Martin Splitt and John Mueller of Google released a Search Off the Record episode on how Search treats PDFs, titled “PDFs in Search: Indexing, AI & Pitfalls.” Search Engine Roundtable’s Barry Schwartz posted the chapter list on October 13.
The episode opens with whether an llms.txt file can be saved as a PDF. It then contrasts PDFs, which are binary containers, with HTML, and walks through Google’s converter pipeline that turns PDFs into HTML for indexing. Further segments address pulling out images and links, crawl limits and file sizes, and when publishing a PDF makes sense.
The back half asks whether PDFs rank in classic results, AI Overviews, and AI Mode. It also covers scanned pages, accessibility, mobile delays, duplicate files and canonicalization, metadata, passwords, updates, CMS-generated PDFs, and restaurant menus. The final chapter treats PDFs as a tool for GEO (generative engine optimization) and AI grounding.
Schwartz notes that Google had addressed most of this earlier and said little on the GEO and AI side. Teams with large PDF libraries can use the chapter timestamps to jump to the segment matching their problem.
Search Engine Roundtable, Barry Schwartz, October 13, 2026.
