When your shopping cart predicts what you need: The next frontier of pricing

Put tuna in your shopping cart. Now, reach for the mayonnaise. What if the price changed because the retailer’s technology had already worked out what you were making?

It is no longer fanciful for a retailer to know what is entering a basket before checkout. Walmart’s Scan & Go service lets eligible customers scan items with their phones as they shop, building a digital basket before payment. Separately, electronic shelf labels can make price updates faster. Neither feature proves that an individual shopper is being charged more because of what is in that basket. But they reveal the next frontier of pricing deserves attention.

On The Weekly Show with Jon Stewart, Groundwork Collaborative’s Lindsay Owens discussed Walmart’s pricing patents. Stewart imagined a shopper assembling Sloppy Joes: Once the meat and sauce enter the basket, the retailer can infer that buns are likely to follow. What if the price of the buns could respond to that information?

The business logic is serious

A December 2025 Walmart patent describes how machine learning can model the effects of price changes on complementary products. It is not evidence that Walmart has deployed personalised basket-based price increases. Walmart CEO John Furner has expressly said the company does not price according to shopping history, urgency or perceived ability to pay: “We price the product, not the person.”

Yet, the controversy exposes a bigger shift. Retailers have long studied which goods move together: bread and butter, printers and ink, burgers and buns. Economists call them complementary goods; marketers use those relationships for bundles, promotions and cross-selling. The old question was straightforward: If you buy A, how do I persuade you to buy B?

AI changes the question. If the retailer can infer from your basket that B has become more valuable to completing your immediate purpose, the relationship between those goods is no longer merely a merchandising opportunity. It becomes a potential pricing signal.

This is where cross-price elasticity, an economics concept usually buried in textbooks, becomes practical. It measures how a change in one product’s price affects demand for another. Walmart’s patent uses a machine-learning cross-elasticity model. The ability to examine such relationships across vast product ranges creates opportunities that conventional pricing teams could scarcely calculate in real time.

There is a distinction between dynamic pricing and surveillance pricing. Airlines and hotels adjust prices according to demand, capacity and timing. The more contentious frontier is using information about a particular consumer or that consumer’s behaviour to influence the price or offer they see. In 2025, the US Federal Trade Commission reported that pricing intermediaries could use granular data, including browsing and shopping behaviour, to tailor prices or promotions.

Should shoppers pay for intelligence

The commercial case for better information is compelling. Retailers can forecast demand, reduce waste, improve promotions, and protect margins. An algorithm might lower the price of pasta sauce when pasta enters the basket, giving the customer a useful discount while increasing sales. Consumers may welcome that intelligence.

But, reverse the incentive and the proposition changes. If the basket reveals that the shopper is unlikely to abandon a planned meal over the final ingredient, should that knowledge justify extracting a little more? The technology may be clever. The customer may call it something else.

This is where the old controversy over tied selling, sometimes called “marrying goods”, becomes unexpectedly relevant. Traditional tying involves conditioning the purchase of one product on buying another. Complementary algorithmic pricing need not impose any such condition. The consumer has voluntarily assembled the basket; the system simply learns from it.

Competition authorities across very different economies already recognise the risks of tying. In the United States, the Federal Trade Commission (FTC) and Department of Justice enforce competition laws; the FTC is also examining surveillance pricing. Britain’s Competition and Markets Authority scrutinises anticompetitive tying and bundling. The European Union adds a distinct transparency requirement: traders must inform consumers when prices are personalised through automated decision-making and profiling.

Global regulators

This is not simply a divide between wealthy and developing economies. India’s Competition Commission examines tie-in arrangements under its competition framework. Brazil’s competition authority, CADE, recognises venda casada, or tied selling. South Africa’s competition regime scrutinises certain bundling and discriminatory-pricing conduct, while Jamaica’s Fair Trading Commission addresses tied selling under its Fair Competition Act. The legal tests differ; the concern about using commercial power to constrain choice is widely shared.

But here is the emerging regulatory challenge. Those frameworks know how to ask whether a seller has forced two products together. What happens when the seller does not force anything, but can infer from the customer’s choices how urgently the next item is wanted? The goods have not been married by the retailer. The customer’s behaviour has introduced them.

The question therefore shifts beyond tying. Should regulators examine the price itself, the data used to set it, the explanation given to the consumer, or the market power that makes such pricing effective? A discount based on basket contents and a surcharge based on inferred need may use similar analytical machinery while producing very different judgments about fairness.

For businesses operating across borders, that difference matters. A pricing model designed for one market may encounter different disclosure rules, competition tests and consumer expectations elsewhere. Executives should understand not only whether an algorithm improves revenue per basket, but whether its logic can be explained and defended in every market where it operates.

There is commercial value in trust. A pricing algorithm may calculate the maximum return from a single transaction while overlooking the lifetime value lost when a customer believes the retailer exploited a moment of need. The strongest competitive advantage may belong not to the business with the smartest algorithm, but to the one that knows where not to use it.

The shopping cart has always told retailers what we buy. Increasingly, it can tell them what we are trying to do. Once a business can price the purpose behind the purchase, the question is no longer only what technology permits. It is what customers, competitors and regulators will consider fair.

Dr Charlene Ashley is an International Business Strategist / Organisational Behaviour Consultant & Marketing Strategist. Email: cashley@theconsultancyinc.com