# Searchable is not accessible: what OCR leaves out Canonical: https://paperlyft.com/articles/searchable-pdf-vs-accessible-pdf By Kirtan Thummar Published: 2026-10-21T12:02:25.244Z Updated: 2026-10-21T12:02:25.244Z OCR makes a scanned PDF searchable, but a screen reader needs much more than text. Here’s what OCR leaves out, why its mistakes matter more for screen reader users, and how to turn a searchable PDF into an accessible one. Many offices run OCR on their scanned documents and assume the job is done. The file is now searchable. You can copy text out of it. It feels fixed. Here is the short answer. OCR makes a PDF searchable, not accessible. It adds text, but a screen reader also needs structure: headings, reading order, table relationships, form labels and image descriptions. OCR adds none of those. In our [first post](), we looked at what a screen reader does with a raw scan. This one goes a step further: what happens after OCR, and what is still missing. ## Key takeaways - OCR adds a hidden text layer. That makes a PDF searchable, not accessible. - An accessible PDF also needs tags, a correct reading order, table and form structure, alt text, a title and a language. - OCR errors matter more for screen reader users, because they only get the hidden text, never the image. ## What does OCR actually do to a PDF? OCR (optical character recognition) looks at the picture of each page and guesses which letters are there. It then places an invisible text layer behind the image, lined up with the words you see. The result is often called a searchable PDF. You can search it, select text and copy words out. That is useful, but only as good as the recognition. OCR still makes mistakes, especially with handwriting, faded scans, stamps and unusual fonts. And even when the text is right, the file is still a picture with loose text behind it. Nothing in it says what is a heading, what is a table, or which column comes first. ## Searchable PDF vs accessible PDF | Features | Searchable PDF (OCR only) | Accessible PDF | | --- | --- | --- | | Text you can search and select | Yes | Yes | | Headings, lists and paragraphs tagged | No | Yes | | Correct reading order | Not guaranteed | Yes | | Tables a screen reader can follow | No | Yes | | Labeled form fields | No | Yes | | Alt text for images, stamps and signatures | No | Yes | | Document title and language set | Usually Not | Yes | | Meets WCAG 2.1 AA and PDF/UA | No | Yes | ## What OCR leaves out **Structure.** OCR doesn't know a heading from a footnote. Without tags, a screen reader user can't jump between sections. They have to listen to everything, start to finish. **Reading order.** Two-column pages, sidebars and text boxes often come out in the wrong sequence. The words are there, but in an order that makes no sense. **Tables.** OCR reads a table as rows of loose words. A screen reader user hears numbers with no idea which column or row they belong to. **Forms.** Field labels and checkboxes become plain text. Nothing tells the reader which label goes with which blank. **Images.** Logos, seals, stamps, charts and signatures get no description at all. **Repeated clutter.** Page numbers, headers and footers get read aloud on every single page. ## Why OCR errors matter more for accessibility OCR makes mistakes. A zero becomes the letter O. A one becomes a lowercase L. A faded word turns into symbols. For a sighted person, this rarely matters. They see the image, not the hidden text. For a screen reader user, the hidden text is all they get. If OCR misreads a date, a dollar amount or a parcel number, they hear the wrong one, and nothing tells them it is wrong. A sighted reader sees the page. A screen reader user hears only what OCR guessed. Handwriting makes this harder. Standard OCR is built for printed text, and handwritten forms, notes and signatures often come out incomplete or wrong. We'll cover handwritten documents in their own post. ## How to turn a searchable PDF into an accessible one - Check the text is accurate, especially names, dates, numbers and amounts. - Tag the structure: headings, paragraphs, lists, tables and form fields. - Set the reading order so the page reads the way a person would read it. - Add header cells to tables and labels to form fields. - Write alt text for images, seals, stamps and signatures. - Set the document title and language, and mark page numbers and headers so they aren't read on every page. - Validate against WCAG 2.1 AA and PDF/UA, then have a person review the result. OCR is the first step on that list, not the whole list. The rest is where most of the time goes when it is [done by hand](). ### Questions and answers **Does OCR make a PDF accessible?** No. OCR adds searchable text, but an accessible PDF also needs tags, reading order, table and form structure, and alt text. **Is a searchable PDF ADA compliant?** Not on its own. Under ADA Title II, PDFs on government websites need to meet WCAG 2.1 AA, which a searchable PDF usually doesn't. *(Link "ADA Title II" to the Title II post.)* **Can OCR read handwriting?** Standard OCR struggles with it. Handwritten documents usually need recognition built for handwriting, followed by careful review. Ran OCR on your files? Check whether they're really accessible. [Check for free]() ## Sources - ["a screen reader also needs structure" in the short answer]()