That’s according to a PYMNTS report in May, which added that Sparky users consistently generate larger average basket sizes than customers who don’t use it and that Walmart U.S. has now recorded nine consecutive quarters of digital growth above 20%.
The retailer’s next move: make Sparky smarter by collecting data it has never captured before—what actually happens inside its 3,500-plus stores.
Walmart’s U.S. Chief Technology Officer Hari Vasudev told The Information the company is now logging in-store app usage, tracking shopper conversations with staff and pulling signals from services like urgent care and auto repair. The goal is to tie shopper intent to what’s actually on the shelf.
Retail intelligence startup Radar raised $170 million at a $1 billion valuation last month on the same thesis. Both investments share a diagnosis: stores are fulfillment centers now, and they’re running on data infrastructure that was never built for that job.
Where the Sales Actually Are
The eCommerce figure of roughly 16% of total U.S. retail has long been used to argue that physical retail is holding its ground. PYMNTS CEO Karen Webster wrote in January that the number relies on a denominator that is a mirage. It masks where consumer choice actually exists and where purchases are shifting.
Excluding autos, gas and most of grocery, Webster wrote, online and digitally influenced transactions are outpacing purely physical sales for the first time. In discretionary, higher-margin categories including apparel, electronics, home, beauty and general merchandise, online already accounts for between 30% and 50% of sales.
The skew sharpens inside specific categories. PYMNTS Intelligence found that online shoppers are 76% more likely to buy hobby items, 61% more likely to buy electronics and 61% more likely to buy sporting goods than in-store shoppers. Of purchases already happening online, PYMNTS reported that 48% of clothing and accessories, 61% of electronics and 75% of sporting goods and hobby purchases now flow through Amazon alone. Grocery is the holdout. Most of retail isn’t.
AI is compressing the timeline further. PYMNTS Intelligence found that 41% of consumers used dedicated AI platforms for product discovery as of December, and 43% said they’d fully replaced their prior methods. Purchase decisions are forming before a shopper opens a retailer’s app. The store’s job has narrowed to one function: fill the order accurately.
What Breaks When the Data Isn’t There
However, brick-and-mortar stores can face inventory challenges.
A product listed as available online may be missing from a shelf; employees can spend hours manually checking stockrooms; popular items may go out of stock before workers notice. Online order fulfillment suffers when store inventory data is inaccurate.
Radar aims address these problems.
Radar uses ceiling-mounted RFID sensors that read tagged items every 8 seconds across the sales floor, stockroom and fitting rooms, the company said in a May press release. The company said it processes more than 100 billion item-level events per day and is deployed across more than 1,400 stores. That feed drives automated replenishment, omnichannel fulfillment routing and loss prevention, and is integrated into systems retailers already run.
Why Walmart Is Capturing Data It Never Bothered With Before
Walmart’s Sparky investment works the same problem from the demand side. Walmart CTO Vasudev told The Information that when a customer browses the Walmart app while standing in a store, or asks a floor employee a product question, that interaction leaves no trace in Walmart’s systems today. Logging those signals, he said, lets Sparky tie shopper intent to available supply.
Walmart is also exploring whether AI partners like OpenAI could share context about what shoppers search for outside its own app, Vasudev told The Information. So far, that data sharing is limited. OpenAI shares some information with outside partners but not actual prompts or conversation history.
