PDF accessibility check
Select a PDF and your browser opens it locally to inspect what a screen reader depends on: the tag tree, the text layer, the document properties. Each gap comes with the place to fix it.
Your browser opens the file on your own device. It is not uploaded and we never see it, so private documents are safe to check.
A PDF is a drawing with an optional outline
A web page starts from structure. The HTML says “this is a heading, this is a list” and the browser decides how to draw it. A PDF works the other way around. At its core it’s a set of drawing instructions: put this glyph at these coordinates, paint this image there. A printer or a pair of eyes needs nothing more. But nothing in there says which glyphs form a paragraph, which column comes first or what a picture shows.
For that, the format has a second, optional layer called the structure tree, or simply tags. It looks a lot like HTML: H1, P, L for lists, Table, Figure with alternative text. Each tag points to the drawn content it describes. Screen readers, the reflow mode of phone viewers and text-to-speech tools follow the tags. When there aren’t any, they fall back on drawing order, and in a two-column brochure that often means jumping between columns, captions and page footers.
What is inspected in your file
| Check | Problem reported when |
|---|---|
| Protection | Encryption forbids both copying and text extraction for accessibility |
| Tags | No structure tree, a tree with fewer than three elements, or tag names outside the standard set with no role mapping |
| Real text | Pages hold under three characters: image-only pages point to a scan, pages made of many vector paths point to text converted to outlines |
| Text encoding | More than 5% of extracted characters are unmapped symbols, the sign of fonts without a Unicode table |
| Title | Empty, or looking like a file name (“Report_v3.docx”), or present while the viewer is not told to display it |
| Language | No document language declared |
| Figures | Tagged figures lack alternative text, or their text is a file name or a word like “image” |
| Headings and tables | No heading tags, no H1, skipped levels; tables without TH header cells; clickable links absent from the tags |
| Forms | Fields without a tooltip, which is the name a screen reader speaks; XFA forms |
| Fonts | Fonts referenced but not embedded |
Text and form fields are read on the first 500 pages, the structure tree on the first 300, images and fonts on the first 60.
Which findings to treat first
A red finding stops a reader completely. A scan has no text to speak, an untagged file has no order, a locked file may refuse to hand its text to assistive software, and a missing language makes the speech engine pronounce every word with the wrong rules. Fix those before anything else.
Amber findings make the reading worse without blocking it. With no title, the window and the screen reader announce a file name. Headings without an H1, or with gaps, make navigation by heading unreliable. A table without header cells is read as a stream of numbers.
If the summary names the program the file came from, start there. Repairing the source document and exporting again is nearly always faster than repairing the PDF.
Fix it at the source
- Microsoft Word, PowerPoint
- Use the built-in heading styles and real lists. Add alt text from the picture’s context menu. Set the title under File › Info. Run Review › Check Accessibility. Then export with File › Save As › PDF › Options and keep “Document structure tags for accessibility” ticked. Never use Print › Save as PDF, which throws the tags away.
- Google Docs
- Apply heading styles and alt text, then File › Download › PDF document. Check the result here, because title and language are often still missing and need a PDF editor.
- LibreOffice Writer
- File › Export as PDF, then tick “Universal Accessibility (PDF/UA)”, which also turns on tagging. Set the title under File › Properties › Description.
- Adobe InDesign
- Map paragraph styles to export tags, set alt text in Object Export Options, order the content in the Articles panel, and export with “Create Tagged PDF” and Display Title set to Document Title.
- Canva
- Add alt text to images and download as PDF Standard without flattening. For anything longer than a page or two, expect to finish the structure in a PDF editor.
- No source file, or a scan
- In Adobe Acrobat Pro: Scan & OCR › Recognize Text for scans, then Accessibility › Automatically tag PDF, then correct the tags and reading order by hand. Title, display setting and language are under File › Properties.
What stays a human job
The check confirms that the pieces are there. It can’t tell you whether they’re any good. Alt text that says “chart” counts as present. Tags in the wrong order are still tags, and a heading tag on a decorative line is still a heading. Color contrast inside the pages isn’t measured, and neither is the logical tab order of form fields.
Passing here isn’t a certificate of PDF/UA or WCAG conformance. For documents with legal weight, run the full checker in Acrobat Pro or the free PAC tool, then listen to a few pages with a screen reader.
Questions people ask
Is my PDF uploaded to a server?
No. A script running in your browser tab opens the file and analyzes it on your own device. Nothing about it is sent to us, so you can check confidential documents safely. To see for yourself, disconnect from the network after the page has loaded. The check still works.
How do I know if a PDF is tagged?
Drop it here and read the Tags line. In Adobe Acrobat Reader you can also open File › Properties and look for “Tagged PDF: Yes” at the bottom of the Description tab. That line only says tags exist, not that they’re correct.
Why does Print to PDF produce an inaccessible file?
Printing only sends drawing instructions to the PDF printer driver, the same ones a paper printer would get. Headings, alt text, language and links aren’t part of a print job, so they’re lost. Use the program’s Export or Save As PDF command instead.
Can I make a scanned PDF accessible?
Partly. Optical character recognition (OCR) adds a text layer so the content can be read aloud and searched, and automatic tagging can then guess a structure. The recognized text needs proofreading, and headings, tables and reading order usually have to be corrected by hand.
What is the difference between PDF/UA and an accessible PDF?
PDF/UA (ISO 14289) is the technical standard that spells out what an accessible PDF must contain: complete tagging, embedded fonts, a displayed title, a language, alt text on figures. “Accessible PDF” is the everyday term, and in practice it means a file that meets PDF/UA and the relevant WCAG criteria.