The landscape of digital discovery is undergoing a tectonic shift. For years, search engine optimization (SEO) was the undisputed king of e-commerce visibility. However, recent data suggests that ChatGPT is quietly pivoting away from the open web, favoring structured data feeds over traditional search results. A comprehensive new report from Profound indicates that a fundamental change in how the AI retrieves product information—implemented around July 10—has sent shockwaves through merchant visibility, signaling a new era for brands operating within the OpenAI ecosystem.
Main Facts: The Transition to Feed-Integrated Retrieval
For merchants, the "how" of AI product discovery has suddenly become as important as the "what." According to Profound’s tracking of over 1.7 million prompt runs, the method by which ChatGPT selects products for users has shifted dramatically. On July 10, the share of ChatGPT Shopping recommendations classified as "feed-integrated" surged from 8.26% to 61.54%.
This shift marks a departure from the reliance on broad web crawling. Instead, ChatGPT is increasingly leaning on the "Agentic Commerce Protocol," a framework that allows retailers to pipe their product data—prices, descriptions, and availability—directly into the AI’s decision-making engine. This change, while subtle to the average consumer, represents a massive realignment of digital retail power. Brands that have successfully integrated their product feeds are experiencing a surge in visibility, while those still relying on traditional web-search indexing are finding their products increasingly relegated to the digital shadows.
Chronology: The July 10 Inflection Point
The transformation did not happen gradually; it arrived with the suddenness of a software update. On July 9, OpenAI released GPT-5.6, stating that a global rollout would occur within 24 hours. While OpenAI’s official release notes remained characteristically quiet regarding e-commerce functionality, the data collected by Profound tells a different story.
The Immediate Aftermath
Between July 7 and July 9, the status quo held steady. However, the period between July 10 and July 12 revealed a massive volatility in merchant visibility. In a sample of 687 customers, 450 experienced at least a 33% reduction in shopping visibility immediately following the update. Conversely, a smaller group of 67 merchants saw an increase of similar magnitude.
The August Consolidation
The trend continued to solidify throughout the summer. By analyzing a broader dataset of 97,725 prompt runs between July 1 and August 24, Profound observed that feed-based retrieval officially overtook web-search retrieval in August. By September 3, the report confirms that approximately 65% of all tracked product recommendations were sourced via feed retrieval, solidifying the new architecture as the primary engine for AI-driven shopping.
Supporting Data: Understanding the Visibility Swing
The statistical correlation between the adoption of feed-integrated retrieval and the loss of traditional web-search visibility is striking. Profound’s analysis of 517 merchants—all of whom experienced significant shifts in their traffic—found that a model tracking the decline of web-search influence versus the gain of feed-integrated influence could explain 83% of the observed variation in visibility.
A Narrowing Pool of Winners
The shift toward feed-integrated retrieval has had a contractionary effect on the diversity of merchants featured. During the transition, the top 10 merchants saw their share of total recommendations jump from 22.5% to 41.8%. Simultaneously, the total number of unique merchants referenced by the AI dropped by over 20%, falling from 13,524 to 10,607.
This suggests that when ChatGPT switches to feed-based retrieval, it defaults to a more restricted, curated pool of data rather than the "infinite shelf" of the open web. Furthermore, the data suggests that Shopify-hosted stores are major beneficiaries of this change; an estimated 35% of all feed retrieval activity in July appears directly linked to the Shopify ecosystem, likely due to the seamless integration provided by the Shopify Catalog.
Official Responses and Platform Mechanics
OpenAI has maintained a position of neutrality regarding how its AI selects products. Its Help Center explicitly states that product results are selected independently by ChatGPT, emphasizing that these selections are not ads and are not influenced by paid partnerships.
The Agentic Commerce Protocol
The mechanism behind this retrieval is the Agentic Commerce Protocol, an initiative extended to product discovery on March 24. According to OpenAI’s documentation, the system pulls data from a variety of sources:
- Direct Integrations: Shopify and Etsy catalogs are already natively connected, requiring no additional technical hurdles for merchants on those platforms.
- Direct Feeds: Retailers not on these platforms can apply for direct feed access, though the company’s merchant page indicates that this process is currently waitlisted.
- Data Providers: OpenAI supports feeds from major providers, including Salesforce and Stripe, to populate its database.
Despite these clear guidelines, OpenAI has yet to provide public data on the specific weighting of feed-sourced data versus open-web data. The lack of transparency regarding the "ranking algorithm" behind these selections remains a point of contention for merchants who are currently flying blind in their optimization efforts.
Implications for the Future of Retail
The implications of this shift are profound for any business selling goods online. The era of "Search Engine Optimization" is being superseded by "AI Feed Optimization."
The Barrier to Entry
The current "waitlist" status for direct feed access creates a significant competitive moat. Brands that are already integrated—particularly those on Shopify or Etsy—are effectively grandfathered into a position of high visibility. For others, the inability to easily connect a feed creates a disadvantage that cannot be overcome by traditional website improvements.
The New Competitive Frontier
Moving forward, mere connectivity will likely not be enough. Profound anticipates that as more merchants gain access to feed-integrated systems, the competition will shift toward the "quality" of the feed. This includes the depth of metadata, the granularity of product descriptions, and the precision of the fields provided to the AI. Brands will need to optimize their data for machine consumption rather than human reading, ensuring that their feeds contain the exact variables the AI uses to match user intent with product reality.
The "Black Box" Problem
It is essential to note that these findings are based on observational data from a single vendor’s tracked prompts. They do not necessarily reflect the totality of all shopper sessions across the globe. Furthermore, the rapid release cycle of OpenAI’s models—such as the recent rollout of GPT-6 Astra—means that the rules of the game are subject to change without notice.
Looking Ahead: A Self-Serve Future
As we look toward the remainder of the year, the landscape is poised for further evolution. OpenAI has announced plans for a self-serve platform that will allow stores to connect their feeds without the current bureaucratic friction of the waitlist.
While this promises to democratize access, it also threatens to flood the feed-retrieval system with new data, potentially diluting the visibility gains seen by early adopters. For the modern merchant, the message is clear: the path to the consumer no longer goes through the browser’s address bar. It goes through the API. Businesses that fail to prioritize their AI-ready data infrastructure will find themselves increasingly invisible in an age where the most important search engine in the world is no longer a search engine at all—it is a conversational agent.
