Google explains how PDFs get indexed and ranked, including in AI Mode
Google converts PDFs into HTML before indexing them, so a PDF is ranked on the HTML version rather than the original file. Martin Splitt, speaking on a Google podcast, said the converter runs during indexing, right after rendering and just before a document can enter the index, and that the same component handles rich text files, spreadsheets and other document formats.
Because the conversion produces an HTML page, links inside a PDF are followed like links on a normal web page, and PDFs can surface in AI Mode and AI Overviews on the same basis as regular search results. Splitt said the file size limit for fetching a PDF is probably larger than the web page limit, similar to video, though he did not give a figure, and that HTML limits likely apply once the PDF has been converted.
Images inside PDFs are a weaker point. Splitt said he believes Google can tell an image is present but cannot process it the way it processes images with URLs, so it may not reach image search, and that this could change. PDFs made of scanned images rather than selectable text cannot be converted and will not rank on their text.
John Mueller said HTML remains the better option where a choice exists, because it is easier to interact with, but that a PDF is preferable to no content at all. For brands chasing citations in AI answers, the practical test is whether text in a PDF can be selected and copied: if it can, Google can convert and index it.
Reported by Search Engine Journal ↗
Topics: AI Mode, Google Search, AI Overviews