Doing things one at a time often makes a lot of sense. As the popular old adage goes, “a journey of a thousand miles begins with a single step.”

Starting a new healthy habit like exercising? Build it one day at a time!

Have a huge backlog of emails to get through? Focus on knocking them out one at a time!

Want to order 10 items on your shopping list? Do so one at a time– 10 separate orders, payments, and shipments!

Hold on a second 🤔… that last one doesn’t sound like a particularly good idea . And that’s exactly what was happening when ChatGPT users tried building an online order during the pilot of Walmart’s ill-fated Instant Checkout feature.

The goal was to increase conversion rates by allowing shoppers to make Walmart purchases mid-conversation in ChatGPT from a catalog of about 200,000 products. This sounds great in theory, but Instant Checkout provided anything BUT a smooth shopping experience. Instead, it broke carts up into individual orders and transactions, leaving customers getting bombarded with tons of small packages as opposed to more logically-packed multi-product ones.

The end result? Conversion rates wound up being three times lower than they were when users actually clicked out to the Walmart website.

Some companies might have spent years on re-programming something new, or maybe even scrapped the partnership and project altogether. But Walmart and OpenAI took a different route– installing Walmart’s existing AI chatbot, Sparky, inside of ChatGPT. Yes, that’s right… a chatbot IN ANOTHER chatbot!

It may not be the most elegant or perfect system, but the important thing is that it works. Users log into Sparky once, and from that point on their shopping basket syncs up across all platforms. A cart built on ChatGPT will be accessible on Walmart’s website and app, and a completed purchase will be fulfilled properly through Walmart’s proven shipping and handling logistics.
Like it or not, AI is here to stay. And the data suggests that plenty of users do indeed like it. Walmart has reported that roughly half of its app users have engaged with Sparky, and customers using Sparky spend approximately 35% more per order than shoppers who aren’t using Sparky.

With Sparky taking its place, it won’t be long until Instant Checkout is another distant memory in the failed tech project graveyard. But before that happens, let’s see what we can learn from the wreckage. Because, as Warren Buffett says: “It's good to learn from your mistakes. It's better to learn from other people's mistakes.”

• Customers are using AI for more complex purchases – For simple products like milk and paper towels, shoppers tend to use the standard Walmart app search feature. AI tools like Instant Checkout and Sparky are used for help and information with more complex queries, which explains why more than half of Instant Checkout’s orders came in these collective categories: automotive, beauty, home, hardware and tools.
• AI offers complete shopping solutions – A customer asking which faucet they should buy for their bathroom sink won’t only get that product recommendation– they’ll also learn about tools they may need to install it, or other pieces that would complement the decor. Considering these points, Walmart sellers should keep bundle opportunities in mind and make sure the product details on their listings clearly indicate how an item is used and what it might pair well with.
• Don’t be afraid to pivot – Instant Checkout wasn’t working, so Walmart and OpenAI pivoted to Sparky instead. This is an important reminder that sometimes the best course of action is to admit defeat and reroute your resources as opposed to stubbornly chasing your losses when something doesn’t go according to plan.

Walmart’s Impressive Growth Continues

Amazon was still the undisputed king of third-party U.S. marketplace sales in 2025 with an estimated $300 billion racked up in total. But last year’s sales numbers also showed Walmart slowly but surely making major progress:

• Walmart’s third-party marketplace sales grew 50% year-over-year from $10 billion in 2024 to $15 billion in 2025.
• Fueled by a 30% increase in seller count during the first half of 2025 alone, Walmart now has over 200,000 active sellers.
• Walmart's catalog features over 420 million products, with marketplace sellers powering 95% of those listings.

As Walmart continues to invest in technology like AI integration and in its growth through seller-friendly programs and fee structures, there’s no reason to believe we won’t see another big leap forward in 2026. And if you are still on the fence, I hope that numbers like this encourage YOU to be a part of that leap!

‍