Yes, PDFs can rank in Google. On a Search Off the Record episode covered on 13 October 2026, Martin Splitt said Google does not use a PDF viewer: a converter turns the PDF into HTML after rendering, and that HTML is what gets indexed and ranked "pretty much the same as any other document". John Mueller added that PDFs can appear in AI Overviews and AI Mode, but that HTML is generally easier and better for SEO. Scanned PDFs without a text layer are the main exception.
Key Takeaways
- Google converts PDFs to HTML during indexing, so text and links in a PDF are processed like web-page text and links (per Splitt and Mueller).
- Ranking uses the same core systems; PDFs can show in AI Overviews and AI Mode, according to Mueller.
- If you cannot select and copy text from the PDF, assume Google cannot read it either.
- Google's documentation says Googlebot crawls the first 64MB of a PDF and the first 2MB of HTML, though the hosts on the podcast could not name the PDF number.
- Google recrawls PDFs rarely. For an update, publish under a new filename or URL.
- Default to HTML; use a PDF when you need print fidelity, forms or offline use.
What was actually said, and where
The source is a Search Off the Record episode titled "PDFs in Search: Indexing, AI & Pitfalls", with Splitt and Mueller. Search Engine Roundtable logged it on 13 October 2026 with a chapter list and a link to the audio. I did not listen to the full episode; I read three written summaries and Google's documentation. The summaries are Search Engine Roundtable, Search Engine Journal and PPC Land. Barry Schwartz notes most of this was already covered in earlier Google statements, and that Google said little about GEO specifically. That is the right calibration: this is confirmation and detail, not news that changes the rules.
How the conversion works
According to the Search Engine Journal summary:
- Googlebot fetches the PDF.
- During indexing, after rendering, a converter turns the PDF into HTML. Splitt: "we are not using a PDF viewer."
- From there the pipeline treats the document as HTML. Ranking signals are evaluated on the converted version.
PPC Land adds that Google uses similar converters for other document formats such as rich text and spreadsheets. Google's indexable file types page lists PDF among encoded files that need a dedicated parser, and says file type is determined from the Content-Type header, so serve application/pdf correctly.
Text, links and images
Text
Selectable text is indexed. Rasterized pages, where each page is a flat image, have no text layer. Splitt and Mueller said such PDFs may not rank for their words. The test is simple: open the file, select all, paste into a notes app. If nothing useful appears, neither will it for Google. Rasterized pages are also harder for screen readers, per PPC Land.
Links
Links inside a PDF work like links on a web page because the converted HTML contains them. Mueller said this is "why things like links also work in PDF files". For you, that means a PDF can pass users and, plausibly, signals to your own pages, but treat that as a bonus rather than a link-building plan.
Images
Splitt was unsure whether embedded images are indexed and leaned toward no, saying it could change. Google's image pipeline normally needs a URL and embedded images do not have one. If an image matters for Image Search, publish it on an HTML page.
Ranking and AI features
Splitt said indexing and ranking work much like any other document. Mueller said PDFs can appear in AI Overviews and AI Mode, and that PDFs "work" for GEO at Google. That is consistent with Google's AI features guidance, which says the requirement is that a page be indexed and eligible to appear in Search with a snippet, with no additional requirements for AI Overviews or AI Mode. If you want a refresher on the product, see my AI Mode explainer.
What was not said: nobody quantified how often PDFs rank against equivalent HTML pages, and neither speaker explained how publishers could separate search visibility from AI training, per PPC Land. Do not read "PDFs can rank" as "PDFs rank as well".
Size limits, and a source conflict
The hosts could not name the PDF size limit on the episode. Google's Googlebot documentation says Googlebot crawls the first 2MB of a supported file type and the first 64MB of a PDF, applied to uncompressed data. PPC Land cites the same 64MB figure. Search Engine Roundtable's summary instead links to a 15MB Googlebot limit "for context". Those do not match, and I'd go with the documentation page I opened for the 2MB and 64MB figures, while noting that Google deliberately avoids over-documenting limits because they change.
Practical read: a 64MB PDF is not a goal. Splitt suggested the HTML limits likely apply to the converted version, which he did not confirm. Keep PDFs small enough to download on a phone anyway.
