reach from your pages. A document nobody links to will not appear, and neither will one that could not be opened for analysis - for example a password-protected file. See Password-protected and restricted documents.

I republished the document but the check still reports it

Results refresh when the document is next tested. Note also that replacing a file at the same address is not always enough on its own: check that the page really links to the new file.

Learn more

Last updated

Was this page helpful?