Walmart Sparky for Sellers: How Walmart's AI Assistant Picks Products

What is Walmart Sparky?

Sparky is Walmart's AI shopping assistant. Walmart introduced it on 6 June 2025 as an "Ask Sparky" button in the Walmart app, working across every category: it searches for items, sums up reviews, compares options, answers product questions, suggests what to buy for an occasion and lets shoppers add its picks to the cart. Walmart said at launch that reordering, booking services and input by image, audio and video would follow (Walmart).

Sparky is one of four "super agents" Walmart set out in July 2025. The others serve associates, developers and, under the name Marty, suppliers, sellers and advertisers (Walmart Global Tech, Supermarket News). For a seller the split is simple: Sparky is the one your customers talk to, Marty is the one you talk to.

The important limit is that Sparky only shops Walmart. It is not a web-wide assistant like ChatGPT or Gemini. If a product is not in Walmart's catalog, or its Walmart item data is thin, Sparky has nothing good to say about it.

Where Sparky answers, as of October 2026

Sparky runs in the Walmart app and on walmart.com (The Paypers), and in 2026 it left Walmart's own surfaces. The steps, with dates:

• 11 January 2026. Walmart and Google announced a Walmart shopping experience built by Walmart and reachable inside Gemini through the Universal Commerce Protocol, US first. Linking a Walmart account is optional and brings purchase history and Walmart+ or Sam's Club benefits into the conversation (Walmart).
• 24 March 2026. Walmart ended its Instant Checkout pilot in ChatGPT, which had offered about 200,000 products since November 2025 and converted roughly three times worse than click-throughs to Walmart's own site. In its place, Sparky runs inside ChatGPT (reports differed on whether free users got it at launch or later), and Gemini was named as next. Shoppers sign in with their Walmart account, carts sync, and the purchase is completed in Walmart's own system (The Paypers, Retail Dive). The wider story is in what happened to ChatGPT Instant Checkout.
• 21 October 2026. Walmart announced the Walmart Agent Gateway, an MCP-based route, in invitation-only beta, for outside agents to shop Walmart and Sam's Club, with customer-set spending limits (Walmart). That route runs through Sparky itself, according to Walmart's beta page, so outside agents now get Sparky's view of the same Walmart catalog (PPC Land). In the same announcement Walmart listed Muse, ChatGPT and Gemini among the conversational commerce integrations it runs today.

For a seller, the consequence is that one set of item data now feeds several assistants. You do not optimise for "Sparky in ChatGPT" separately. You fix the Walmart item, and every surface that reads it improves.

What Walmart says Sparky is doing for sales

Walmart has shared a few figures, all from earnings calls and investor events, all as Walmart's own numbers:

• February 2026. On Walmart's fourth-quarter earnings call on 19 February, CEO John Furner said Sparky users' average order value was about 35% higher than non-users', and Walmart said about half of its app users had tried Sparky (Constellation Research, Retail Dive).
• September 2026. Walmart U.S. CEO David Guggina told an investor conference that weekly Sparky engagement grew 60% quarter over quarter, that Sparky users spend about 40% more per order, and that visual search in Sparky converts 57% better than text search (Retail Dive, 8 October 2026).

A higher order value among Sparky users is a correlation: people who use an assistant may be the ones who were going to buy more anyway. Critics have also asked whether bigger baskets mean higher prices. Walmart's answer, in an open letter on 5 October 2026, is that it does not use AI to upcharge customers and that information shared with Sparky does not change the price a shopper pays (Fox Business). For a seller the useful reading is narrower: Walmart is investing in Sparky and steering shoppers to it, so its share of product discovery on Walmart is rising.

Sparky only recommends what Walmart sells. Your Walmart item data is the one input to it you fully control.

Kris Krokos

How Sparky picks products: what is known

Walmart has not published how Sparky ranks or chooses products, and as of 28 October 2026 there is no seller-facing Sparky guide in Walmart's Marketplace documentation. What can be said with confidence:

It reads Walmart's catalog. Sparky recommends from Walmart's assortment, so the item data on it (titles, descriptions, attributes, images, price, stock and reviews) is what it has to work with. Walmart has not said which fields it weighs or whether it treats Walmart-sold and Marketplace items differently.

It summarises reviews. Review synthesis was a launch feature. Recurring complaints in your reviews can surface in an answer even if your listing copy is strong.

It favours what Walmart can deliver. Walmart's pitch for its Gemini partnership leans on speed, including local delivery in under three hours and as fast as 30 minutes (Walmart). An item that is out of stock or slow to ship is a weak candidate for an assistant that is judged on getting the shopper what they asked for.

It is answer-first. A shopper asks "snacks for a school trip, nut-free, under $25" and gets a few suggestions, not 40 results. Attributes that answer the question (allergens, pack size, age range, dimensions) decide whether your item can be suggested at all. This is the same pattern as Amazon's assistant, described in Amazon listing optimization for AI shoppers, and the reason structured product attributes matter more than keywords.

