The problem with scanned documents
A scanned PDF contains pictures of pages, not text. Screen readers announce nothing — at best "graphic", at worst silence. The document can't be searched, text can't be resized or reflowed, and no amount of tagging fixes it, because there is nothing to tag. Archives, signed contracts, legacy reports, and incoming correspondence are the usual offenders — and they are just as covered by Section 508, the ADA, and the European Accessibility Act as anything else you publish.
OCR alone is not accessibility
Running automatic OCR (in Acrobat or a scanner driver) produces a searchable text layer — useful, but nowhere near compliant. Automatic OCR gives you no heading structure, no reading order, no table semantics, no alt text, and frequent recognition errors that a screen reader will read aloud verbatim. That's why our process is OCR plus full remediation:
- High-accuracy OCR with manual review of every uncertain word
- Correct text order rebuilt for multi-column layouts and sidebars
- Complete PDF/UA tag structure: headings, lists, tables, links
- Alt text written for stamps, signatures, logos, photos, and figures
- Document language, title, and metadata set correctly
- Validated to PDF/UA, HSS, and WCAG 2.0 / 2.1 / 2.2 in PAC 2024
- Delivered with an issue report and signed ACR/VPAT
Typical use cases
Government archives going online, scanned board minutes and policies, legacy manuals and forms, signed agreements that must be published, historical publications, and incoming scanned mail that public-sector bodies must serve accessibly. Batch pricing is available for large archives — and large batches are exactly where DeepTag, our in-house tagging software, earns its keep: repetitive structure work goes faster, expert review stays thorough.
Frequently asked questions
Why is a scanned PDF not accessible?
It's a photograph of a page — there is no text for assistive technology to read. OCR creates the text; remediation adds the structure (headings, order, tables, alt text) that screen readers need to present it sensibly.
Is automatic OCR enough?
No. Auto-OCR yields searchable text but no reliable structure, and recognition errors go straight into the screen reader output. PDF/UA requires correct tags, order, and metadata — verified by a person and validated in PAC 2024.
Can poor-quality or handwritten scans be fixed?
Usually yes. We correct low-quality OCR manually, and transcribe handwriting where the engine can't read it, so the delivered document is complete and accurate.
Find out if your scan is readable — free
Upload it to the checker; the score tells you instantly whether a text layer exists.
Run the free check →