Walmart Put Checkout Inside ChatGPT. It Converted Three Times Worse Than Sending People to Its Own Site.
The retailer ran the experiment everyone told creators to run. The numbers killed it — and pointed to where your buy button should actually live.

For a year the advice was the same: make your product buyable inside the chat window before someone else does. Then a retailer big enough to measure it properly published a number.
Starting in November, Walmart offered about 200,000 products through OpenAI's Instant Checkout, letting users complete purchases inside ChatGPT without visiting Walmart's site. Walmart EVP Daniel Danker said those in-chat purchases converted at one-third the rate of click-out transactions, called the experience "unsatisfying," and confirmed Walmart is moving away from it. In plain terms: moving checkout into the AI interface cut conversion by about 66%.
Where the coverage disagrees
The number is consistent across outlets. The paper trail around it is not.
Accounts differ on where Danker said it — one report places the remarks with Wired in March 2026, another at a Morgan Stanley conference on March 4, another at Shoptalk 2026 — and even on his title, given as EVP of product and design in some coverage and EVP of global technology in others.
Adoption counts are messier still. By February 2026, roughly 30 Shopify merchants were actually live on Instant Checkout, per Forrester principal analyst Emily Pfeiffer — against the "over a million" once promised. Other write-ups put the figure at fewer than 15. Treat the direction as solid and the decimals as contested.
The part that should make you happy
AI traffic isn't the problem. In-chat checkout was.
On Shopify's Q2 2026 earnings call on August 5, AI-driven traffic and AI orders both grew 3x year over year platform-wide, AI sessions landing on product pages converted 2.5x higher, and new-buyer orders from AI channels ran about 2x other channels. Adobe Digital Insights data reported by eMarketer found AI-referred retail visitors converted 42% better than non-AI traffic in March 2026, spent 48% more time on-site and produced 37% higher revenue per visit.
Scale check before you rebuild anything: one portfolio of ten established stores found ChatGPT, Perplexity, Gemini, Copilot and Claude combined driving 0.10% of sessions and 0.10% of revenue in 2026 so far. Strong intent, small volume, fast growth.
OpenAI already moved
OpenAI retired Instant Checkout in March 2026, about six months after launch, and repositioned ChatGPT toward product discovery, with checkout moving back to merchants' own environments. Asked what "complete purchases inside ChatGPT" now means, OpenAI told Modern Retail the phrase describes product discoverability: "Instant Checkout is moving to Apps, where purchases can happen more seamlessly." Walmart's replacement is its own chatbot, Sparky, embedded inside ChatGPT — users log into Walmart, sync carts across platforms, and pay inside Walmart's system.
Consumers were never asking for the original version. Forrester's March 2026 research found completing a purchase inside an answer engine was the least adopted use case across all demographics.
If you sell a $497 course
You were never the target customer for any of this — the pilots ran on physical retail SKUs. But the strategic read transfers cleanly.
AI is a discovery channel. Your checkout page is where the money is made, because that's where the order bump, the upsell, the payment plan, the email capture and the student data live. None of that survives a handoff to somebody else's interface.
What does deserve your afternoon: making sure the assistants describe your offer correctly. Merchants found that the products and prices AI displayed needed constant synchronization, and when they drifted the customer experience fell apart. Analysts tracking AI referral bounce rates point to the same cause — the AI described the product inaccurately, the shopper arrived expecting something different, and conversion turned out to be downstream of AI accuracy rather than traffic quality.
Publish the price. Publish what's inside. Publish the refund window. Then let the robot send the traffic and keep the buy button yours.
Sources
- Walmart: ChatGPT checkout converted 3x worse than website — Search Engine Land
- Walmart says ChatGPT checkout converted 3x worse than its own website — MarTech
- Why AI Checkout Stalled: Discover in AI, Buy on Site — Digital Applied
- ChatGPT Instant Checkout: What Happened to It in 2026 — Hypotenuse AI
- What went wrong with ChatGPT's Instant Checkout — Modern Retail
- OpenAI Kills Instant Checkout: What This Actually Means for Merchants — Agentic Commerce Feed
- State of Ecommerce 2026: AI Traffic, Shop App & Channel Data — Ethercycle
- ChatGPT AI visibility statistics for 2026: referrals, citations and conversion — Cognizo
- The Real ChatGPT Referral Conversion Rates and AI Search ROI for Shopify Ecommerce (2026) — Metricus
- ChatGPT Shopping Instant Checkout Guide 2026 — Evolve Media Agency
- AI Was Supposed to Buy Products For Your Customers. It's Sending Them to Your Store Instead. — Earning A Living Online
- Walmart ChatGPT Checkout Converted 3x Worse | Analysis — BuildMVPFast
Editor’s note: Half the audience has been guiltily planning to make their course buyable inside a chatbot. Walmart just measured what that costs.
