Launched with great fanfare in the fall of 2025 as "the next step in agency commerce," Instant Checkout was introduced as a feature that allowed direct product purchases through ChatGPT. However, this feature was quietly discontinued just a few months later. OpenAI shifted direction, retail partners adapted, and the entire industry learned lessons from an experience that revealed the real challenges of shopping with artificial intelligence.

Highlights:

  • Instant Checkout has officially ended. OpenAI has moved away from the native payment system in ChatGPT, directing users to specialized applications that redirect them to the retailer's website to complete their purchases.
  • There are many reasons for the failure: limited catalog, frequently outdated product data, widespread errors, and according to Walmart, conversion rates that were three times lower than direct purchases made on a retailer's website.
  • Retailers are regaining control. Walmart is integrating its own assistant Sparky into ChatGPT and Gemini; Etsy is developing its own ChatGPT application. The trend is clear: major brands want to control customer experience and data, not hand it over to OpenAI.
  • No one has yet found the formula. According to analysts, buyers are increasingly using chatbots to gather information about products, but they are still not completing purchases within the chat interface.

A Noisy Announcement, A Much More Complex Reality

When OpenAI introduced Instant Checkout in the fall of 2025, the online commerce world held its breath. The goal was for ChatGPT users to be able to purchase products without leaving the chat window. Etsy, Walmart, and Shopify rushed to join this system. Shopify's president Harley Finkelstein spoke of a "new frontier" for online retail at the time.

Five months later, the tone changed radically. Walmart's senior vice president of artificial intelligence Daniel Danker strikingly summarized the situation at Morgan Stanley's Tech, Media & Telecom conference in early March 2026: Instant Checkout, he said, "was just a very temporary moment." He added that this experience would end within a month.

Why Did It Fail?

The issues with Instant Checkout were both technical and structural. OpenAI partially relied on scraping and crawling methods from retailers' websites for product data. As a result: stock, delivery times, or price information were often incorrect or outdated. Emily Pfeiffer, chief analyst at Forrester, states directly: "Crawling and scraping are not sufficient to obtain the depth of product data necessary for a satisfying shopping experience."

Integration with vendors also turned out to be much more challenging than expected. Despite promises that there would be more than a million Shopify vendors available initially, by the time Instant Checkout ended, only about 30 vendors were available. Walmart had offered approximately 200,000 products online; while this was a significant number, it was insufficient to meet a modern retail catalog.

Bob Hetu, an analyst at Gartner, clearly states the situation: "OpenAI underestimated the difficulty of enabling transactions. This is somewhat surprising on one hand, but not easy for retailers either."

Killing Data: Conversion Rates

The figures shared by Walmart are extremely striking. The conversion rates for products sold directly through ChatGPT via Instant Checkout were three times lower than those recorded when the user was redirected to the retailer's website to complete the purchase. A March 2026 Adobe-Semrush study confirms this trend in a panel of over 1,000 American consumers: only 22% reported having purchased a product directly through an AI tool. On the other hand, half stated that they made a purchase after obtaining information using AI.

The message is clear: chatbots have become powerful tools for exploration and product research, but they are not yet being used to complete purchases.

New Model: Retailers Hold the Key

In light of this failure, OpenAI and its partners decided to make a fundamental change. Instead of centralizing transactions in ChatGPT, the new model is based on "ChatGPT Applications" developed by each retailer. When a user finds a product through the chatbot, they are redirected to the brand's website to complete the purchase; this occurs either in an integrated browser within the ChatGPT mobile app or in a separate web tab.

Walmart exemplifies this transformation best. The large retail chain is directly integrating its shopping assistant Sparky into ChatGPT and Google Gemini. The mechanics are simple but strategic: when a user asks ChatGPT to find a deal on a product, the request is directed to Sparky, which queries Walmart's inventory, presents options, and processes the order, all while staying within the ChatGPT interface. OpenAI receives an exposure and potentially access fee, but customer data, transaction, and post-sale relationship remain with Walmart.

The strategic importance is significant: "We learned that our customers want consistency at every touchpoint," says a Walmart spokesperson. Data from pilot tests of Sparky in ChatGPT shows that users accessing this tool completed purchases at approximately 70% of the rates observed directly on Walmart.com; this is a much more satisfying result than Instant Checkout. According to Walmart, the key is trust: customers know they are interacting with a Walmart representative even within another application.

Etsy is also adopting the same logic. The marketplace is developing its own ChatGPT application; this will not only allow it to better control the shopping experience but also grant earlier access to buyer data during the purchasing process; Instant Checkout only provided visibility at the final transaction moment.

Shopify also confirms that the e-commerce experience in ChatGPT needs to be rethought. Shopify vendors will be included in the chatbot, but payment will no longer be native: it will be conducted in the vendor's online store. Good news for retailers: they will not need to develop a special application for ChatGPT to benefit from this integration.

Structural Tension Between AI Platforms and Retailers

This transformation reveals a fundamental tension between two incompatible logics. Platforms like OpenAI and Google aim to be a universal interface layer where commerce occurs. Retailers, on the other hand, have spent decades building customer relationships and do not want to become merely interchangeable catalogs in someone else's application.

Amazon exemplifies this defensive stance perfectly: the Seattle-based company has blocked access to its website for many AI tools, including ChatGPT, and has sued Perplexity for attempting to scrape its pages. Perplexity claimed that Amazon only wanted to protect its advertising revenue because AI tools "cannot see" that revenue. Meanwhile, Amazon continues to invest in its tools: developing shopping chatbot Rufus, the "Buy for Me" assistant, and the "Shop Direct" function that scans products outside its catalog.

Google, on the other hand, is in a more comfortable position. Because it is deeply integrated into shopping through search and advertising, hosting third-party shopping tools in Gemini does not threaten its business model. On the contrary, it keeps users within its ecosystem.

A Series of Statements Filled with Question Marks

The abandonment of Instant Checkout does not occur in a vacuum. As some experts point out, this transformation is part of a series of setbacks for OpenAI: the $500 billion Stargate project announced with Donald Trump in January 2025 is reportedly experiencing significant delays; the $100 billion Nvidia partnership announced in September 2025 was quietly reduced to $30 billion in early March 2026; or situations like the agreement with the Pentagon that was urgently announced by Sam Altman and partially rolled back after a few days of general criticism.

These recurring events are raising concerns among investors, as OpenAI is considering going public in a potential artificial intelligence bubble environment. Several major tech companies that went public in 2025 are now trading about 60% below their IPO prices. Altman acknowledged that these recent events were "a good learning experience"; this seems a bit hollow after more than a decade of communication about artificial intelligence.

The market reaction regarding the abandonment of Instant Checkout is also noteworthy: stocks of many specialized platforms such as travel bookings and food delivery rose between 3% and 13% in the following hours. Investors feared that ChatGPT would eliminate these services, but found some peace.

An Unshakable Sector, In the Construction Phase

However, reading the abandonment of Instant Checkout as the death of agency commerce would be excessive. Emily Pfeiffer from Forrester is clear on this: "It’s not the death of agency commerce." Everyone thinks that others have found the solution or are further ahead. The reality is that no one has yet found the formula.

Google has updated its shopping tool with features such as real-time product data uploads, which ChatGPT has not yet mastered: adding multiple products to the cart simultaneously, preventing stockouts and price errors, and connecting to loyalty programs.

The seemingly shaping model of "owning the agency, renting the distribution" may become the norm in the sector. Retailers with sufficient resources will invest in their own agencies and use them as distribution channels on artificial intelligence platforms. For others, solutions will be offered that provide visibility without requiring special development, as suggested for Shopify's vendors.