Google AI Mode Is Citing 74% Fewer Sources. Now What?
Google AI Mode went from about 16 cited sources per answer to about 4, roughly 74% fewer, according to one new dataset. Paris Childress covers three stories in today's GEO briefing: fewer AI Mode citations, how ChatGPT picks ads, and Sundar Pichai's agentic layer.
In this episode:
- 74% fewer sources. Maeva Cifuentes's independent analysis of Scrunch tracking data finds AI Mode citing about 4 sources per answer by mid-October, down from about 16 in the summer. Only around 9% of previously cited pages kept their citation, and pages in Google's top three kept theirs far more often. Caveat: this is one dataset, not yet replicated, Google hasn't commented, and one brand made up 42% of the prompt-page pairs.
- How ChatGPT picks ads. OpenAI says an orchestrator model nominates candidate ads, a separate auction ranks and prices them, and a final check can still drop them, all in a system separate from the one that writes answers. The separation is OpenAI's own description and has not been independently audited.
- Ads still shape trust. Even if the systems are separate, an ad shown next to an answer still shapes what users trust and click. Paris's advice: write ad creative for the problem a user works through across several turns, not a single query.
- The agentic layer. Sundar Pichai says agents will reach websites by browsing pages and through protocols like MCP, payment protocols and Google's Universal Commerce Protocol (UCP). Paris's advice: list your site's key actions and check whether an agent could complete them through a feed, an API or a protocol.
The GEO Show is hosted by Paris Childress, founder of Hop AI and co-founder of GEOforge. getgeoforge.com