One vendor analysis claims that most sources Sparky cites sit outside the Walmart listing, such as earned media and brand sites (Azoma). Azoma sells optimisation services and Walmart has not confirmed this, so treat it as a hypothesis worth testing, not a rule.

The closest official signal: Listing Quality Score

Without a Sparky guide, the best official proxy is Walmart's Listing Quality Score, the 0 to 100 score Walmart gives each item and catalog in Seller Center. Walmart says it considers content, discoverability, offer, ratings and reviews (Walmart Marketplace). The dashboard also reports post-purchase quality (late deliveries, cancellations and returns), which Walmart says does not affect the score (Walmart Marketplace Learn). The API breaks the score into component scores, including content, price competitiveness, shipping speed, whether the item is published and in stock, and ratings and reviews (Walmart Developer).

Even so, every component measures something an assistant needs to recommend an item with confidence, and Walmart's own recommendations under it (correct category and back-end attributes, clear titles, detailed descriptions, high-resolution images, expedited shipping, competitive prices) are the same work.

Ads in Sparky

Walmart said in January 2026 that it was testing sponsored prompts in Sparky, shown when shoppers ask for recommendations, and that advertisers can only promote their own Walmart-listed products (B&T, Adweek). There is no general availability date yet. If you already buy Sponsored Search through Walmart Connect, ask your contact whether your campaigns can join the test. How Sparky's ads compare with Amazon's sponsored prompts and ChatGPT's ads is covered in ChatGPT ads and retail media in AI answers.

What to do as a Walmart seller

1. Start with the items that already sell. Open the Listing Quality dashboard in Seller Center and fix the content and discoverability gaps on your top sellers before the long tail.
2. Fill attributes beyond the minimum. The required fields get an item listed. The optional ones (materials, allergens, dimensions, use cases, compatibility) are what let an assistant match it to a specific question.
3. Write for questions, not keywords. A title stuffed with search terms does not help an assistant that reads meaning. Key features that state who the product is for and what it solves do.
4. Fix variants. Broken variant groups can lead an assistant to suggest the wrong size or pack. The same logic as in why messy product catalogs fail in AI commerce applies.
5. Treat delivery speed as content. Price competitiveness, shipping speed and in-stock status are part of the Listing Quality Score. An item Walmart cannot get to the shopper quickly is a weaker answer.
6. Check consistency off Walmart. If Sparky and other assistants also read your brand site and reviews elsewhere, conflicting specs work against you.
7. Test it yourself. Ask Sparky the questions your customers ask, in the app and inside ChatGPT, once a month. Record which products it suggests and why. That log is the only Sparky ranking report you will have until Walmart publishes one.

Sparky vs Alexa for Shopping, for a seller

Both are closed-loop assistants: each recommends only from its own store, both summarise reviews, and both now sell sponsored prompts (Amazon's are billed in the US since March 2026 (Acadia); Walmart's are in test). The difference is reach. Amazon's assistant sits in the main Amazon search bar (TechCrunch); Walmart has taken Sparky into ChatGPT and lists Gemini and Meta's Muse among its live integrations, so a Walmart listing can now be read by people who never opened a Walmart app. Our AI shopping assistants comparison puts both next to the open assistants. More on marketplaces in Retail Platforms & Marketplaces.

What we do not know yet

• How Sparky weighs relevance, price, delivery speed, reviews and ads against each other.
• Whether Marketplace seller items and Walmart's own items are treated the same way.
• When sponsored prompts in Sparky leave testing, and how they are priced.
• How far Sparky itself reaches inside Gemini and Muse, and for which users.

FAQ

What is Walmart Sparky?

Sparky is Walmart's AI shopping assistant, launched in the Walmart app on 6 June 2025. It searches Walmart's assortment, summarises reviews, compares products, answers questions and lets shoppers add its picks to the cart.

Is Walmart Sparky free?

Yes, for shoppers. It is part of the Walmart app and site, and linking a Walmart account adds purchase history and membership benefits when it is used inside other assistants.

Can I use Walmart Sparky in ChatGPT?

Yes. In March 2026 Walmart replaced its Instant Checkout pilot in ChatGPT with a Sparky-powered Walmart app. Shoppers sign in with their Walmart account and check out in Walmart's own system.

How do I get my product recommended by Sparky?

Walmart has not published a Sparky ranking guide. The closest official signal is the Listing Quality Score in Seller Center: complete content and attributes, competitive price, fast shipping, in-stock items and good reviews. Then test your category's questions in Sparky yourself.

Can sellers advertise in Sparky?

Walmart said in January 2026 that it is testing sponsored prompts in Sparky, for products listed on Walmart. There is no general availability date as of October 2026.