NextFin News - Amazon’s high-stakes experiment with conversational advertising on OpenAI’s ChatGPT is delivering a wealth of consumer intent data but failing to trigger the "buy" button, according to internal pilot results and advertiser feedback. Six weeks after OpenAI launched its U.S. advertising pilot with a $60 CPM (cost per thousand impressions), the partnership between the e-commerce giant and the AI powerhouse is revealing a fundamental friction in "agentic commerce": users are happy to chat with AI about products, but they are not yet ready to let the chatbot close the deal.
The pilot, which began in February 2026, has seen OpenAI’s ad revenue hit an annualized run rate of $100 million. However, for Amazon and its third-party merchants, the return on investment remains lopsided. While the ads—often appearing as contextual product recommendations within a ChatGPT dialogue—generate high engagement rates, the conversion to actual sales on Amazon’s platform has been described by several participating brands as "negligible." The disconnect stems partly from a technical retreat; on March 4, 2026, OpenAI quietly removed the "Instant Checkout" feature that had previously allowed for seamless transactions within the chat interface, forcing users back to traditional browser-based checkouts.
Catherine Perloff of The Information, who has closely tracked the intersection of generative AI and digital marketing, reports that the primary value for Amazon currently lies in "top-of-funnel" intelligence. By analyzing how users describe their needs to ChatGPT—often in far more detail than a standard search query—Amazon is gaining unprecedented insight into consumer pain points. Yet, this data comes at a steep price. The $60 CPM is significantly higher than traditional display ads, and without the "Instant Checkout" functionality, the friction of jumping from a chat app to the Amazon app is killing the impulse to buy.
The struggle to monetize chat-based search highlights a growing divide in the advertising industry. Criteo, which became the first ad-tech partner to integrate with OpenAI’s pilot in early March, argues that the contextual nature of these ads is superior to traditional search because the user is in an "active problem-solving mode." However, this optimism is not yet reflected in the balance sheets of retail brands. According to data from ALM Corp, while retail and grocery brands claim 44% of ChatGPT’s current ad inventory, many are treating the spend as an experimental R&D cost rather than a performance marketing channel.
For U.S. President Trump’s administration, which has emphasized American leadership in AI as a cornerstone of economic policy, the commercial viability of these platforms is a matter of national competitiveness. However, the current "data-rich, sales-poor" reality suggests that the transition from search engines to "answer engines" will be more profitable for the platform owners than for the merchants. Amazon’s willingness to continue the pilot despite low sales indicates a defensive strategy: it cannot afford to let OpenAI or a resurgent Google define the future of shopping without being at the table.
The pilot is now set to expand into international markets, including Canada and Australia, despite the lingering doubts over conversion. For now, the "power couple" of Amazon and OpenAI remains in a state of awkward courtship. They have successfully captured the consumer's attention and their data, but the final step of the transaction—the actual exchange of currency for goods—remains stubbornly tethered to the old world of clicks and carts rather than conversations.
Explore more exclusive insights at nextfin.ai.
