Forbes Names Columbia Data Scientist to its 30 Under 30 List
The process of making a product and then selling it to the public is usually handled independently within a company. But as more people shop online, companies are experimenting with ways to manage supply chain costs and grow sales at the same time.
“How can we come up with models to maximize profits across both of these functions?” said Adam Elmachtoub, an operations researcher at Columbia Engineering and a member of the Data Science Institute. Elmachtoub was recently named a 30 Under 30 innovator in science by Forbes magazine.
Companies typically cut prices to increase sales of slow-moving products, but this approach can backfire by alienating customers who paid full price. The growth of online retail has led to a way around this. The rise of so-called opaque products allows companies to reduce inventory without annoying their best customers. Customers have a choice: pay full price for the product as specified, or pay slightly less for an opaque product whose brand, color, or other attribute is hidden until the product arrives at their door.
A popular way of selling high-end hotel rooms and flights, the opaque marketing approach is starting to enter the world of retail goods. In an ongoing study, "Retailing with Opaque Products", Elmachtoub and his colleague show that reducing inventory with opaque products produces meaningful savings; if as few as 10 percent of customers purchase opaque products, companies could trim costs by 3 percent, a substantial amount in an industry with notoriously slim profit margins.
“Companies are starting to realize that managing inventory and consumer demand together can have large benefits,” said Elmachtoub, who joined the Engineering School in August 2015. “As competition increases with Internet sales, companies are looking for new ways to make and sell their products more efficiently.”
Elmachtoub became interested in operations research, or OR, as researchers in the field call it, as an undergraduate at Cornell University. He liked the idea of solving real-world problems with mathematical models. In one project, he and his colleagues searched for the most cost-effective way to save the endangered Red-cockaded Woodpecker’s habitat. They came up with four land-purchase options, ranging from $150 million to $400 million, which would allow the federal government to join together pieces of fragmented habitat in the southeastern United States.
For his PhD thesis at MIT, Elmachtoub developed a tool to help manufacturing companies decide whether to take a customer’s order. Each additional order brings in more revenue, but those benefits can be easily negated by complex supply chain costs. In a paper to be published in Operations Research, Elmachtoub developed two algorithms that allow manufacturers to quickly calculate which orders will be most profitable to fulfill.
After graduating from MIT, Elmachtoub worked at IBM, in Yorktown Heights, N.Y. There, he and his colleagues developed a tool for pitching promotions to prospective airline customers as they shopped online for tickets. By analyzing customers’ past preferences, the algorithm served up promotions designed to clinch the deal—free baggage for some customers, access to the private lounge for others. The work is ongoing, but he estimates their approach could raise revenues by 2 to 3 percent.
In another ongoing project, Elmachtoub is using data about IBM’s corporate clients to pinpoint which ones are most likely to buy IBM servers. This information can help IBM’s limited sales staff prioritize sales calls. The underlying algorithms also take into account inventory, including servers that may no longer be in production. “The key is to manage inventory availability and customer satisfaction while trying to maximize overall revenue,” he said. “We’ve come up with an effective way to do this.”
BS, Cornell University (2009); PhD, MIT (2014)
—by Kim Martineau