Make a scanned PDF accessible for screen readers

Is your PDF a scan? A ten-second test

1. Open the PDF and try to select one word with your mouse, as if to copy it.
2. Or press Ctrl + F (⌘ + F on a Mac) and search for a word you can see on the page.

If you can't select the word, or the search finds nothing, the page is a picture of words. A screen reader says nothing, or just "image", and nobody can search it, copy from it or make the text bigger without it going blurry.

Agendas, minutes and packets that were printed, signed and fed through the copier almost always fail this test. Which of them the ADA Title II rule covers is in PDFs, agendas and minutes under the ADA Title II rule.

The easiest fix: post it as a web page

For agendas, minutes and notices, typing or pasting the text into a page on your website is usually less work than repairing a scan, and it reads better on a phone.

• Ask whoever wrote it for the original file (the document they typed), and paste the text into a new page.
• Make the meeting name and each agenda item a heading, not bold text: Headings in order shows how.
• Give links words that say where they go: "Item 4: budget amendment", not "click here" ("Click here" and other link text that fails).
• Keep the signed PDF as well if your records policy or your state's open-meetings law asks for one. The web page is the copy most people will read.

When it has to stay a PDF: OCR, then tags

Two words you'll meet. OCR (optical character recognition) means the program reads the picture and puts real, selectable text behind it. Tags are hidden labels inside the PDF that tell a screen reader what each part is (a heading, a paragraph, a list, a table) and what order to read it in. A scan needs both.

1. Look for the original first. If the document was typed, the typed file saved as a tagged PDF beats any repaired scan. In your word processor use Save as or Export and tick the accessibility or tagged PDF option; "Print to PDF" usually throws the structure away.
2. Run OCR. In your PDF program, look for a feature called something like Recognize text, Scan and OCR, or Make searchable. It usually needs a full PDF editor, not just a free reader. Many office copiers can also scan straight to a searchable PDF: look for that setting before you scan the next one.
3. Check the words it recognized. OCR misreads smudges, stamps, handwriting and small print. Pay most attention to numbers, dollar amounts, dates and names: a wrong figure in a budget line is worse than no text. Copy all the text into a blank document and read it through.
4. Add tags. Your PDF program's accessibility tools will have something like Add tags, Autotag or Make accessible. It guesses the headings, lists and tables, so check its guesses.
5. Set the title and language in the document's properties, so a screen reader announces "Planning Board agenda" in English, not a file name.
6. Describe what isn't text. Give maps, photos and charts a short description (alt text). A signature can be described as "Signed by the town clerk".
7. Run the accessibility checker in your PDF program and fix what it flags.

Check it the way a reader would

An automatic checker confirms the tags are there. It can't tell you the recognized words are right or that the reading order makes sense. So also:

• Read the text you copied out in step 3, start to finish. It should be in the same order as the page.
• Open the PDF with the screen reader built into your computer (Windows and Mac both have one, in the accessibility settings) and listen to the first page.
• Try the select-a-word test again. Every word should be selectable now.

If your web person or a vendor mentions PDF/UA, that's the international standard for accessible PDFs (ISO 14289). Tags are what it's built on.

When a scan isn't worth repairing

Some documents won't come out well however long you spend: handwritten notes, faint copies, hand-drawn maps, a 200-page packet from 2009. For those:

• If people still use it, retype the parts they need into a web page.
• If it's only kept for the record, the Title II rule may not require it (the guide above explains when).
• Either way, say on your accessibility page how people can ask for a copy they can use, and answer quickly when they do.

What PageKiwi can check

PageKiwi doesn't open PDFs, so it can't check one. Once an agenda is a web page, Accessibility Check looks at that page's headings and links, and Readability marks long sentences and jargon.