Why AI Pricing Broke Down in Grocery Retail

When Instacart shut down its AI-driven price tests last week, the company said it was responding to concerns raised by consumers, advocates, and regulators.

The experiments were designed to allow retailers to test different item prices on Instacart’s platform using Eversight, a pricing technology company Instacart acquired in 2022.

The tests drew public attention after an investigation by Consumer Reports and the Groundwork Collaborative found that shoppers were sometimes charged different prices for the same grocery item at the same store. In some instances, the reported price differences exceeded 20 percent. Consumer Reports said the findings were based on data collected from hundreds of volunteer shoppers across multiple retailers.

Following the report, consumer advocacy groups criticized the lack of disclosure around the pricing tests. Members of Congress called for oversight. The Federal Trade Commission opened an inquiry into Instacart’s pricing practices. Instacart said the tests had “missed the mark” and announced it would end all item-level price experimentation on its platform.

Much of the reaction focused on artificial intelligence. But the controversy is more about how price experimentation intersected with consumer expectations around grocery shopping.

Grocery pricing and the expectations consumers bring

Dynamic and algorithmic pricing are widely used in other sectors. Airlines adjust fares continuously. Hotel rates change based on demand, timing, and availability. Ride-hailing platforms vary prices through surge pricing. In each case, consumers are generally aware that prices fluctuate and that variability is part of the transaction.

Grocery shopping has historically operated differently. Food purchases are frequent and essential, and prices are often compared closely across households and neighborhoods. Consumer advocates and regulators have long treated grocery pricing as an area where consistency and transparency matter, particularly because of its impact on household budgets.

That context shaped the response to Instacart’s pricing tests. Although Instacart has said retailers ultimately set prices and that the tests did not rely on personal or demographic data, shoppers experienced different outcomes for identical items without being informed that price testing was occurring.

Price experimentation itself is not new in grocery retail. Retailers routinely test promotions, discounts, and coupons. What distinguished Instacart’s tests was that the experimentation occurred at the item level and was not disclosed to shoppers in advance.

Similar reactions have occurred in other sectors when pricing algorithms intersect with periods of stress or limited choice. Ride-hailing companies, for example, have faced public and regulatory scrutiny when surge pricing has been applied during emergencies or extreme weather events, even though the underlying pricing mechanisms were already well known.

In each case, the backlash has focused less on the existence of algorithms and more on when and how they are applied.

When AI affects prices, scrutiny follows

The response to Instacart’s pricing tests reflects broader regulatory attention on algorithmic pricing. The FTC has previously examined the use of automated pricing tools by large platforms, including whether such systems can harm consumers or competition when deployed without transparency.

In Instacart’s case, regulators and consumer advocates treated the pricing tests as a consumer protection issue rather than a product innovation. The FTC’s inquiry signaled that algorithmic pricing in essential consumer categories is likely to face closer examination, particularly when experimentation is not disclosed.

At the same time, Instacart has continued to expand its use of AI in other parts of its business. The company has invested in machine learning for search, product recommendations, advertising, and in-store technologies, areas that have not drawn comparable criticism. These applications influence shopping behavior but do not directly change the base price of goods.

Industry analysts and market observers have noted that this distinction matters. AI systems that optimize logistics, ads, or recommendations generally operate in the background. Pricing, by contrast, is the most visible and sensitive point of interaction between a retailer and a consumer.

As a result, retailers appear to be approaching pricing-related AI with greater caution than other applications. There is little evidence that companies are abandoning AI-driven optimization altogether. Instead, public statements and product roadmaps suggest a focus on areas where algorithmic decision-making is less likely to trigger fairness concerns.

Instacart’s pricing tests did not establish new legal standards on their own. But they did demonstrate how quickly undisclosed price experimentation can escalate into regulatory scrutiny when it involves everyday consumer goods.

Whether retailers will revisit item-level price testing with clearer disclosures, or whether regulators will impose new guardrails around such practices, remains an open question. What is clear from the response is that when AI systems influence prices, they draw a level of attention and scrutiny that other retail technologies do not.