Walmart is building the technology to compile intimate profiles of shoppers and use that data to charge them accordingly.
By Lindsay Owens and Elizabeth Pancotti
Last month, Walmart CEO John Furner published a letter to customers in the face of growing backlash against the company’s pricing practices. In recent months, customers and consumer advocates have sounded alarms about the data the retail giant collects on shoppers and how it’s used and sold, as well as the high-tech systems it’s deploying to change how we shop.
Walmart plans to install digital price tags in all U.S. stores by the end of the year. These devices, called electronic shelf labels (ESLs), can change prices in mere seconds. Using ESLs, Walmart can deploy price changes remotely, in many stores at once, depending on the time of day, demand for the product, or even who is standing in front of the product in the store. As a result, Walmart shoppers could be charged a different price at the register than the one they saw when they put an item in their cart. Or worse, paying a different price than the customer in front of them in the checkout line.
Last year, Walmart launched “Sparky,” its in-house AI shopping assistant. Customers can ask Sparky to recommend products, generate shopping lists, and even make purchases on their behalf. The tool is now used by roughly half of the tens of millions of monthly active users of Walmart’s mobile app. Customers must log in to their Walmart accounts to use Sparky, allowing the retailer to store information from their interactions. Further, unlike other in-house AI agents (like Amazon’s Alexa for Shopping), Sparky connects to external platforms like ChatGPT. The investment in these technological developments is already paying off for Walmart. Executives have bragged about how Sparky users spend more than those who do not use the agent: according to Dave Guggina, president and CEO of Walmart U.S., “when customers engage with Sparky, their average order value jumps 40%.”
In August, Walmart updated its customer privacy notice, spelling out what data the retailer collects on shoppers and how that data is used – and in some cases, sold. Among the many pieces of information it collects on shoppers, the privacy notice lists biometric information like retinal imagery, demographic information like ethnicity, financial information like income, vehicle information like license plate numbers, and precise location information that can identify where you are standing in a store. What Walmart doesn’t collect itself, it buys from third-party vendors. And what is Walmart doing with all of this data? Training its AI models, personalizing advertisements, making inferences about shoppers, and in some cases, selling it. In several places, the privacy notice notes that Walmart may receive payment for the data it gives to external companies.
Furner is indignant: “We price the product, not the person.” While Furner assures customers their “income, shopping history, urgency” or what Walmart thinks they could pay “won’t change the price,” the company has spent the last decade developing technology that would allow it to do the opposite.
Walmart is building the technology to compile intimate profiles of shoppers and use that data to charge them accordingly.
Across several patent applications, Walmart details the troves of data it can collect on users. These are just some of the invasive facts it could file away in customer dossiers:
Combined with the miles-long list of personal data it collects on shoppers, outlined in its most recent privacy notice, these patents show Walmart building detailed consumer profiles – and the infrastructure to deepen them – seemingly so it can price people, not products.
Walmart can de-anonymize cell phone location data it gets from cell phone carriers. When customers walk into a Walmart store or pull into a Walmart parking lot, the retailer can collect their phone’s MAC address – effectively a digital fingerprint for the device. Walmart can also obtain anonymized location data from cellular carriers or other vendors. But according to one patent filing, if it merges that anonymized data with MAC address data and store purchase data (such as names on credit/debit cards swiped in stores), it can decode the anonymized ID of a customer’s cell phone from the external data source. For instance, if a customer visited and made three purchases at a Walmart store last month, the likelihood of another cell phone being in that Walmart location at those exact times – and only those times – last month is very small, allowing Walmart to match the name from the customer’s credit card with their anonymized cell phone ID. They may not even need a customer to visit three times; the patent notes that it can link a customer to a device “in many cases, as a result of only a single customer visit.”
This lets Walmart track customers well beyond the walls of its stores – it can see not only when someone was at Walmart, but data “representing the location and travel” of customers’ devices in “real-time, near-real time […] and/or historical.” The patent details how Walmart can use this data to identify the restaurants customers frequent, where they live, and how they commute to work. This data is then stored as part of customer profiles, which determine what discounts and coupons individual customers get – or do not get.
