The largest supermarket chain in the United States, Cincinnati-based Kroger, is using data and analytics to improve operations, cut costs, and increase customer satisfaction.
The grocer began using analytics to evaluate and improve operations in 2007, when it put together its first operations research team. The team consisted of mathematicians, MBAs, former Marine helicopter pilots, and a guy who used to shoot down missiles for a living. Today, the team is focused on how to move people through stores faster, improve customer service, and optimize warehouse layouts.
In 2010, the team implemented a project aimed at in-store pharmacies that would cut down on out-of-stock items, improve the customer experience, save money, and increase revenue. To date, they have reduced inventory by $120 million and the number of out-of-stock prescriptions by 1.7 million.
“So that’s 1.7 million more customers walking out the door with a script,” Doug Meiser, Kroger’s Operations Research manager, said at the University of Cincinnati’s Analytics Summit 2014 in May.
This effort has realized $10 million in annual savings and $80 million in increased revenue.
But perhaps the group’s biggest breakthrough is QueVision, a data analytics package that combines historical shopping data with real-time information about the number of customers in the store and infrared camera technology that counts the number of people waiting in line. Once certain thresholds are reached, managers are alerted to open new checkout lanes.
The project started because of a single line item in a project about how they could scan items faster at checkout.
“We put a little flag in the model that said ‘dynamic lane planning,’ ” said Meiser. “It was more of a curiosity than anything. What would happen to our labor if we could open up a lane exactly when we needed it? And how would that impact customer service and labor?”
The answer to labor question was: Not much. Using QueVision to determine when to open checkout lanes did not impact their employees’ ability to do their jobs effectively. The answer to the customer-service question was: A lot.
“By being able to do that prediction, we can get our customer wait times down to 30 seconds [from four minutes],” said Jim Holtman, the former missile-man turned operations analyst.
Once management saw the benefit of the technology, it was very quickly rolled out to all 2,400 of the company’s stores.
While improving customer wait times is important, managing the back-end operations that move products from warehouse to store is a far more daunting challenge. The company has been working for years on how to design its warehouses so products can be picked and palletized in a way that is logical at the store level.
Except for a few recently optimized facilities, distribution centers are currently designed in a way that is warehouse-centric. As a result, a single pallet could have items on it that go into many different areas of a store. Compounding the problem is the fact that each distribution center is unique and that not all stores share the same layout.
“The basic cause of the problem was local optimization – how to put the pallets together neatly,” said Holtman. “So I put a case of corn, dog food, and baby products on the same pallet … And the stores were laid out in the way the customer shopped.”
The way customers shop is by looking for related items, not related sizes. So corn, dog food, and baby products frequently will not be shelved near one another in the store.
A typical store receives between 15 and 20 pallets per day. Multiply this by 2,400, and the scope of the problem becomes clear.
“What we really want to look at is the relationship between the products in the store – how far apart are they?” said Holtman. “This is where cluster analysis comes in very handy. It helped us figure out how things need to be grouped.”
Holtman has worked with around 20 different distribution centers, each different from the others. In only one instance was he able to work with a warehouse under construction. Most of the time they are trying to optimize a warehouse’s layout while it is in operation. This is where the real challenge lies, he said.
“The analysis is easy,” he said. “A lot of this stuff is just math. I know from experience the math I used to do large-scale [missile intercepts] works very well to redo distribution centers.”
Now a freelance writer, in a former, not-too-distant life, Allen Bernard was the managing editor of CIOUpdate.com and numerous other technology websites. Since 2000, Allen has written, assigned and edited thousands of articles that focus on intersection of technology and business. As well as content marketing and PR, he now writes for Data Informed.com, Ziff Davis B2B, CIO.com, the Economist Intelligence Unit and other high-quality publications. Originally from the Boston area, Allen now calls Columbus, Ohio, home. He can be reached at 614-937-2316 or email@example.com. Please follow him on Twitter at @allen_bernard1, on Google+ or on Linked In.