Fix It in the Document, Not in the PDF

Remediating a PDF is slow, manual and expensive. Fixing the Word file it came from takes minutes, and the fix survives every future export. Almost nobody does it in that order, which is why document remediation is an industry.

This is about the stage before the PDF. If you only take one thing from it, take the checker.

The Checker Built Into Office

Word, PowerPoint and Excel all ship an Accessibility Checker, under the Review tab. It flags missing alt text, poor heading structure, low contrast and unclear link text, and it runs in seconds.

Its value is where it sits. A document fixed before it is published never becomes one of the thousands that need remediating later, which is the difference between a habit and a project. It does not judge whether alt text is correct, only whether it exists.

Word

• Use the built-in heading styles. Text made large and bold is not a heading, and the structure is what a screen reader navigates by. This is 1.3.1 Info and Relationships in the place it is cheapest to satisfy.
• Do not skip levels. Heading 2 followed by Heading 4 leaves a gap in the outline.
• Give every image alt text, or mark it decorative. Right-click, View Alt Text. Covered in how to write alt text.
• Use real lists, from the list buttons, rather than typed hyphens.
• Mark the header row on tables, and avoid merged cells, which are hard to follow in a linear reading order.
• Write link text that says where it goes. A pasted address read aloud is a string of characters.
• Set the document language, and mark any passage in another one, which is 3.1.1 Language of Page in document form.

PowerPoint

Slides have a problem the other formats do not. What you see is a canvas, and what a screen reader gets is an order somebody has to have set.

1. 1Give every slide a title, even if you hide it visually. It is how somebody navigates a deck, and an untitled slide announces as nothing.
2. 2Check the reading order in the Selection pane or the Reading Order pane. Content added outside a layout placeholder lands at the end, however it looks on screen.
3. 3Start from the built-in layouts. They carry placeholders that already have a sensible order. Boxes drawn onto a blank slide do not.
4. 4Alt text on every image, chart and grouped shape. A grouped shape needs its own description, since its parts mean nothing individually.
5. 5Do not put the substance only in the slide. Speaker notes and a transcript reach people the slide does not.

Excel

Spreadsheets get published as if they were documents and read as if they were tables, and the gap between those two is where the problems live.

• Name every sheet. Sheet1, Sheet2 and Sheet3 tell nobody anything, and the tab name is how a screen reader user moves between them.
• Use Format as Table so the header row is marked rather than merely bold.
• Avoid merged cells and blank rows. Both break the structure assistive technology uses to work out where a table starts and ends.
• Do not carry meaning in fill colour alone. A red cell is a colour, not a value, which is 1.4.1 Use of Color. Add a status column or a symbol.
• Give charts a text summary, because a chart image in a spreadsheet is exactly as opaque as one on a web page.

Exporting to PDF Without Losing It

The tags you have just created survive only if you export the right way. On Windows, Save As PDF with "Document structure tags for accessibility" enabled, or use the Acrobat add-in. Printing to PDF discards every tag and produces exactly the untagged file the whole exercise was meant to avoid.

Check the result rather than trusting it. Our PDF checker reads a file and reports whether it has tags, a language, a title and real text, and how to tag a PDF covers what to do when it does not.

What About the Documents You Already Have

Counting them usually decides the strategy on its own. A few dozen can be re-exported from source. Several thousand is a programme, and the honest first step there is deciding which files still deserve to exist.

1. 1Stop the inflow first, by making the checker part of publishing. Everything else is bailing.
2. 2Rank what exists by how often it is opened, and work from the top. Nobody needs a tagged copy of a memo from 2015.
3. 3Re-export from the source file wherever the source survives, which is faster than repairing the PDF.
4. 4Consider turning the most-used documents into web pages, which are easier to maintain and more accessible than any PDF. A page beside a PDF only counts as a replacement when four conditions hold at once, so check those before you retire the file.

The same reasoning, applied to an intranet, is in our SharePoint guide. If the volume is beyond what your team can absorb, the job goes out to a tagging supplier, and how to scope and buy document remediation is the page that tells you what to ask them.