ChatGPT Scales Back Shopping Push, “AI for Discovery, Platforms for Checkout” Splits E-Commerce Power

Limits of standalone AI commerce model come into focus
Renewed emphasis on proprietary stores and platform competitiveness
Expansion of experiments across new channels including SNS live commerce

OpenAI has scaled back direct payment functionality in its ChatGPT shopping service, shifting instead toward product search and comparison. As low conversion rates within the chatbot and persistent concerns over consumer trust converged, the structure in which actual transactions take place on retailer platforms has regained prominence. In response, major companies such as Walmart have moved to reinforce strategies that keep payment control within their own systems. Meanwhile, some firms including Gap continue experimenting with models that combine AI platforms and checkout, signaling continued expectations for the expansion of conversational commerce.

Only 22% complete purchases within AI

According to IT publication The Information on the 25th (local time), OpenAI recently decided to significantly scale down its “Instant Checkout” shopping feature offered through ChatGPT. The service, which allowed users to purchase products directly within the chatbot interface, had drawn early attention when it launched last September with participation from major retailers such as Walmart, Shopify, and Etsy. However, delayed updates and frequent errors led to declining user engagement, prompting OpenAI to pivot toward strengthening search, comparison, and recommendation functions instead of direct payments.

This shift reflects underlying usage data. According to figures disclosed by Daniel Danker, Walmart’s vice president of AI acceleration, product, and design, the rate of completed purchases within ChatGPT was only about one-third of transactions completed through Walmart’s own website. A survey by digital marketing analytics firm Semrush similarly found that just 22% of respondents had completed purchases directly within AI tools, while more than half reported using AI for product discovery before finalizing transactions through separate channels.

These realities have directly shaped retailer responses. Walmart, while maintaining its partnership with OpenAI, has integrated its in-house AI shopping assistant “Sparky” both within ChatGPT and its own platform. Even if users explore products within ChatGPT, account linkage, reward application, and payment are all steered back to Walmart’s environment. Danker said, “Shopping can begin anywhere, whether in the Walmart app or ChatGPT,” adding that “regardless of where it starts, it is critical that customers experience our speed and personalization.”

Industry observers view OpenAI’s move as evidence of the limitations of establishing a unified commerce platform centered on payment functionality. The inability to replace existing retail structures—built on consumer trust, payment stability, and brand-driven experience—has become evident. A retail industry source noted that “it takes years to build consumer trust, but it can collapse instantly,” adding that “even a single security issue or error at the payment stage can undermine confidence in AI shopping as a whole.” As a result, ChatGPT is increasingly positioned as a “shopping decision hub” focused on discovery and comparison.

Brand ecosystems shift into “lock-in” mode

By contrast, traditional e-commerce companies appear to be gaining traction by embedding conversational commerce within their own platforms. Earlier expectations had suggested that an “AI commerce” model—where AI handles everything from product discovery to checkout—could diminish the role of established platforms and brand-owned stores. However, actual market developments point in a different direction. Many retailers are integrating AI into their own systems rather than outsourcing it entirely to external platforms, thereby maintaining control over transaction structures. In this model, AI assists consumer decision-making, while final purchases and payments remain within existing retail networks.

The case of Gap, a U.S.-based multinational apparel and accessories retailer, illustrates this approach. Gap has chosen to integrate direct payment functionality into Google’s AI platform Gemini. Users can not only search for products through conversations with Gemini but also complete purchases of Gap products immediately within the interface. Unlike traditional flows that require navigation to a website or app, discovery and checkout are unified within the AI environment. This stands in contrast to Walmart’s approach, which pulls all steps beyond product discovery back into its own system, although Gap likewise maintains its own brand infrastructure as the payment entity.

Other major retailers are pursuing similar strategies. Companies including Target, Nordstrom, Lowe’s, Home Depot, and Wayfair provide product data to external platforms such as ChatGPT while retaining payment authority within their own sites. Sephora also offers personalized recommendations and membership integration through a dedicated ChatGPT app but has deferred payment functionality to later stages. This reflects a strategy of embracing AI-driven discovery while keeping transaction data and customer relationships within proprietary ecosystems.

Experts broadly agree that conversational commerce will only deliver meaningful results when it is fully integrated with existing e-commerce structures. Lee Kyung-hoon, vice president at Channel Corporation, said, “Predictions that platform-centric models would replace existing channels have surfaced repeatedly in the past, but none have materialized,” adding that “as shopping encompasses discovery and experience before purchase, brands are continuing to expand diverse pathways while maintaining their own channels.” AI, in this context, is likely to settle into a supporting role within these multiple pathways.

Challenges such as data consistency remain

Conversational commerce is considered highly effective in influencing consumer purchase decisions, particularly in the pre-checkout phase. A joint survey by the IBM Institute for Business Value and the National Retail Federation (NRF) found that while 72% of global consumers still prefer offline stores, 45% reported using AI during the purchase process. Among them, 41% used AI for product research, 33% for interpreting reviews, and 31% for finding discounts and benefits. This indicates that consumers increasingly narrow down options and complete comparisons through AI interactions before entering a store or launching a brand’s app.

This dynamic explains why many brands are experimenting with conversational tools across multiple channels. As touchpoints expand across messaging platforms, apps, voice interfaces, text, and push notifications, companies are focusing investment on areas that can most effectively influence consumer decisions. New York department store chain Bloomingdale’s operates a one-on-one personalized styling service combining video calls and live chat, while Four Seasons Hotels and Resorts offers a conversational concierge through its app. Norwegian-based Berg Sports uses an AI chatbot to help customers find bicycles and related products.

These developments represent a continuation of retailers’ long-standing efforts to test new media channels, following earlier transitions such as home shopping networks. More recently, live commerce on social media platforms—including Facebook, Instagram, and X—has expanded, enabling real-time product demonstrations combined with live chat and immediate consumer feedback to drive purchases. While SNS live commerce has not achieved the large-scale conversion rates initially anticipated, it has introduced a significant shift by linking discovery, comparison, inquiry, recommendation, and payment into a continuous process.

However, conversational commerce is unlikely to establish itself immediately as a dominant channel. In the NRF and IBM survey, consumers showed clear willingness to use AI, but 54% of brand executives reported difficulties with data consistency across channels and systems, while 51% cited a lack of in-house AI expertise as a major obstacle. Although conversational commerce has strong potential to shape consumer decision-making, achieving full integration across payment trust, comparison accuracy, data connectivity, and channel alignment is expected to require considerable time.