You can spend weeks fixing your website’s HTML, pass an automated scan with a green checkmark, and still hand a blind customer a wall they can’t get through. It’s usually the same wall every time: a PDF. The takeout menu. The monthly statement. The “download and fill out” intake form. These files feel like an afterthought, which is exactly why they’re the most common accessibility gap we find — and one of the easiest for a plaintiff to point at.
Why PDFs slip through the cracks
A PDF is a separate file that happens to be linked from your site. Your accessibility scanner crawls the page, sees a link, and moves on — it almost never opens the document and reads its internal structure. So a site can score perfectly while every menu, brochure, and form behind a download link is completely unreadable to a screen reader.
That blind spot matters legally. The DOJ’s 2024 Title II rule explicitly lists “word processing, presentation, PDF, and spreadsheet files” as covered web content that must meet WCAG 2.1 Level AA, with large public entities required to comply by April 26, 2027 and smaller ones by April 26, 2028. That rule binds government, not the corner restaurant — but it has set the expectation that documents count. For private businesses under ADA Title III, courts and demand letters increasingly treat a downloadable PDF as part of the website it lives on. Digital accessibility filings have run in the thousands per year, with more than 4,000 filed in 2024 alone, and PDFs are a recurring item in the complaints.
What actually makes a PDF accessible
The visible page tells you nothing. Two PDFs can look identical on screen — same fonts, same layout — while one is fully accessible and the other is a flat image a screen reader announces as “blank.” The difference lives in five things underneath, per the W3C’s PDF techniques and the PDF/UA standard (ISO 14289):
- Tags. A hidden tag tree labels every piece of content —
<H1>,<P>,<L>(list),<Table>,<Figure>. Tags are the structure assistive technology actually reads. An untagged PDF is the single most common failure: there’s simply nothing for a screen reader to interpret. - Reading order. Tags have to be in the order a human would read them. Autotagging routinely scrambles multi-column layouts, sidebars, and pull quotes, so a restaurant menu can read its prices before its dish names, or jump from column one to a footer and back.
- Alt text. Every meaningful image, chart, and logo needs alternative text that conveys its content and purpose. Decorative flourishes get marked as artifacts so the screen reader skips them instead of announcing noise.
- Form fields. Interactive PDFs — the intake forms, the applications — need every field labeled with a name and tooltip, in a logical tab order, so it can be completed by keyboard alone. This is where PDFs are hardest, and where our accessible-forms guidance applies directly.
- Document language and metadata. A declared language tells text-to-speech engines how to pronounce the words, and a real title shows up in the screen reader instead of “document1-final-v3.pdf.”
A file that gets all five right generally satisfies the relevant WCAG criteria. Miss the first two and the document is effectively unreadable, no matter how polished it looks.
The three documents that cause the most trouble
Menus are usually designed in InDesign or Canva and exported with no tags at all — often as a single flattened image. A screen reader user gets nothing. For a restaurant, that’s the entire reason a customer visited the site.
Statements and notices — invoices, billing statements, benefit letters — are frequently generated by back-office systems that emit untagged output by the thousands. Because they’re personalized, you can’t remediate one and call it done; the template has to produce tagged PDFs.
Forms are the worst offenders because they combine every problem: scanned origins, scrambled reading order, and unlabeled fields. A form that can’t be completed independently forces a disabled user to call, email, or give up — the exact “barrier to access” complaints are built on.
How to test your own PDFs
You don’t need a lab. Start here:
- Try to select the text. Open the PDF and drag your cursor across a paragraph. If nothing highlights, it’s a flat scan with no real text — the worst case, and you’ll need OCR before anything else.
- Run an automated check. Adobe Acrobat Pro’s “Full Check” flags missing tags, alt text, and language. The free, more thorough PAC checker validates against PDF/UA and WCAG and gives a detailed report.
- Read it with a screen reader. Open it in NVDA (free, Windows) or VoiceOver (built into Mac) and navigate by headings, then through a table. This is the step that catches reading-order and meaning errors no automated tool can — the same reason automated scans miss so much on the HTML side.
If you only do one thing, do step three. A document can pass an automated check and still read like word salad.
Why overlays can’t help — and what does
This is the part that surprises people who bought an accessibility widget. Overlays are JavaScript that runs on HTML pages. The instant a user clicks a download link and opens a PDF in their own viewer, your overlay isn’t running anymore. It cannot add a single tag, alt attribute, or reading-order fix to the file. The barrier is untouched.
The only real fix is remediating the file itself — ideally by building structure into the source document first, then exporting a clean tagged PDF, and verifying with a checker plus a real screen reader. Our full accessible-PDFs guide walks through every step, and our manual remediation service does it for you when the document count is high or the originals are gone.
A note on the legal side: this is general information, not legal advice. ADA case law moves fast and varies by circuit and state, so if you’ve received a demand letter or want to gauge your exposure, talk to a qualified accessibility attorney.
The good news is that PDFs are a contained problem. Unlike a sprawling site, you can inventory your documents, prioritize the high-traffic ones — menu, top forms, current statements — and clear most of your risk in a focused pass. The hard part is just knowing they’re there. Start with a free scan and we’ll help you find the files your site is quietly hiding.