Remediating an inaccessible PDF means adding the structure a screen reader and other assistive technology need to make sense of the file — real text, tags, headings, reading order, alt text, and labeled forms. The most important principle comes before any of that, though: whenever you still have the source file, fix accessibility at the source and re-export, because that is faster, cheaper, and more reliable than patching a finished PDF. This walkthrough covers both: doing it right at the source, and remediating an existing PDF step by step.
This is a hands-on companion to our ADA PDF compliance checklist — use the checklist to confirm you're done, and this guide to actually do the work.
Start at the source whenever you can
If the PDF was created from Word, InDesign, PowerPoint, or a similar tool, and you still have that file, the single best thing you can do is fix the source and export a tagged PDF. Authoring tools understand structure natively, so the work you do there carries straight through into the PDF's tag tree.
In the source document:
- Use the program's built-in heading styles (Heading 1, Heading 2, and so on) rather than just making text big and bold.
- Add alt text to images.
- Build tables with the program's real table feature and mark header rows.
- Use real bulleted and numbered lists instead of manually typed dashes.
- Set the document language and title in the file's properties.
- Export using the "tagged PDF" / "PDF for accessibility" option, not "print to PDF," which usually strips structure.
Fixing the source is almost always cheaper than remediating the output, and it means the next revision exports cleanly too. Reach for the steps below when there is no source file, or when the source is locked or owned by someone else.
Remediating an existing PDF, step by step
These steps are ordered deliberately. Some early steps make later ones easier, so work roughly top to bottom rather than jumping around.
1. Set the document title and language
Start with two quick metadata fixes. Set a meaningful document title so assistive technology announces the document by name rather than a filename, and configure the document to display that title instead of the filename. Set the document language so the screen reader uses the correct pronunciation. These are small, high-value changes and a common reason files fail an automated check.
2. Make sure there is real text
A scanned document is just an image of text — there is nothing for a screen reader to read. Run OCR (optical character recognition) on any scanned pages to convert the image into selectable, readable text. Always proofread the OCR output, since recognition errors are common with poor scans, unusual fonts, or low contrast. If a page is genuinely just a picture of text, no amount of tagging will fix it until OCR has run.
3. Auto-tag, then correct the tag tree
Tags are the hidden structure that tells assistive technology what each element is — a heading, a paragraph, a list, a table. If the PDF has no tags, run the tool's auto-tag feature to generate a first draft. Auto-tagging is a starting point, not a finish line: it guesses, and it gets plenty wrong. Open the tag tree and correct it — fix mislabeled elements, remove empty or duplicate tags, and make sure every piece of real content is tagged with the right type. For a deeper explanation of how tags work, see our guide to PDF tags and reading order.
4. Fix the reading order
The reading order is the sequence in which assistive technology presents content. It often differs from the visual layout — multi-column pages, sidebars, and callouts are common culprits. Check that content reads in the logical order a person would expect: headline, then body, then sidebar, not a jumble. Reading order and the tag tree are closely linked, which is why both are covered in the tags and reading order guide.
5. Mark real headings
Headings are how screen reader users navigate — they jump heading to heading the way a sighted reader scans a page. Make sure visual headings are tagged as actual heading levels (H1, H2, H3) in a logical hierarchy, without skipping levels. Don't tag something as a heading just because it's bold; tag it as a heading because it is one.
6. Add alt text to images
Every meaningful image needs alternative text that conveys what it communicates. Write concise, descriptive alt text for informative images, charts, and logos. Describe the information, not just the appearance — for a chart, summarize what it shows. Our guide to alt text for images in PDFs covers how to write good descriptions and how to handle complex graphics.
7. Tag tables correctly
Tables are one of the hardest elements to get right. A correctly tagged table identifies its header cells and associates them with the data cells they describe, so a screen reader can announce "Revenue, Q3: $4.2M" instead of reading a wall of disconnected numbers. Check that header rows (and header columns, where relevant) are marked as headers and that the table's row-and-column structure is intact. See accessible tables in PDFs for the details, including how to handle merged and nested cells.
8. Tag lists
Bulleted and numbered lists should be tagged as real list structures, not as a series of stray paragraphs that happen to start with a dash. Correct list tagging lets assistive technology announce the list and the number of items, which helps users orient themselves.
9. Label form fields and set tab order
If the PDF contains a form, every field needs an accessible label (a tooltip/name) so a screen reader can tell the user what to enter. Then set a logical tab order so keyboard users move through fields in a sensible sequence, and make sure required fields and validation are conveyed accessibly. Forms have enough nuance that we cover them separately in accessible PDF forms.
10. Mark decorative content as artifacts
Not everything on a page carries meaning. Background images, decorative rules, page borders, and repeated headers or footers should be marked as artifacts so assistive technology skips them. This declutters the experience — a screen reader user shouldn't have to wade through "decorative line" on every page.
11. Add bookmarks for long documents
For long documents — reports, manuals, contracts — add bookmarks that mirror the heading structure. Bookmarks give everyone, including assistive technology users, a fast way to jump to a section instead of scrolling through dozens of pages.
12. Run an automated checker
Once the structural work is done, run an automated accessibility checker. Automated tools quickly catch missing titles, untagged content, missing alt text, and language errors. They're excellent at flagging mechanical problems — but understand their limit: they can confirm an image has alt text, not whether that alt text is meaningful. Fix what the checker flags, then re-run it until it's clean.
13. Test with a real screen reader
The final step is the one people skip and shouldn't: test with an actual screen reader. Tab through the document, listen to how it reads, and confirm the reading order, headings, tables, and forms all make sense out loud. Automated checkers verify the mechanics; a real screen reader verifies the experience. If it's confusing to listen to, it isn't done.
When to automate and when to do it manually
Not every step deserves the same effort, and not every document deserves the same approach.
- Automate the mechanical, high-volume work. Setting titles and language, detecting untagged content, generating a first-pass tag tree, OCR, and flagging missing alt text are all repetitive tasks that automation handles well — especially across large batches of similar files.
- Reserve manual work for judgment calls. Writing meaningful alt text, getting reading order right on complex layouts, and correctly tagging intricate tables and forms still benefit from human review.
- Match the approach to volume. A handful of one-off documents can be remediated by hand. A large back-catalog of thousands of files is where automation, with human review reserved for the genuinely complex pages, pays for itself — a tradeoff we quantify in the true cost of PDF remediation.
If you produce documents at scale — common in regulated and high-volume sectors covered in our PDF accessibility by industry overview — the most durable fix is to stop the problem at the source so new files are accessible from the moment they're created.
Key takeaways
- Fix it at the source whenever you have the original file — structure your Word/InDesign document well and export a tagged PDF rather than patching the output.
- For existing PDFs, work in order: title and language, real text (OCR), tags, reading order, headings, alt text, tables, lists, forms, artifacts, and bookmarks.
- Run an automated checker to catch mechanical failures, then test with a real screen reader to confirm the document actually makes sense out loud.
- Automation handles repetitive, high-volume tasks; reserve manual review for alt text, complex reading order, tables, and forms.
- The bigger your document volume, the more it pays to remediate at scale and prevent new inaccessible files at the source.
