According to Sam’s Club officials, their model uses a variety of data. Things like local temperature (warmer weather often means fewer pies bought); whether a Sunday football game is home or away (home games may require more pie); how popular pecan pies are this year (more pecan pies can translate into fewer pumpkin pie sales).
Those data points and others are then plugged into an artificial intelligence model of their own. It gives recommendations to each store manager, such as how many pies their store should have per hour. Last year, Sam’s Club sold enough pumpkin pie to fill 450 football fields, officials said. (They declined to give an exact figure.)
Demand needs to be forecasted with precision because competition to retain customers is fierce and profit margins are tight, officials added.
“If members can’t get what they need, they’re not going to renew with us,” said Pete Rowe, Sam’s Club vice president of technology and store associate, whose family bought both pumpkin and pecan pie for Thanksgiving this year. “It’s very important for us and our model to be sure.”
Pumpkin pie or pecan pie? You don’t have to choose with these recipes.
In recent years, sophisticated AI models have become commonplace in grocery stores. With the pandemic and supply chain challenges, it is rapidly changing the grocery shopping experience: from AI-powered shopping carts that automatically recognize the items you pick up to chef bots that create recipes based on your purchases.
According to grocery experts, the increase is due to a combination of factors. Stores already have access to a wealth of data, including third-party brokers and shopper loyalty programs. Computer processing power is cheaper and faster. Machine learning models, software that computers use to learn and adapt themselves, have evolved. The pandemic played a big role.
Retail and Technology Center CEO Gary Hawkins said that in the pre-pandemic era, stores used software to manage inventory, staff and predict when goods would be in stock. But after the pandemic hit, “supply chains exploded, demand went through the roof,” and grocery stores were unprepared and needed smarter systems, Hawkins said.
“It literally blew up all the models because they weren’t sophisticated enough,” he said. “Pretty quickly, especially the older guys, they said, ‘We need something better here.’ “
In April 2019, Walmart launched an Intelligence Research Lab, where cameras and sensors connect to algorithms to track how shelves are stocked. In March, Kroger launched an AI lab where technology can monitor vegetable freshness. Ketchup maker Kraft Heinz is now using machine learning to track demand for its products leading up to events like the Super Bowl. Amazon opened a fully automated Whole Foods this year that uses deep learning software to let customers shop and check out without the need for a cashier. (Amazon founder Jeff Bezos owns The Washington Post).
Startups have also increased. New York-based Caper Cart makes AI-powered shopping carts that automatically recognize what customers are picking up and checking out. Seattle’s Shelf Engine tells stores how much product they need daily. Hivery, based in Australia, has a model for advising grocers on where to put products on the shelves.
“AI is advancing into almost every technological capability,” Hawkins said.
This World Cup is controlled and powered by artificial intelligence
Dominic D’Agostino, a 30-year member of Sam’s Club in Dayton, Ohio, said he didn’t know the company used such sophisticated technology to forecast demand for pumpkin pie.
Although she’s not a fan of food and probably won’t bring anything to her sister’s house for the holidays — “the only pie I really like is pizza,” she said — D’Agostino is intrigued, and a little worried, that AI is being used this way.
“It’s creepy,” she said. “He’s also fascinating.”
Sam’s Club decided to use AI shortly before the pandemic, Rowe said. The chain used software to guide its operations, but felt it could be better.
In past years, for example, Rowe said, “We’re going to make a lot of pumpkin pies, a lot of croissants, and [would lead] We have to waste our colleagues’ time, and we also have to throw away inventory.
Now, the company uses machine learning to predict inventory for everything they make at home, like pies and chicken. They also have “autonomous floor cleaners,” or self-driving robots, to scan shelves and send alerts about which items need to be restocked first when delivery trucks arrive.
Rowe said this has helped the store be more than 90 percent accurate in forecasting demand, and he wants it to be even higher.
A Google engineer who thinks the company’s artificial intelligence is coming to life
Despite the appeal of artificial intelligence, it has its risks. Researchers at the University of Arkansas said the algorithms are avoiding the bulk of customer data, raising privacy risks. This can also lead to bias.
“Even without formal access to an AI algorithm for race or gender,” they wrote, “an AI application can infer race/gender from other data and use that to give ‘higher value’ to specific demographics.”
Others point out that AI is not a one-size-fits-all solution, and stores may spend money on fancy software to keep up with the hype.
“You can’t be too impressed with the shiny object element of AI,” said Mike Hanrahan, former CEO of Walmart’s Intelligence Research Lab, in a tech publication. “There are a lot of shiny objects out there doing things that we think are unrealistic in scale and probably not long-term, consumer-friendly.”