All the sessions from Transform 2021 are available on-demand now. Watch now.


Say you have a product at a store you want customers to be able to grab off the shelf and buy. How do you balance that with selling the same inventory to people ordering online? How do you decide how much of your inventory you sell online? When someone makes an order online, how do you decide whether to fulfill the order with inventory from stores or your company’s warehouses?

These are just some of the many questions that retail giant Kohl’s wrestled with. The answer the retailer came to, according to Paul Gaffney, Kohl’s chief technology and supply chain officer, was to let AI take a shot at the decision-making.

“When you start allowing machine learning algorithms to make decisions, they sometimes make decisions that aren’t intuitive. They aren’t what the people would make,” Gaffney said.

AI makes a decision

Usually, the deciding factor when trying to pick where to ship from would be shipping costs, Gaffney said at VentureBeat’s Transform 2021 virtual summit. However, it also became clear to the company that when an item was left in inventory at a location where it takes longer to sell, it will eventually wind up being marked down, and that would hurt the bottom line.

“We had this nagging suspicion we were incurring more markdowns than we needed to. Could we be smarter and say, ‘Hey, how about if we sell the merchandise that we might have placed months ago in a spot where we now know it’s probably not going to sell in that store … so let’s pick it from that store and avoid the future markdown,'” Gaffney said.

Kohl’s turned to partners to develop solutions for their supply chain optimization. Then came the leap of faith.

“What opened a bunch of doors for us was the willingness to say, ‘OK, we’re willing to risk a certain amount of money in the belief in the algorithm, and even if it doesn’t work, that investment in learning was good enough,'” Gaffney said. “And it turned out that it paid off.”

With successes in hand, Kohl’s is reflecting on its usage of AI, developing their in-house capability to exercise more control over their AI tools, and also considering further ways to optimize their stores beyond backend inventory management. For example, the data showed that each store has a different make up of customers, so the AI decides what kind of things to display to account for the different group of customers. Allowing the algorithm to suggest making changes to the products on sale at different stores based on customer data, resulted in “enormous positive upside,” said Gaffney.

Human experience

People should “educate themselves” on what machine learning can do, but also to understand how these advanced technologies can disrupt people’s work patterns. Enterprises need to think about ways to “purposefully re-engage” people in activities that aren’t conducive to machine learning.

“It’s tempting to treat the adoption of machine learning AI and big data as a technical problem,” Gaffney said. “But it is much more so a human change management problem as well.”

VentureBeat

VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative technology and transact. Our site delivers essential information on data technologies and strategies to guide you as you lead your organizations. We invite you to become a member of our community, to access:
  • up-to-date information on the subjects of interest to you
  • our newsletters
  • gated thought-leader content and discounted access to our prized events, such as Transform 2021: Learn More
  • networking features, and more
Become a member