Walmart is significantly recalibrating its approach to artificial intelligence-powered shopping, pulling back from OpenAI’s Instant Checkout feature within ChatGPT after internal data revealed substantially lower conversion rates compared to purchases made directly on its own website. The move, first reported by Wired, underscores the challenges of integrating complex e-commerce transactions within conversational AI platforms and signals a broader industry reassessment of handing over the crucial “last mile” of sales to third-party AI providers.
The initial partnership, launched in October, aimed to allow ChatGPT users to purchase roughly 200,000 Walmart items without leaving the chat interface. This represented an early foray into what’s known as “agentic commerce,” where AI assistants proactively facilitate purchases on behalf of users. Still, the experiment didn’t deliver the anticipated results, prompting Walmart to shift its strategy toward a more controlled integration of its own AI capabilities.
According to Daniel Danker, Walmart’s executive vice president of AI acceleration, product and design, conversion rates within ChatGPT were only about one-third of those seen on Walmart’s native platforms. He described the experience as “unsatisfying” in his conversation with Wired. This performance gap wasn’t simply a matter of initial teething problems; CNBC reported that the feature suffered from a limited product selection, occasionally inaccurate item information and design limitations that hindered more complex shopping scenarios.
The Friction of a Third-Party Checkout
The core issue, it appears, was friction. Walmart deliberately excluded high-consideration purchases, like televisions, from the Instant Checkout feature because the chat-based interface couldn’t adequately support the bundling of products and the presentation of relevant accessories. The linear, single-query format of ChatGPT as well didn’t align with the typical multi-item shopping baskets that drive revenue for Walmart. Customers accustomed to Walmart’s established ecosystem were also hesitant to utilize an unfamiliar payment flow within a third-party interface.
Industry analysts suggest OpenAI may have underestimated the complexities of enabling secure and seamless transactions. Gartner analyst Bob Hetu told CNBC that OpenAI “underestimated how demanding the enablement of transactions was going to be.” The challenge isn’t just technical; it’s about building trust and replicating the familiar shopping experience consumers expect.
Sparky Takes the Reins
Walmart is now pivoting to a model that keeps control of the shopping experience firmly within its own systems. The company is deploying its in-house AI assistant, “Sparky,” directly within ChatGPT. Users will interact with Sparky inside the ChatGPT environment, but will be prompted to log into their existing Walmart accounts to complete purchases. This approach synchronizes shopping carts across Walmart’s website, mobile app, and ChatGPT, ensuring a consistent experience and allowing purchases to reflect ongoing shopping behavior.
This shift represents a significant change in strategy for OpenAI, which initially envisioned its platform as owning the transaction layer in agentic commerce. The Information reported that OpenAI is now acknowledging the need for a more merchant-centric architecture, where retailers retain control over the cart, customer identity, loyalty data, and the economics of the transaction. OpenAI’s role, in this revised model, becomes focused on product discovery and distribution.
Early data suggests the new approach is yielding better results. According to Danker, Sparky within ChatGPT is currently converting at roughly 70% of the rate seen on Walmart.com, a substantial improvement over Instant Checkout. While these figures are preliminary, they indicate that maintaining control over the core shopping experience is crucial for driving conversions.
A Broader Retailer Response
Walmart’s experience is already influencing the broader retail landscape. Shopify has announced it will no longer support native checkout within ChatGPT, opting instead to redirect transactions back to merchant storefronts. Etsy has confirmed to CNBC that it is developing a dedicated ChatGPT app, also with the goal of maintaining control over its shopping experience. OpenAI itself has acknowledged the stronger user engagement in product discovery and search, suggesting a refocusing of its efforts in those areas.
Amazon, however, is taking a different tack, keeping AI commerce tightly integrated within its own ecosystem. The company continues to expand tools like its Rufus shopping assistant and “Buy for Me” agent, alongside features like “Shop Direct,” which allows customers to browse and purchase items beyond Amazon’s core marketplace. Amazon’s strategy emphasizes embedding discovery, recommendation, and transaction flows within its own platform, reinforcing its control over the entire shopping journey.
The evolution of AI-powered commerce is clearly still unfolding. Walmart’s initial experiment with OpenAI’s Instant Checkout provided a valuable, if somewhat cautionary, tale. The company’s subsequent move to integrate its own AI assistant, Sparky, highlights the importance of control, seamless integration, and a familiar customer experience in driving successful AI-driven sales. The coming months will be critical as retailers and AI platforms alike refine their strategies and navigate this evolving landscape.
Looking ahead, Walmart plans to extend the Sparky integration to Google’s Gemini, further expanding its reach within the conversational AI space. The company will continue to monitor performance and refine its approach, seeking to leverage the power of AI while maintaining a customer-centric focus.
What are your thoughts on the future of AI-powered shopping? Share your comments below, and let us know how you envision the role of AI in your own shopping experiences.
