Anything Engine Optimization The rolling record of the AI-search industry.
Minimalist photo of a blue sky with a white wall and an antique lantern, ideal for backgrounds.
Photo by Jan van der Wolf on Pexels
Profound study — vendor's own product data, not independent research
Profound research blog · Sep 3 EnginesResearchAEO

ChatGPT shopping mode: feed retrieval overtakes web search

What this answers

Does ChatGPT shopping mode use product feeds or web search?
As of August 2026, ChatGPT's Shopping mode pulls the majority of product recommendations from merchant product feeds rather than web search, per Profound — a reversal from earlier in the year.
What caused ChatGPT's shopping retrieval mix to shift?
Profound ties the shift to the ChatGPT 5.6 release on July 10, 2026, after which feed-integrated retrieval jumped from about 8% to 62% of product recommendations in a single day.

Feed-integrated retrieval accounted for roughly 65% of ChatGPT Shopping mode product recommendations as of late August, overtaking web search as the leading source, per a Profound analysis of 687 customers tracking shopping-mode prompts. The shift was abrupt: on July 10, the day ChatGPT 5.6 shipped, feed retrieval jumped from about 8% to 62% of recommendations. Top-10 merchants’ share of shopping-mode product cards rose from 22.5% to 41.8% over the same period, per Profound.

Why it matters: A single-day retrieval-method flip tied to a model release means product-feed structure, not web-page SEO, is now the dominant lever for surfacing in ChatGPT's shopping results.

The record: ChatGPTOpenAIProfound

Via Profound research blog ↗

Posted to the wire September 3, 2026. Edited by Joe Balewski.