A recent Consumer Reports investigation showed that Instacart was running AI price experiments in the app, charging some consumers as much as 23% more than others for the same goods. While Instacart disputed some of the findings initially, the grocery delivery company announced on Monday that it will end all AI price experiments. Buyers may continue to see price differences for the same products between different retailers, similar to what happens in brick-and-mortar stores, but Instacart won’t use AI to modify the prices users see in the app when ordering.
“We’ve listened carefully to feedback from our customers. And we understand that the tests we ran with a small number of retail partners that resulted in different prices for the same item at the same store missed the mark for some customers,” the company said in a blog post on Monday. “At a time when families are working exceptionally hard to stretch every grocery dollar, those tests raised concerns, leaving some people questioning the prices they see on Instacart. That’s not okay -– especially for a company built on trust, transparency, and affordability.”
Instacart said that “effective immediately,” all item price tests will conclude. Retailers will not have access to the Eversight AI technology that was used for the price experiment. “Now, if two families are shopping for the same items, at the same time, from the same store location on Instacart, they see the same prices — period,” Instacart said.
What Consumer Reports found
Consumer Reports ran a monthlong investigation in partnership with Groundwork Collaborative and More Perfect Union to see whether Instacart was using algorithmic prices, a strategy that a regular shopper would not be able to determine on their own. Consumer Reports selected 437 volunteers, dividing them into several groups to run Instacart order simulations, and test in-store prices for the same goods. The consumer advocacy group found that every volunteer was an “unwitting participant” in the AI pricing experiments.
Consumer Reports said Instacart offers as many as five different prices for the same product, with variations ranging from $0.07 to $2.56. Even small price variations “could add up to big differences” in the total cost of groceries. “The price of the same basket of food at a Seattle-area Safeway on the Instacart platform, for example, ranged from $114.34 to $123.93 — roughly a $10 difference,” Consumer Reports wrote. “And only 8 percent of those shoppers got the lowest basket total.” Put differently, Consumer Reports said that the average price variations observed in its experiment would translate to paying $1,200 extra on groceries a year. The figure is based on Instacart’s typical spend on groceries for a family of four.
Consumer Reports also discovered another troubling pricing strategy during the experiment. Instacart showed buyers different “original” prices for the same discounted product. This made the savings appear larger or smaller, potentially affecting one’s purchase decision. “For example, most volunteers shopping on Instacart at a Safeway in Seattle were shown original prices for Premium brand saltine crackers of $5.93, $5.99, or $6.69, while the final sale price was the same for everyone — $3.99,” Consumer Reports noted.
Did Instacart use AI to enable dynamic pricing?
Following Instacart’s announcement that AI price experiments would end, Consumer Reports also detailed the “blowback from U.S. consumers, regulators, and politicians,” over the initial Instacart shopping experiment, including a Reuters report that said the FTC had launched an investigation. While the FTC has not confirmed the action, Instacart has offered some details on how its AI price test worked.
In the same blog post announcing the end of the experiment, Instacart said it had not used “dynamic pricing or surveillance pricing,” which were some of the worries laid out in the Consumer Reports investigation. Dynamic pricing refers to using algorithms to set different prices for the same goods, like airline tickets, hotel rooms, and rideshares. It’s a technology consumers accept for those types of products, but not food.
Surveillance pricing is a strategy that relies on using personal user data, like one’s shopping history, to set unique price points and maximize profits. Instacart said the prices “were never based on supply or demand, personal data, demographics, or individual shopping behavior.” Earlier this year, Delta Air Lines came under fire after remarks from the company that it would use AI for airfare prices. In response to backlash, Delta said that it’s not using personal data for its AI prices.