Walmart can tell your favorite Ben & Jerry’s flavor by seeing which one makes your heart race. One patent application details an invasive system to determine user preferences. First, customers would be tracked within the store using facial recognition on store cameras, drones flying around the store, or cell phone signals when you’re connected to the store’s Wi-Fi. Then, data from heart rate and glucose monitors could reveal how customers feel about ice cream flavors. If pulse or glucose levels are different when a customer looks at two different ice cream flavors, it may reveal “which flavor the customer is more likely to purchase.” To determine which flavor made your pulse or blood sugar spike, the technology can compare them to your “baseline physiological reaction” – your heart rate or blood sugar levels when you are sleeping.
Walmart can tell when you miss your usual workout class. One patent application details how Walmart could establish a “general sense of” a person’s daily routine by “monitoring a person’s behavior over time,” such as “the person’s time at their place of employment or their time spent at their child’s sports practices,” as well as less regular activities, “such as visits to local stores, movie theaters, and the homes of nearby friends and relatives.” The patent filing describes a process that would detect changes to an established routine. As an illustrative example, it describes “detecting that the person has not arrived at their usual 6PM-Wednesday dance class.” How on earth would Walmart know how long you stick around your kid’s soccer practice or if you miss a gym session? By collecting GPS data from your cell phone, or worse, “surveillance cameras, social networking posts and updates, [and] traffic cameras.”
From the moment you bring a new baby home from the hospital, Walmart can track their every milestone – and steer you toward pricier products accordingly. A patent that Walmart was granted in 2024 describes a system that tracks the items you purchase for a child – size 3 diapers, stage 2 baby food, 6-month onesie – and the items you stop purchasing – trading bottles for sippy cups – to “determine specifically the number and ages of juveniles in the customer’s household.” And the retailer doesn’t just size up your kids once. The system can update estimates of their ages “on a periodic basis such as monthly or yearly” as they grow. Walmart is candid about what it hopes to do with this data – steer customers toward different price points based on data about their children. The patent states that a customer “with a large number of children may be offered bulk sizes […] or less expensive product, whereas a household with a smaller number of juveniles may be offered […] smaller quantity packaging or more expensive products.” And it isn’t just collecting data from online purchases or parents – the patent details that Walmart can identify parents upon “entering a store” and link “aunts, uncles and grandparents” to the same household profile.
Walmart can understand exactly how much you’re willing to pay by running pricing experiments on you. In a patent Walmart was granted in 2019, the retailer admits that while it can build a profile of individual customers’ values, affinities, aspirations, and preferences, how strongly a customer holds them “can be difficult to accurately determine,” so it starts from “an estimated or presumed” profile and refines it. To pinpoint exactly how much they’ll pay, it must “probe” customers. The patent describes a system that runs experiments on customers by marketing a product that is outside of a customer’s typical price range – one that “costs the customer a premium” above “what the customer historically has absorbed.” If the customer purchases the product and “absorbs” the premium, Walmart now knows they will pay more than it previously assumed. These results allow Walmart to store estimates of price sensitivity in customer profiles.
In a separate patent filing, Walmart admitted that it tested a price experimentation algorithm over a five-week period on more than 20,000 items. The patent outlines how price experiments could test daily price changes, offering an example of price swings of up to 10% per day. Walmart acknowledges that these experiments “can harm the user experience,” and though these experiments did not estimate product demand based on shopper-level data, the patent laments how difficult it is to track users who switch devices, log out, or use Incognito mode – all motivation for the robust set of customer-tracking technologies discussed above.
Walmart has designed a system that withholds coupons from customers who will buy products at full price. In a patent application titled “Systems and methods to generate coupon offerings,” Walmart describes sorting customers by how likely they are to buy a product at full price. Customers who are likely to buy a product without a discount receive “marketing materials without the coupon.” But if a coupon might change a customer’s mind about making a particular purchase, the discount they’re offered is “distinct.” As a result, the price a customer sees is calibrated to how much Walmart thinks they can charge and still make the sale. In fact, the patent notes that when a customer has a 90% probability of purchasing a product, “a coupon that provides a reduced purchase price for the product is not needed.”
