The Checkout That Converted Three Times Worse Than a Link

Walmart made approximately 200,000 products available through OpenAI’s Instant Checkout beginning in November, allowing ChatGPT users to complete purchases without leaving the chat interface. The experiment produced one of the first hard performance figures in AI-mediated commerce: conversion rates for products sold directly inside the chatbot were three times lower than those requiring users to click out to Walmart’s website, according to Daniel Danker, Walmart’s executive vice president of AI acceleration, product and design, in a report published by WIRED on March 18. Danker called the experience “unsatisfying.” This week, Walmart replaced it.

The new experience, now live in web browsers with iOS and Android app access to follow, is built around Sparky, Walmart’s commerce AI agent. It supports the full shopping journey, from discovery inside ChatGPT to account linking, loyalty program integration, and payment within a Walmart environment. Checkout no longer happens inside the chatbot. It happens at Walmart.

Instant Checkout Failed on Structure, Not Concept

The conversion gap was not random. The single-item checkout model created a structural mismatch with how shoppers actually behave. Danker put it plainly to WIRED: shoppers feared that interacting with multiple product recommendations would result in separate boxes arriving when they wanted everything in one order. They did not want a checkout experience that separated one chatbot-purchased item from the other products already sitting in their Walmart cart.

Walmart knew this going in and managed its exposure accordingly. The retailer deliberately excluded certain products from Instant Checkout. Televisions, for example, were kept out because buyers typically need accessories, and Walmart’s website can handle bundling in ways ChatGPT could not. Checkout is not a commodity step at the end of a transaction. It is where loyalty programs, bundle logic, shipping optimization, and customer identity converge. Instant Checkout abstracted all of that away, and shoppers noticed.

OpenAI acknowledged the gap directly, stating that the initial version of Instant Checkout did not offer the flexibility it aimed for, and is now directing merchants to run their own checkout experiences while concentrating its own efforts on product discovery. The company is expanding the Agentic Commerce Protocol to support richer catalog data, visual product grids, and side-by-side comparisons, all upstream of purchase.

The revised architecture addresses the core problem. Baskets from Walmart’s website, the Walmart app, and ChatGPT will synchronize, allowing purchases to reflect ongoing shopping behavior rather than a single isolated transaction. Danker described the philosophy to WIRED as Sparky traveling to external platforms and bringing the Walmart store to the customer, rather than routing them into a disconnected experience.

Discovery Is Now the Contested Ground, and Catalog Data Determines Who Wins It

For 3P Marketplace sellers, the shift from checkout to discovery changes where the competitive work happens. Under Instant Checkout, the question was whether your item could transact inside the chatbot. Under the expanded ACP, the question is whether ChatGPT can ingest your structured catalog data, understand your inventory, and surface your product accurately in context.

ACP integration starts with a structured product feed covering titles, descriptions, images, price, and availability, shared with OpenAI on a daily basis with real-time updates sent throughout the day via API, per OpenAI’s developer documentation. For Marketplace sellers whose item content flows through Walmart’s systems, the practical question is whether that data is clean and current enough to be read accurately by an AI model making autonomous product selections. An agent that encounters stale availability data or an ambiguous product title does not ask for clarification. It moves to the next result.

Shopify merchants receive automatic integration through Shopify Catalog, enabling their products to appear in ChatGPT conversations without additional setup, per OpenAI’s March 24 announcement. Marketplace sellers who rely on Walmart’s item data infrastructure should understand that automatic pass-through is not currently available to them through the same path. ACP integration requires deliberate feed management outside of what Walmart’s systems provide natively.

For 1P suppliers, the catalog question is different but equally consequential. Sparky does not surface Walmart’s assortment indiscriminately. Categories requiring accessories, complex fulfillment, or high-consideration purchase behavior were deliberately excluded from Instant Checkout. The implication for category and replenishment teams is that item content quality, meaning accurate attributes, clear variant modeling, and current availability, will increasingly determine whether a product appears in Sparky-mediated sessions, not just in conventional search.

The Shopper Arriving Through ChatGPT Is Not the Shopper Walmart Already Has

Danker disclosed at Morgan Stanley’s Technology, Media and Telecom conference earlier this month that the top two product categories purchased through Sparky in ChatGPT are vitamin supplements and protein supplements, purchases he attributed to conversational queries that did not begin with direct purchase intent. The pattern he described: a shopper asking what they need to know after starting GLP-1 medication, receiving product recommendations, and completing a purchase without having entered a commerce environment intentionally.

According to WIRED’s reporting, the five best-performing categories through Instant Checkout were vitamins and protein supplements, automotive parts, beauty products, hardware, and tools, which together accounted for more than half of all Instant Checkout orders. That list is not random. These are categories where a shopper can make a confident single-item purchase without needing to evaluate accessories, compare fit, or coordinate with other items in a basket. They are also categories where a health, lifestyle, or how-to query can naturally surface a product recommendation without the session beginning as a shopping trip.

WIRED also reported that ChatGPT is now bringing Walmart roughly twice the rate of new customers compared to search engines. Danker attributed that partly to the demographic profile of heavy ChatGPT users, who tend not to be typical Walmart shoppers. Suppliers in categories that over-index with higher-income or health-conscious consumers should register that this channel is not simply recirculating Walmart’s existing customer base.

A Gemini integration is scheduled for April, according to Danker. The suppliers whose products Sparky surfaces consistently across both platforms will be determined in large part by item content decisions being made now, before either integration reaches full scale.

Winning With Walmart

Winning With Walmart is an independent platform for suppliers, sellers, solution providers, and industry experts.

Articles are developed by staff editors, contributing experts, and trusted partners.

Some are researched and drafted with the assistance of AI tools and are reviewed, fact-checked, and edited by Winning With Walmart editors before publication.

Editorial judgment, sourcing decisions, and final approval rest with the publication's human editors in every case.

We are committed to accuracy and fairness. If you believe this article contains an error, we welcome your feedback.