The PDF Nobody Can Read

The three levels, worst first

A scan. A photograph or scan of a printed page, saved as a document. There is no text in it at all, only an image. A screen reader announces nothing, search engines index nothing, and a phone user pinches and drags across a page-sized image.

This is the most common form of small business document and the least usable thing on most sites.

A generated document that was never structured. Real text, so it can be read aloud and searched, but with no headings, no reading order, no table structure and no alternative text. Better, and still hard work to navigate.

A tagged document. Text with structure: headings marked as headings, tables with header cells, images described, and a defined reading order. This is what an accessible document actually means.

Most businesses believe they are at the third level and are at the first.

The question to ask first

Why is this a document rather than a page.

For most small business content the honest answer is habit. A price list, a service menu, an opening hours notice, a form, a brochure and a specification sheet are all better as web pages.

A page is readable by everybody, findable in search, works on a phone, costs nothing to update, and does not require anybody to download anything.

Documents make sense for things genuinely meant to be printed, filled in by hand, or preserved as a fixed record: a signed agreement, a certificate, a technical drawing, a form somebody takes to an appointment.

Everything else is a page that got saved in the wrong format.

What it costs to leave it as a document

Beyond accessibility, which is reason enough.

Content inside a scanned document is invisible to search, so a price list published that way ranks for nothing and answers no query.

It cannot be linked to at the section level, so somebody asking about one service gets sent a fourteen page brochure.

It is heavy, which matters on a phone on mobile data more than most people account for.

And it goes stale silently, because updating it means finding the original file, which after two years frequently nobody has.

If the document has to exist

The practical requirements, in order of impact.

1. Generate it from a text source rather than scanning. Export from the word processor, do not print and scan.
2. Use real headings in the source document, not text made large and bold.
3. Add alternative text to any image carrying information.
4. Mark table headers so cells can be associated with them.
5. Set the document language and title in its properties.
6. Check the reading order, which multi-column layouts commonly scramble.

Most of that happens in the source document rather than in the finished file, which is the useful insight: an accessible document is produced by writing it properly, not by fixing it afterwards.

Where you have an existing scan and no source, optical character recognition can extract the text, and the result needs checking rather than trusting, since it handles tables, columns and unusual typefaces badly.

Offering both

The arrangement that works when a document is genuinely needed.

Publish the content as a page, and offer the document alongside it for anybody who wants to print or keep it.

That serves everybody: the screen reader user reads the page, the search engine indexes the page, the phone user reads the page, and the person who wanted a printable copy gets one.

It is more work than uploading a file and it is the difference between content that works and content that merely exists.

Linking to one honestly

Small courtesies that matter.

Say it is a document in the link text, and say how large it is. Somebody on a limited connection deserves to know before tapping.

Do not open it in a new window without saying so, since an unexpected context change is disorienting for screen reader users and confusing for everybody else.

Avoid link text that is only the filename, which tells the reader nothing about what they are opening.

The public sector angle

Worth knowing if you sell to one.

Public sector bodies in several jurisdictions have obligations covering the documents they publish, not only their web pages, and that obligation flows down to suppliers producing material for them.

Which means a supplier who can deliver properly tagged documents has something to say that most cannot, and a supplier who delivers scans is creating a problem for their client.

Requirements differ by jurisdiction and are set by the applicable authority, so the specifics are worth confirming for the buyer you are dealing with rather than assumed.

The general content-level work this builds on is set out in accessibility without the jargon.

Frequently asked questions

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.