What Did Google Say About PDFs?
On October 13, 2026, Google's Search Central podcast, Search Off the Record, published episode 115, “PDFs in Search: Indexing, AI & Pitfalls,” with Martin Splitt and John Mueller of Google's Search Relations team (episode page). Google publishes a transcript, and the quotes below come from it (transcript).
Splitt describes a component called a converter, which “happens right after rendering and right before something potentially makes it into the index.” Google has converters for several formats, and for PDFs it is “producing an HTML version of it so that the processing in the rest of the pipeline pretty much only works with HTML.” Asked whether PDFs therefore rank normally, he says: “Yeah, indexing and ranking is pretty much the same as any other document.” When Mueller adds that they can “also appear in AI overviews, AI mode, all of those things,” Splitt answers: “Because it is basically internally a normal HTML page, nothing wild.”
The episode page sums up the advice in five words: “HTML is preferred, and PDFs work.” Late in the episode Mueller says that for AI features, “at least at Google, yes, they do” work, and Splitt adds that “good SEO is good GEO.”
How Does Google Handle a PDF?
| Question | What Google's hosts said | Source |
|---|---|---|
| Is a PDF indexed as a PDF? | A converter produces an HTML version right after rendering | Transcript (Splitt) |
| Does it rank like a web page? | “indexing and ranking is pretty much the same as any other document” | Transcript (Splitt) |
| Can it appear in AI Overviews and AI Mode? | Yes, because internally it is “a normal HTML page” | Transcript (Mueller, Splitt) |
| Are links in a PDF followed? | Links are kept in the HTML version, so “we can see things like links” | Transcript (Splitt) |
| Are images inside a PDF indexed for image search? | Splitt does not think they are processed like normal images, but says that “might change or might have changed” | Transcript (Splitt) |
| Which is better, PDF or HTML? | “HTML is preferred, and PDFs work” | Episode page |
What Are the File-Size Limits?
Google's crawler documentation answers this directly. “Googlebot crawls the first 2MB of a supported file type, and the first 64MB of a PDF file,” and “The file size limit is applied on the uncompressed data.” At the limit, Googlebot “stops the fetch and only sends the already downloaded part of the file for indexing consideration” (Google Search Central).
MB (uncompressed, first N MB fetched)
The limit is on the first N MB of a file; Googlebot stops fetching at the limit and passes on only what it downloaded. Each referenced CSS or JavaScript resource is fetched separately under the 2MB limit. Googlebot Image and Googlebot Video may have different limits. The episode page states no thresholds.
The chart shows that PDFs get a much larger fetch limit than ordinary pages, but the cap still exists, so a very long annual report or media book may be only partly read.
On the podcast, the hosts did not cite that figure. Splitt said, “I don't know the specific limit for PDF files,” guessed that “probably there's a limit for the web page kind of representation,” and said Google is “not super keen on documenting” limits because they may change. Google's Googlebot page, last updated February 3, 2026, does document the 64MB figure, so we rely on it. If a PDF is very large, our recommendation, not a Google rule, is to split it or publish the key text as an HTML page.
When Can a PDF Hurt You?
Splitt calls rasterized pages “the worst one”: a PDF whose pages are effectively images, so “you lose all text content.” His test is to open the PDF, select all text and paste it into a document or text file. “If it's not there, then that is a problem. Then we can't see the content inside the PDF.” He adds that such pages are also hard for screen-reader users unless the PDF was made accessible. Mueller notes that scanned pages are the same case, and that optical character recognition is something “you kind of can't rely on.”
The hosts raise four more issues (transcript):
- Duplicates. Mueller says a copied PDF is “one-to-one identical,” which makes canonicalization harder. He mentions a
rel="canonical"link header in the HTTP response, and linking to your own homepage from the top of the PDF so the link survives conversion. - Metadata and protection. Splitt suggests filling in author and description metadata, and notes that a password-protected PDF cannot be read by search engines.
- Updates. Mueller says Google does not refetch PDFs often, so when a PDF changes “it can make sense to just use a new file name.”
