How to add alt text to a PDF
Alternative text is what a screen reader says when it reaches an image. Getting it in place is mechanical. Writing a good one takes thirty seconds of judgment — and that is the part worth learning.
Where it goes
In Word, before you export
Right-click the image → View Alt Text → type the description in the panel. If the image carries no information — a decorative divider, a background flourish — tick Mark as decorative instead. That tells a screen reader to skip it silently, which is correct and much better than an empty description.
Then export with File → Save As → PDF → Options → "Document structure tags for accessibility." Printing to PDF discards every description you just wrote.
In Acrobat, on a finished PDF
All tools → Prepare for accessibility → Set alternate text. Acrobat walks through every image in the document one at a time, with a box for the description and a checkbox for decorative. For a document with a handful of images this takes a few minutes.
Auto-tagging does not write alt text
Acrobat's automatic tagging creates structure but leaves every image without a description. Neither do the AI tools that promise one-click remediation — or when they do, the result is generic. This step needs a person who knows what the image is for.
What to actually write
The rule is not "describe the image." It is "convey what the image conveys." The same photograph gets different alt text depending on why it is on the page.
The four cases you will actually hit
Decorative
A stock photo of a sunset next to a parks announcement, a divider line, a background texture. Mark as decorative. Do not describe it. A screen reader user does not need to hear about the sunset; they need to get to the announcement.
The seal, the logo, the letterhead
Write what it is, not what it looks like: City of Springfield seal. Not
circular blue emblem with an eagle and a banner reading Founded 1847. Nobody needs
the heraldry.
A chart or a graph — this is the one people get wrong
Do not describe the picture. State the finding.
| Weak | Useful |
|---|---|
| Bar chart showing budget by department | Bar chart: Public Safety is the largest share of the 2026 budget at 42%, followed by Public Works at 23% and Administration at 11%. |
| Line graph of water usage | Line graph: residential water usage rose steadily from 2019 to 2024, with a sharp drop in 2021 during the conservation order. |
If the chart is complex enough that a sentence cannot carry it, put the underlying numbers in a real table elsewhere in the document and let the alt text point to it.
A map or a site plan
These are genuinely hard. Describe the purpose and the answer, not the geography:
Map showing the proposed rezoning area: the six parcels on the north side of Elm Street
between 4th and 6th Avenue. If the map is the entire point of the document — a zoning
map, a flood map — the honest answer is that alt text cannot substitute for it, and you should
provide the same information another way: a list of affected addresses, a table of parcels, a
phone number for someone who will read it out.
Length and phrasing
- Aim for one sentence. Two if the image is carrying real information.
- Skip "image of" and "picture of." The screen reader already announced that it is an image. Starting with those words wastes the listener's time on every single image.
- Don't repeat the caption. If a caption below the image already says it, mark the image decorative — otherwise the user hears the same thing twice.
- Include text that appears inside the image. A flyer saved as a picture is invisible to a screen reader; every word on it has to be in the description, or better, the flyer should not be an image in the first place.
The check that takes two seconds
Read your document out loud, and where an image appears, read your alt text instead of looking at the picture. If the sentence still makes sense and you did not lose anything, the description works. If you find yourself glancing at the image to follow along, it doesn't yet.
A note on scanned documents
If someone scanned a paper flyer and posted it, the entire page is one image. Alt text is not the fix — you would be writing a description of an entire document. The real fix is to recreate it as a real document, or run text recognition and tag the result. Our checker flags this case specifically, because it is common and because it is the one that leaves a reader with literally nothing.
See where your own documents stand
Enter your website and we'll inventory every PDF on it — how many fail, how many pages, and which ones are forms. Free, no signup.
Check your site