A PDF brochure looks like content, but it does not act like a page.
I check PDFs the way I check a website: pull the file, then look for an actual text layer sitting under the design, not just words that look right on screen.
A scanned PDF with no text layer carries no words to anything that reads it afterward. Not to Google. Not to an AI crawler. Not to a person trying to copy a phone number out of it. That part is not a judgment call. It is how a scanned image works.
The more common problem is not the scan. It is the PDF built the right way, with real text, that still does almost nothing for how a business gets found.
What a Text Layer Actually Is
A text layer is the part of a PDF that holds actual letters, the kind a person can highlight, search, and copy. A scanned page is a photograph of text, nothing more. It looks identical on screen. Underneath, one is words and the other is a picture of words.
Open the file and try to select a sentence. If nothing highlights, there is no text layer. That single test tells you more than looking at the page ever will.
Fixing it is OCR, optical character recognition, not a redesign. Run it once, and a file that was closed to every crawler opens back up.
What I Check Inside a PDF
- Whether the text is selectable at all, not just visible
- Whether the file has a real title and basic metadata instead of a scan name
- Whether the same information already lives on the site as a real page
Usually it does not. The PDF is the only copy of that information anywhere, which makes it an island.
Most of these files started as something else. A print brochure, scanned once and never revisited. A rate sheet exported flat from a design program because flat looked cleaner. Nobody set out to hide the information. It happened as a side effect of how the file was made.
Why an Island Is the Real Problem
A page earns trust from what sits around it. Internal links pointing to it. Structured data describing what it is. A URL a person can land on, bookmark, and find again without opening a download.
A PDF usually has none of that. It sits outside the site's internal linking, so our post on internal linking does not reach it. It carries no schema most of the time. It opens into a download folder instead of a page someone remembers.
I have not seen documentation from any AI engine stating exactly how well it reads a PDF's text, so I will not claim it cannot. What I can say is that even a perfectly parsed PDF loses everything that made the page around it worth citing.
Where a PDF Still Makes Sense
A PDF is right when someone needs to print it, sign it, or carry a static copy with no connection. A rate card taken into a meeting. A one-page spec sheet attached to an email.
It is the wrong format for anything a person or a machine should find through a search box: services, pricing ranges, hours, location, a plain list of what you actually do. Put that information on a page first. The PDF, if it exists at all, becomes a convenience copy, not the source.
The Fix I Actually Recommend
Keep the PDF if it genuinely needs to be printed or carried. Rebuild the page underneath it in real HTML, carrying the same facts a person would read in the download. Link the two together, page to file, so each one points at the other.
Give the file a real name and a real title. Run OCR on anything scanned, even the old ones already sitting online. It costs little, and it turns a locked file back into text.
None of this needs new photography or a new brand system. It needs someone to open files nobody has opened since the day they were uploaded.
A site with a handful of PDFs linked from old pages is not a handful of small jobs. It is one afternoon of checking, sorted into keep, rebuild, or delete. Treat it the same way you would treat a stale page: on a schedule, not once and forgotten.
Checking every PDF for a text layer, a title, and a page to stand next to is part of how I run an SEO / AEO / GEO Foundation engagement. The file was never the problem. The silence around it was.