If you can select individual words, your PDF contains real text. Use Part 1 to make sure the rest of the document is in good shape.
If clicking and dragging selects the whole page as one block, or nothing at all, your PDF is an image. Go to Part 2.
The best time to make a PDF accessible is before it becomes a PDF. Everything below happens in the source document (Word, Google Docs, etc.) and takes far less effort than fixing problems later.
Export from the digital original, never from paper. If the content exists as an editable file, export the PDF directly from it. Printing a document and scanning it back in throws away the real text and replaces it with a picture.
Use real heading styles. Use the built-in heading styles (Heading 1, Heading 2) for section titles instead of manually making text big and bold. Headings carry structure into the PDF so learners and assistive tools can follow how the content is organized. Don't skip levels.
Describe meaningful images. Any image that carries information (a diagram, a chart) needs alt text: a one-sentence description of what it shows. In Word and Google Docs, right-click the image and look for "Alt text." Ask yourself: if a learner couldn't see this image, what would they need to know?
Keep text readable. Use a clean, common font at 11pt or larger. Keep strong contrast: dark text on a light background, never text over busy photos. Don't rely on color alone to convey meaning ("items in red are required" fails for anyone who can't distinguish the color).
Use real lists and tables. Use the document's built-in list and table tools rather than typed dashes or spaced-out columns. Keep tables simple: one header row, no merged cells.
Export the right way. In Word, use Save As or Export to PDF with "Document structure tags for accessibility" checked (the default). In Google Docs, File > Download > PDF. Avoid "Print to PDF," which strips out the structure you just built.
Sometimes you only have a scan: a photocopied handout, an old workbook, a PDF made from paper years ago. The fix that works best is recreating the document.
Before anything else, ask whether anyone still has the original file in Word, Docs, or a PDF editing tool. Check with the author or whoever sent you the file. Exporting a fresh PDF from the original takes a minute and gives a perfect result.
If the original is gone, retype it. For most handouts and short documents this is faster than people expect, and the result is dramatically better than any repair: smaller, more reliable on tablets, and fully accessible once you follow Part 1. A recreated document always beats a repaired scan.
For long documents you can't realistically retype, OCR tools (like Adobe Acrobat's "Scan & OCR" feature) can read the page images and add a real text layer. Treat this as a last resort, and know its limits:
OCR makes mistakes, especially with handwriting, decorative fonts, and low-quality scans. Always check the result: select text on several pages and skim for garbled words.
OCR adds text but not structure. The result won't have proper headings, image descriptions, or reliable reading order.
Never upload documents with personal or sensitive information to free online OCR services.
For content learners will use heavily, recreation is always your best option.
I can select individual words with my cursor.
Section titles use real heading styles.
Meaningful images have short text descriptions.
Text is 11pt or larger with strong contrast.
The file was exported from the source document, not printed and scanned.