A scanned PDF is a picture of text, and many exported PDFs have no headings or reading order. Put rates, rules, hours and forms on the page as text, and export the remaining PDFs properly.

Documents from the office copier

On many small business websites, the most important information is not on a page. It is in a PDF: the rate sheet, the cancellation policy, the application form, the rules.

Often the PDF came from the office copier. Somebody laid the printed document on the glass, pressed scan, and uploaded whatever came out.

On screen it looks exactly like the paper. To a screen reader, the software blind and partially sighted people use to hear a page read aloud, it is a photograph with nothing in it to read.

Other PDFs are exported from Word or a spreadsheet. They contain real text, and still often fail.

Two kinds of unreadable PDF

A scanned PDF is an image of a page. Unless somebody has run text recognition on it, the file holds no letters at all, only dots arranged to look like letters.

An exported PDF usually has the letters but not the structure. A well-made one carries hidden tags marking headings, lists and table columns, and the order to read things in.

Without those tags, a screen reader guesses. Headings sound like body text, two columns are read straight across as one, and a table becomes a stream of unconnected numbers.

The WCAG accessibility guidelines are generally applied to a website's documents as well as its pages.

What a screen reader user meets

The first is the most common on small business sites, because scanning is the quickest way to put a document online. It is also the most complete failure.

Awkward for everyone else too

A PDF is designed for a sheet of paper, and a phone is not one.

On a phone it may open in a separate viewer or download as a file, and the visitor pinches and drags across a printed layout looking for one figure.

Nothing in a scan can be tapped, searched or copied, so even a phone number on it has to be retyped.

Put the important content on the page

Most documents on a small business site are doing a job a web page does better.

Rates, hours, rules, policies and instructions belong on pages as ordinary text, with real headings and proper tables. A page reads aloud correctly, fits a phone, and takes a minute to update.

Forms belong on the page too. A form that has to be printed, filled in by hand, scanned and emailed back is a chore for nearly everyone, and impossible for some.

A PDF can still sit beside the page for anybody who wants to print. It should be a copy of the page, not the only version.

A worked example

A co-working space above a row of shops posted four PDFs on its membership page: a rate sheet, the house rules, a membership application and the membership agreement.

The rate sheet was exported from a spreadsheet without tags, the house rules were a crooked scan of a photocopy, and the application had to be printed and filled in by hand.

A prospective member who uses a screen reader wrote to say she could hear none of the house rules and only a jumble of the rates. Nobody there had ever checked.

The rates became a table on the membership page, the house rules a page of their own with headings, and the application a web form.

The membership agreement, which has to be signed, stayed a PDF, exported again from the original Word file with tags, a title and a language set.

Making the remaining PDFs readable

Some documents should stay as PDFs: agreements to sign, brochures meant for print, product sheets from a manufacturer.

The fix starts in the original file. In Word, use the built-in heading styles rather than bold text, describe the images, and check that the accessibility tags option is ticked when saving as PDF.

Printing to PDF, rather than saving as one, usually throws that structure away. InDesign can export tagged PDFs too.

Set a title and a language in each document's properties, and keep the source files, so the PDF can be remade in minutes when something changes.

Links to a PDF should say so, with the length: membership agreement, PDF, two pages. Visitors on phones then know a download is coming.

When only a scan exists

When the original file is long gone, text recognition, built into Adobe Acrobat Pro and much scanning software, can turn a scan into real text. For a short document, retyping is often quicker and more reliable.

Check the result against the paper either way. Recognition misreads figures, and a price that quietly changes from 45 dollars to 46 is worse than no text.

Checking a PDF in two minutes

Open the file and try to select a sentence. If nothing highlights, it is a scan. Search for a word you can see; if the search finds nothing, neither will a screen reader.

The free Acrobat Reader has Read Out Loud in its View menu. It is not a screen reader, but it reads a scan as silence and an untagged file in whatever order it guesses.

Acrobat Pro adds an accessibility checker that lists missing tags, titles and descriptions. Whoever prepares your documents can run it before anything is uploaded.

The counter-case

Not every PDF on a site needs attention this month.

An archive of old newsletters that nobody opens matters less than the rate sheet everybody does. A PDF offered beside a page with the same content is a convenience, not a barrier.

A site with hundreds of documents can begin with the ones customers open most.

Either way, put the content customers need on the page, and fix the PDFs that remain at the source.

Document by document

  1. List every PDF linked from the site.
  2. Try to select the text in each one.
  3. Move rates, rules and hours onto pages.
  4. Rebuild forms as web forms.
  5. Re-export the rest from the original files.
  6. Set a title and language on each.
  7. Say PDF and the length in every link.

Step three is the one that helps the most people at once. A page works for screen readers, phones and search engines together, and it is the version that gets kept up to date.

Another way a page can leave a screen reader with nothing to say is covered in buttons that are only an icon.

Frequently asked questions

Why can a screen reader not read a scanned PDF?

A scan is a picture of a page. Unless text recognition has been run on it, the file holds no letters at all, only an image of them.

What is wrong with an exported PDF?

Often nothing you can see. Without tags marking headings, lists, tables and reading order, a screen reader guesses, so columns run together and tables become a stream of numbers.

Which documents should become web pages?

Rates, hours, rules, policies, instructions and forms. On a page they read aloud correctly, fit a phone, can be searched, and take a minute to update.

How do I make a readable PDF from Word?

Use the built-in heading styles, describe the images, and save as PDF with the accessibility tags option ticked. Printing to PDF usually throws the structure away.

What if only a scan exists?

Run text recognition on it or retype it, then check every figure against the paper. Recognition misreads numbers, and a wrong price is worse than no text.

How can I check a PDF quickly?

Try to select a sentence and search for a visible word. If neither works, it is a scan. Read Out Loud in Acrobat Reader gives a rough idea of reading order.

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.

Rates and forms posted as scanned PDFs?

Send us the documents. We will tell you which belong on a page and rebuild the rest so they can be read aloud.

Start a Conversation