Harps Food Stores is using artificial intelligence (AI) software to improve and optimize decision-making around pricing and promotions by studying sales data from years past.
The 87-store chain launched a partnership with Toronto-based Daisy Intelligence in early 2017 to help it review which products should be featured in its circulars. Harps, which does not have a frequent shopper card program, provided Daisy Intelligence with two years’ worth of information from its T-log. “That is all of the transactional information across all of our stores,” said David Ganoung, vice president of marketing at Springdale, Ark.-based Harps. He added that Harps also gave Daisy Intelligence pricing information including ad items or items that may have been an in-store special.
“We are now getting ready to hop into price optimization and forecasting where they can go back and tell us what in a future ad will be our best item and allow us to target those with our manufacturers to get the best deal,” Ganoung said.
“Daisy Intelligence now gives us the ability to see all of the associated sales related to every item,” Ganoung said. “We can look across all transactions and see what is the impact of that one item in the basket.”
“Our goal is to help retailers decide what products they should promote, what prices they should charge and how much inventory they should allocate,” said Gary Saarenvirta, CEO of Daisy Intelligence. “Those are the three core decisions that retailers make every week.”
Daisy’s AI tells Harps what were the baseline sales of an item in an ad, along with the incremental lift the chain got by promoting it, along with associated sales that were driven by having the item in the ad, and what other like items the featured item cannibalized.
“It tells us the net promotional effectiveness of that item,” Ganoung said. “It helps us to score, or rank, our ad blocks on what they did to drive sales.”
When working with a retailer, Daisy Intelligence will study several years’ worth of sales, examining every single transaction receipt. “We know what products are bought together and how frequently. We know about seasonality and purchasing habits,” Saarenvirta said. “We are taking into account literally millions of attributes about how consumers purchase products. Our objective is to double a grocer’s profits. If we can grow their sales by more than 3% we can double their net margin, and we’ve been able to do that with some core clients.”
If a retailer has a frequent shopper program, Daisy Intelligence can look into the consumer’s pantry, refrigerator and freezer and see patterns that span years of product usage. “Even without that we’re able to double a retailer’s profitability even without knowing their customer,” said Saarenvirta (left).
Retailers sign a one-year contract and then pay Daisy Intelligence a monthly fee. “It takes us one or two months to get up and running and then in month three we start delivering decisions and recommendations,” Saarenvirta said. “We typically work with the category managers and we have client teams that work with the client to make sure they are getting the most value out of our recommendations.”
In addition to Harps, Daisy is working with Walmart Canada and several other retailers in its native Canada, along with Earth Fare and four other clients in the United States.
“Our system recommends a decision and the retailer can follow that decision or override it if they feel they know something that our system didn’t,” Saarenvirta noted.
Because Harps buys off a promotional calendar from its warehouse, Associated Wholesale Grocers, there are some “built-in handcuffs” that have kept Daisy’s AI from being 100% optimized, Ganoung said. However, the information has already greatly helped Harps’ buyers.
“At their fingertips, our buyers have the ability to type in a UPC and they can see how elastic the pricing of that item is and how it impacts movement,” Ganoung said. “They can see the associated lift by item. We can’t wait for this to take us to the next level.”
“The application of AI in improving the promotional impact is a natural,” said Bill Bishop, chief architect of Brick Meets Click, a consulting firm based in Barrington, Ill. “They can tell if people have a tendency to buy ahead, so if the transaction shows a couple of purchases, but one of them was borrowed from three weeks in the future, they account for that. There is a high likelihood that this kind of thing will be the way that ads are done in the future.”