Alt text in slide decks and PDFs: the files nobody audits

Somebody posts Tuesday's board deck to the website. Fourteen slides, three charts, a campus map, a logo wall. It goes up as a PDF and every image in it is silent — a screen reader reaches the chart, says nothing useful, and moves on. Nobody flagged it, because a deck feels like a handout rather than a page.

It stops being a handout the moment it's on the site. The Justice Department's Title II web rule defines web content to include documents, and its plain-language fact sheet works through the exact case: a town that posts a PowerPoint after its compliance date does not get the preexisting-documents exception, and the deck has to meet WCAG 2.1 Level AA. The remedy is the one you already know from social — describe the images before you publish — applied to a file type nobody put on the checklist.

The field is already in the app you're using

In PowerPoint, right-click an image and choose View Alt Text, or select it and go to Picture Format > Alt Text. The same pane holds a Mark as decorative checkbox, which is the honest answer for the gradient shape behind your section title. In Google Slides it's Image options > Alt text, or Ctrl + Alt + Y (⌘ + Option + Y on a Mac).

Both apps will sometimes fill that box in for you, and both vendors quietly tell you not to trust it. Google's help page says to verify that automatic alt text is correct. Microsoft's pane makes you tick Approve alt text before generated text counts as yours. Treat what's in the box as a first draft written by something that has never seen your agenda.

The export is where the words disappear

Here's the part that catches careful teams. You can write perfect descriptions on every slide and still ship a PDF with none of them, because alt text only survives the export if the file is tagged. In Office on Windows, that's File > Save As, then Options, then the Document structure tags for accessibility checkbox. Microsoft also recommends running the Accessibility Checker first, because the PDF tags are built from what that check finds.

What's actually inside these files

Decks are unusually image-dense, and the images are unusually load-bearing. A quarterly review is mostly charts and infographics, where the number on the slide is the whole point and a label like "Q3 results" describes nothing. Then there are org charts, screenshots of a dashboard, scanned forms, and a title slide of partner logos.

The counterweight is that a lot of what's on a slide is genuinely ornamental. Section dividers, background textures, the repeated corner mark — those get marked decorative, and choosing to write nothing is a real decision rather than a skipped step.

Where to start if you have hundreds

Don't start at the top of the folder. Start with the files people are actively using: applications, current schedules, benefit forms, anything someone needs in order to do something. The fact sheet is explicit that a document still being used to apply for or access a service doesn't qualify for the preexisting exception even if it was posted years ago — and the clock for larger public entities is the April 2027 Title II date.

After that, work backward through whatever gets downloaded most. It's the same triage as catching up on a year of posts: you're not trying to reach zero, you're trying to stop publishing new files that need fixing later.