PDF ranking at Google works through a conversion step: Google transforms PDFs it can process into HTML before indexing them, which is why text-based PDFs can appear in regular search, AI Overviews (Google’s AI-generated answer panels in search results), and AI Mode on the same basis as ordinary web pages. Google’s Martin Splitt and John Mueller discussed these details on the Search Off The Record podcast, covered by Search Engine Journal on October 14, 2026.
How Does PDF Ranking Work in Google Search?
Google does not use a PDF viewer to read documents. Instead, a dedicated converter component transforms each PDF into HTML right after rendering and right before it enters the index. From that point, Google’s standard web pipeline processes the HTML version, not the original PDF file. Splitt noted Google runs similar converters for other document formats: rich text files, document files, Excel sheets, and Lotus Notes files.
Because the conversion aims to be “as true as is possible to that original format,” content embedded in the PDF, including links, is preserved in the HTML output. Splitt explained: “We convert it into an HTML page and then the rest of our systems are like, ‘I know how to deal with an HTML page,’ and they just do their stuff.”
Do Links Inside PDFs Work Like Normal Links?
Yes. Because links appear in the HTML version Google creates, they are treated like ordinary hyperlinks on a web page. Splitt confirmed: “We can see things like links, for instance, from PDFs.” As Mueller summed it up on the podcast, that conversion is “also why things like links also work in PDF files.”
What Happens to Images Embedded in PDFs?
This is the area where Splitt expressed the most uncertainty. Normally, Google passes an image URL to its image search pipeline, the separate system that processes and indexes images. Images inside a PDF don’t have standalone URLs, so that handoff can’t happen in the usual way.
Splitt’s current understanding: Google likely detects that an image exists inside the PDF but cannot process it like a normal web image. “I think we understand there is an image, but I don’t think we can actually process it like a normal image for image search,” he said, adding that this could change in the future.
Can PDFs Rank in AI Mode and AI Overviews?
Yes, and for the same reason they rank in regular search. The HTML conversion means Google’s AI-powered systems see the document as a web page. Splitt confirmed PDFs “can also appear in AI overviews, AI mode, all of those things.” One condition applies: the PDF must contain selectable text. A PDF that is essentially a scanned image cannot be meaningfully converted to HTML and therefore cannot rank for its content.
Splitt’s practical test: try highlighting all text in a PDF and pasting it into a plain text editor. If the text transfers cleanly, Google can work with the document.
Are There File Size Limits for PDFs?
Splitt said he believes Google applies a larger file-size limit when fetching PDFs than it does for standard HTML pages, comparable to how it handles video and image files. Once the PDF is converted to HTML, standard HTML processing limits likely apply. He acknowledged he does not know the exact PDF limit, and noted Google may revise limits if publishers start uploading significantly larger files.
Should You Publish Content as a PDF?
Splitt and Mueller converged on a pragmatic position: HTML is the preferred format, but a PDF is far better than no page at all.
- Use a PDF when the format genuinely matters: specific print layouts, fillable forms, or documents that must look exact in print, such as menus, registration forms, or product spec sheets.
- Prefer HTML when you have the choice: Mueller noted that an HTML version of the same content “makes everything a little bit easier.”
- Check text selectability first: if you cannot highlight and copy text out of your PDF, Google cannot index it for ranking.
Mueller framed the choice plainly: “If the option is have a menu as a PDF or don’t have a menu at all, then yes, I would prefer to have a menu. But if the option is menu as a PDF versus a normal web page, I kind of like normal web pages because it’s easier to interact with them.”
What This Means for AI-Search Visibility
Google’s pipeline treats your PDF much like a web page from the moment of conversion. So the real question isn’t can a PDF rank? It’s what does the HTML version of my PDF actually look like to Google and to AI retrieval systems?
Consider a concrete example. A company publishes a product catalog as a PDF. If the catalog is text-based and well-structured, Google’s conversion produces readable HTML with product names, descriptions, and working links. That content can surface in AI Mode the same way a well-written web page would. But if the catalog is a graphic-design export saved as a flat image layout, Google’s conversion produces little or nothing usable.
Two implications stand out from an AI-search visibility perspective.
First, images in PDFs may be a blind spot. Based on Splitt’s current understanding, Google knows an image is there but may not process it like a normal web image, so if your PDF relies on images to convey key information, such as data charts, tables rendered as graphics, or labeled diagrams, that information may not be used the way text is. Moving that information into the text layer of the PDF, or into an HTML page, is the practical fix.
Second, and this is Hingewise’s assessment rather than something the podcast states explicitly: even when a PDF converts cleanly to HTML, that HTML is structurally flat compared to a well-built web page. A good HTML page carries semantic markup (HTML elements that signal meaning to search engines and AI systems, like heading tags and structured content sections) that helps AI retrieval systems identify what a passage is about and anchor citations to specific claims. A PDF-converted HTML document may carry less of that hierarchy, depending on how the PDF was built. In Hingewise’s view, that gap could make a converted PDF a weaker source for AI citations than a well-structured page, even when Google can read it.
For content you want cited in AI-generated answers, a well-structured HTML page is the stronger choice. PDFs serve a clear purpose for format-dependent documents, and Google’s explanation clears up much of how PDFs are indexed. Mueller and Splitt’s candid acknowledgment that some details, particularly around image handling and size limits, remain uncertain is a fair signal that the pipeline continues to evolve.
Before publishing a PDF for search or AI visibility, check:
- Can you highlight and copy all key text from the PDF into a plain text editor? If not, Google may not be able to use the text for search.
- Does the PDF contain images that carry critical information, such as data tables as graphics or labeled diagrams? Move that information into the text layer if possible.
- Does equivalent content already exist as an HTML page on your site? Decide which version you want ranking and whether the PDF adds distinct value.
- Are there links embedded in the PDF that you want flowing through to Google’s link graph? Confirm those links are in the text layer, not only in image-based elements.
- Is the PDF very large? Splitt believes limits apply both when Google fetches the file and after conversion, but he does not know the exact thresholds.
Source: Search Engine Journal, October 14, 2026, reporting on the Search Off The Record podcast with Martin Splitt and John Mueller.
Lam Nguyen · Hingewise
This article was drafted with AI assistance from the sources cited above and checked against them before publishing. The "What this means for AI-search visibility" section is Hingewise's editorial analysis, built on the frameworks and positions we use in our GEO work. How we work: Editorial policy.