Josh Blyskal, Special Projects at Profound, returns to the show for a data-rich, in-person conversation with Ross Hudgens about what’s actually true in AI search right now, and what’s just noise. 

Josh makes the case that the industry’s obsession with tracking frequency is mostly a distraction (running a prompt 10x a day vs. once barely moves your visibility score) and that the real hard problem is still attribution. He walks through a new model some advanced teams are using, treating agent attention as a proxy for human attention and heavily discounting citations into “clicks,” and shares why ChatGPT referral traffic is up 60% since mid-March, especially for CPG.

From there they get into why GEO is fundamentally a relationship business (the Reddit–Google negotiations, why Claude cites Reddit almost never, and the walled-garden problem), what actually separates slop from signal, and why listicles still do the “caloric work” of teaching a model an industry.

Show Notes

0:00 How often should you track Prompts?
08:14 The new attribution model
14:05 GEO as a relationship business
20:05 What separates slop from signal
26:43 Product pages, versus queries and battle cards
30:21 ChatGPT vs. Claude
34:12 Future-proofing your stack
40:55 Video, social and the language arbitrage
49:18 Imitation vs. creation & where GEO is headed

Show Links

 

The definitive guide to AI SEO.

Ross Hudgens’ new book from Wiley breaks down how to rank, get cited, and win in AI search.

Secret recipes sold here.

Fresh out of the oven.