The pitfalls Google listed
| Pitfall | What Google's hosts said | What to do |
|---|---|---|
| Rasterized or scanned pages | No text layer, may not rank for text | Export from the source with real text; run the copy-paste test |
| Mobile experience | Whole file downloads before display; often does not reflow | Offer an HTML page as the main path |
| Duplicate PDFs | Re-hosted copies are byte-identical, hard to canonicalize | Use a rel="canonical" HTTP header, fill metadata, put a visible link to your homepage in the PDF |
| Password protection | Bots cannot read it | Do not password-protect what you want found |
| Robots.txt blocks | Crawling is blocked | Allow crawling of PDFs you want indexed |
| Rare recrawling | Google assumes PDFs seldom change | New filename or URL for updates |
| CMS-generated PDF plus HTML | Out-of-sync indexed copies | Pick one canonical version; keep the other noindexed |
The rel="canonical" HTTP header, noted as rarely used by publishers, is a Link header. For blocking indexing of PDFs, Google's robots meta tag documentation says to use an X-Robots-Tag header, since a meta tag cannot go inside a PDF, and warns that a robots.txt block prevents crawlers from ever seeing that header.
Freshness: the hidden risk
Mueller said Google refetches PDFs infrequently, assuming they will not change for a long time. Mueller's suggestion for updates was a new filename, because a new linked URL is crawled as new content. Splitt suggested caching headers that match how often the file changes, such as months for a quarterly catalogue. If you publish "Course-Brochure-2026.pdf" and overwrite it every quarter, search may show the old text for a while.
That interacts with a content-decay problem I covered in my content refresh and AI citation half-life post.
When a PDF is the right call
Both speakers said PDFs are fine when content needs fixed formatting, print fidelity or forms. Splitt's example: a restaurant PDF menu is "better than not having your menu on the website at all", while Mueller would choose a normal web page for interactivity, zoom, reflow and speed. They also advised against publishing llms.txt as a PDF, per PPC Land; keep machine-readable files as plain text.
A simple decision rule
- Needs to be read on screen, searched, updated or linked to: HTML.
- Needs to be printed exactly, signed or filled offline: PDF, plus an HTML landing page that summarizes it and links to it.
- Both needed: make the HTML canonical for the topic, offer the PDF as a download, and keep the two in sync.
I cover the production steps in my PDF SEO checklist for whitepapers, brochures and prospectuses.
What I would audit this week
- Search
site:yourdomain.com filetype:pdfand note every PDF Google has indexed. Google's indexable file types page mentions thefiletype:operator for this. - For each, run the copy-paste text test.
- Check whether the same information exists on an HTML page, and whether they conflict.
- Check last-modified dates against what the PDF says. Look for pricing or fee tables that are out of date.
- Look at your analytics for PDF downloads versus the HTML equivalents.
- Decide, file by file, whether to keep, replace, or noindex.
FAQ
Do PDFs rank in Google?
Yes. Per Martin Splitt on Search Off the Record, Google converts PDFs to HTML and indexes and ranks that version much like any other document. PDFs can also appear in AI Overviews and AI Mode, according to John Mueller.
Does Google read text inside images in a PDF?
Not reliably. Scanned or rasterized PDFs have no text layer and may not rank for their content. Test by selecting and copying text from the file; if you cannot, treat it as unreadable.
Do links in PDFs count?
They are crawled, because they survive conversion to HTML. Mueller said links work in PDF files for that reason. Do not expect them to substitute for links on a regular page.
What is Google's PDF size limit?
Google's Googlebot documentation says it crawls the first 64MB of a PDF and the first 2MB of other supported file types. The podcast hosts could not name the PDF limit, and one trade summary cites a different 15MB figure, so rely on Google's page and keep files small.
Is a PDF or an HTML page better for SEO?
Mueller said HTML is generally easier to use and better for SEO. PDFs are not impossible, just harder. Use PDFs for print fidelity, forms and offline needs.
How do I stop a PDF from being indexed?
Use an X-Robots-Tag: noindex HTTP header on the file, per Google's robots meta tag documentation. Do not block the file in robots.txt, because then Google cannot see the header.
How do I update a PDF so Google notices?
Mueller suggested publishing the update under a new filename, since a new linked URL is crawled as new content. Overwriting the same URL may be recrawled slowly.
Can the same PDF on two domains cause problems?
Identical PDFs are byte-for-byte duplicates, which makes canonicalization hard. Mitigations discussed were a rel=canonical HTTP header, filled-in metadata and a visible link back to your home page inside the PDF.
Should my llms.txt be a PDF?
No. Splitt and Mueller advised against it, as reported by PPC Land. Keep such files as plain text.
Want a content audit that includes your PDFs?
I'm Younus Fardeen, an organic growth, SEO and AEO strategist with 4+ years of marketing experience across edtech and startups. If your site hides important fees, syllabi or whitepapers inside PDFs, I can help you decide what to move, what to keep and how to measure it. See my work and reach me through the contact form at younusfardeen.in. Details above are as of 14 October 2026 and drawn from the sources linked.