ChatGPT Dropped Instant Checkout. Here's What AI Shopping Looks Like Now

Short answer: in March 2026, OpenAI scaled back Instant Checkout, the feature that let shoppers buy directly inside ChatGPT, after weak merchant adoption and low conversion. ChatGPT now focuses on product discovery and on retailer apps that send buyers to their own checkout, while Google (through its Universal Commerce Protocol) and Perplexity (through Instant Buy with PayPal) keep pushing purchases inside AI answers. For e-commerce teams, the lesson is that AI shopping is real but fragmented: discovery happens in AI engines everywhere, checkout depends on the engine, and complete product data and clear product pages matter on all of them.

What happened to Instant Checkout

OpenAI launched Instant Checkout in September 2025 and announced on March 24, 2026 that it would scale it back in favor of checkout run by merchants. The results had been thin. Etsy saw minimal sales through the feature, and Walmart reported conversion rates three times lower for purchases made inside the chatbot than for clicks out to its own site. Shoppers researched in ChatGPT, then bought elsewhere (Modern Retail).

Retailers also pushed back on what the feature lacked: real-time inventory, loyalty programs, promotions, combined orders and ownership of the customer relationship. "The initial version of Instant Checkout did not offer the level of flexibility that we aspire to provide." - Modern Retail (OpenAI statement, 2026) - company statement.

What replaced it in ChatGPT

OpenAI shifted to merchant-controlled experiences. Retailers including Walmart, Sephora, DoorDash, Instacart and Lowe's now build their own apps inside ChatGPT, which handle discovery in the conversation and then send customers to their own checkout. The Agentic Commerce Protocol that OpenAI built with Stripe remains available as infrastructure for agent-driven purchases (Hypotenuse).

In practice, ChatGPT has become a discovery and recommendation layer. That makes how products are described, compared and cited in its answers more important, not less.

Where in-answer buying is still growing

Other platforms are betting that buying inside AI can work with the right infrastructure. In January 2026, Google announced the Universal Commerce Protocol, an open standard developed with Shopify, Etsy, Wayfair, Target and Walmart. It lets US shoppers complete purchases for eligible products in AI Mode and the Gemini app using Google Pay, and Google added new Merchant Center attributes to help products show up across AI surfaces (TechCrunch).

Perplexity took a payments-partner route. In November 2025, it launched Instant Buy with PayPal for US users, starting with merchants such as Abercrombie & Fitch, Ashley Furniture, Fabletics, Adorama and Newegg (eWeek). "This kind of serendipity is where the best of commerce happens." - TechCrunch (Shopify CEO on UCP, 2026) - industry news.

What this means for e-commerce content teams

Instant Checkout's retreat does not mean AI shopping failed. It means the purchase step is still being sorted out, while the research step has already moved into AI. Wherever the checkout ends up, the product that gets recommended is the one the AI understands best.

Example: a mid-size outdoor gear retailer sells a tent that fits a common query, "best lightweight tent for two people under $300." In ChatGPT, it needs the tent described clearly enough to be recommended, with a link that lands on a product page ready to convert. In Google AI Mode, it also needs complete Merchant Center data so the tent is eligible to be bought in the answer. The same product needs both.

A practical checklist

1. Complete your product feeds. Fill every relevant attribute in Google Merchant Center and other feeds, including size, materials, use cases, compatibility, price and availability, and keep them in sync with your site.
2. Write product pages for questions, not just specs. Answer who the product is for, how it compares and what its limits are, in plain sentences an AI can quote.
3. Make policies easy to find. Shipping, returns and warranty terms affect whether an AI suggests a product and whether a shopper trusts the suggestion.
4. Invest in reviews and third-party coverage. AI answers lean on reviews, comparisons and community discussion, not only on retailer pages.
5. Optimize the landing experience. Shoppers arriving from AI answers have already done research, so product pages should load fast and get them to checkout quickly.
6. Track AI shopping by engine. Separate referrals and conversions from ChatGPT, Google AI Mode, Gemini and Perplexity, since each handles checkout differently.
7. For large retailers, evaluate engine-specific options, such as a ChatGPT app or UCP-based checkout in Google, based on where your shoppers already research.

What to watch next

Watch how widely UCP checkout expands beyond the US and beyond its launch partners, whether more retailers build ChatGPT apps, and whether conversion inside AI answers improves now that merchants control more of the experience. The Walmart data point, three times lower conversion in-chat, is the benchmark to beat.

The bottom line

The first attempt at buying inside ChatGPT stumbled, but AI-driven product discovery is here to stay, and in-answer checkout is still expanding on Google and Perplexity. E-commerce teams don't need to bet on one model. They need product data and pages that every AI engine can understand, and a clear way to measure where AI shoppers come from and where they buy.