- Mobile. Splitt says PDFs must download in full before showing, and “some PDFs are really weird when you try to zoom in on them—they don't reflow.”
On restaurant menus, Mueller is pragmatic: a PDF menu beats no menu, but he likes normal web pages because “You can zoom better, it flows on mobile, loads a little bit faster.”
Should a Press Kit Be a PDF?
Google does not address press kits, so this section is our recommendation, not a Google rule, built on what the hosts said. A press kit is often a PDF because journalists download and forward it. Following Google's stated preference for HTML, publish the substance as a web page (company facts, bio, boilerplate, contact details, links to coverage) and offer a PDF as a download of that page. Then:
- The page is easy to read on a phone and easy for a search engine or an AI system to quote.
- The PDF is a convenience copy, with real text and not images of text.
- Both carry the same facts, so a changed phone number or title is fixed in one place and re-exported. Given Mueller's point that PDFs are refetched rarely, give an updated PDF a new file name and link to it from the page.
- Put a link to your homepage near the top of the PDF, as Mueller suggests, so a forwarded copy still points back to you.
Our guides to what a press kit is, press kit versus media kit and how to create an online newsroom cover the contents. For checking whether Google has the page, see how to check if a page is indexed by Google.
What This Means for Anyone Buying a Placement
If a vendor sends a PDF as proof of coverage or as a deliverable, ask for the live page URL. Coverage that exists only as a PDF clipping is not a link, and nothing in Google's material says a PDF passes any benefit to your site. Do not read the PDF news as a reason to host your coverage as flattened scans. Splitt's own test applies: if the text will not copy out, Google cannot see it. Nothing here promises that any page, PDF or HTML, will rank or be cited.
Methodology and Limitations
Four limits apply. First, we quote the hosts from Google's published transcript, not the audio; hosts' remarks on a podcast are not formal documentation, and Splitt flagged some answers as his “current understanding.” Second, where the podcast and the documentation differ on file-size limits, we follow the documentation. Third, Google's file-size page was last updated February 3, 2026, and may change. Fourth, we did not test how Google indexes any PDF, and the press-kit advice is untested house recommendation.
Conclusion
A PDF can rank and can appear in AI Overviews and AI Mode when it contains real, indexable text. Google's stated preference is HTML wherever you have the choice. For a press kit, that means a web page first and a text-based PDF second.
Frequently Asked Questions
Can a PDF rank in Google Search?
Yes. On Google's Search Off the Record episode of October 13, 2026, Martin Splitt said PDFs are converted to HTML and that “indexing and ranking is pretty much the same as any other document.”
Can PDFs appear in AI Overviews and AI Mode?
Yes, according to Google's Martin Splitt and John Mueller on the October 13, 2026 episode, because a converted PDF is internally treated as a normal HTML page.
Does Google prefer PDF or HTML?
The episode page says "HTML is preferred, and PDFs work."
How big a PDF will Googlebot read?
Google's crawler documentation says the first 64MB of a PDF, against the first 2MB of other supported file types, on uncompressed data.
How do I know whether my PDF is just images?
Google's Martin Splitt suggests selecting all text and pasting it into a text file. If nothing comes through, the pages are likely images and Google cannot see the content.
Should I put my press kit online as a PDF only?
Our recommendation, not a Google rule, is to publish the content as an HTML page and offer a text-based PDF download as a copy.
Sources
This guide is built from Google's episode page, its published transcript and Google's Googlebot documentation, all read on October 14, 2026. The press-kit and buyer sections are house recommendation. We did not test PDF indexing.
- official Google Search Central, Search Off the Record, PDFs in Search: Indexing, AI & Pitfalls. Supports date, hosts and the "HTML is preferred, and PDFs work" summary. Read October 14, 2026.
- official Google Search Central, Search Off the Record, episode 115 - Transcript. Supports all quotes from Martin Splitt and John Mueller. Read October 14, 2026.
- platform documentation Google Search Central, Googlebot. Supports the 64MB and 2MB limits and cutoff behavior. Last updated February 3, 2026. Read October 14, 2026.