The Facebook posts and tweets you like may determine if you get a coupon. Another patent filing outlines a system that can be used for “determining appropriate pricing” using social media likes, shares, and posts about products to estimate if customers are likely to buy a product at full price and only sends a coupon to those who are unlikely to buy without one. The patent notes that “if a customer ‘likes’ an item on a social media service, the marketing application can choose to send that customer a coupon for that item, if the customer is unlikely to purchase it, or may send a communication simply describing the item” – but not discounting it – “if the customer scored as likely to purchase it.”
Walmart has developed technology that can adjust prices based on what items are already in your cart. Walmart’s app has a “Scan & Go” tool, which allows customers to ring up items as they place them into their cart. A patent granted to Walmart in 2023 details how the company might use it to charge different prices to customers shopping for the same item, at the same store, at the same time. The patent says that “labels and prices may be changed in a dynamic fashion” based on what products customers have already scanned and put in their cart by pushing price adjustments to electronic shelf labels. It offers an illustrative example of how prices might be customized: “if a customer has tuna fish, they may be offered a different price for mayonnaise.”
When you search for an item on Walmart’s app or website, the price you see could be based on your identity. In a patent granted to Walmart in 2021, titled “Dynamic pricing systems and methods,” Walmart describes a technology that can set prices based on customer profiles, purchase history, and how close they are to a store when they look up an item. The patent states that the system “determine[s] an identity of a user performing the search” and sets “current pricing of the items to be displayed based on the user identity.” The patent gives the example of a range of prices that could vary by 10%. In other words, some customers could be charged $110 for a product while others are charged $90.
Walmart can estimate how price sensitive you are in each aisle of the store. One patent application details a “customer insight computing device” that can build a shopper profile based on in-store and online purchase data, as well as online browsing data, such as what items you click on and add to your cart, even if you don’t buy them. Using this data, it can estimate each customer’s price sensitivity for each product category. For example, a shopper may want to get “the best bang for the buck” on clothes, but “tends to buy the best products” in tech.
A separate patent application notes that these price sensitivity estimates can be used to target product recommendations and discounts. The patent details an “algorithm that can utilize personal information [to] help identify” customers who are “interested in paying a lower price.” It notes that it can target value-conscious customers (those who buy discounted and generic items more frequently) with “larger discounts and savings” and “additional price discounts,” while customers who are less price sensitive get smaller discounts. The consequences of such price discrimination extend to all shoppers. The mechanism for disciplining prices in our economy is the bargain shopper. Right now, companies have to accommodate coupon-clippers, which keeps overall prices lower. With personalized pricing, they can identify price-sensitive shoppers, push them a discount and make sure they get a better deal, and soak the rest of shoppers.
Walmart can set the price you pay based on how far you drive to the store or how much weight your car gains between when you enter the parking lot and when you leave. One patent application details how cameras in parking lots could photograph vehicle suspensions on entry and exit to measure the gap between the wheels and wheel wells to infer whether the car got heavier from the customer’s purchases. Combined with license plate readers, the system can determine how far the customer drove from home to the store and tie purchases to license plates using names on credit cards and vehicle registrations. The system can generate a per-vehicle profile and “extend customized sales and/or offers.” As a result, if the person in front of you in the checkout line drove from farther away or had a heavier grocery haul last week, you could pay more than them for the same items.
When Walmart screws up your order, its response can vary based on how much of a “Karen” it thinks you are. Even if Walmart charges you the same price as your neighbor when you order an item, you might not get the same compensation when your orders go wrong. One patent application describes an “appeasement” engine that decides whether and how much to offer customers when an online order is canceled or delayed. The model takes into account “persona of customer” and “demographics” before deciding to offer a high-value compensation, like a gift card or a discount on a future purchase, or nothing at all.
Walmart can charge customers more for urgent orders. One recent patent application notes that “a customer may be charged an increase” when a customer places a curbside pickup order and flags that it’s urgent.
Walmart executives can claim that they don’t personalize prices, but they’re spending considerable resources to develop technology that would do just that. Corporations do not spend the time and money, or hire highly specialized staff, to develop and patent technology only to file it away on a hard drive never to enter a store or be coded into the back end of a website. They do it to increase their bottom lines and deliver the most profits they can to their shareholders. Walmart isn’t building detailed profiles on shoppers just to serve them more relevant advertisements. It’s developing the architecture to personalize prices so it can charge customers exactly what they’re willing to pay – and not a penny less.