Researchers claim that machines that learn the responses to various mall promotions can optimize marketing budgets far more effectively than human intuition. The researchers also suggest that spending more during off-peak shopping periods may generate a better return than during the holiday marketing campaigns. They analyzed marketing campaigns and two years of traffic from 25 enclosed malls in China. They measured traffic using sensors embedded with artificial intelligence chips. Marketing campaigns guided by machine learning yielded an 11.2% increase in customer traffic, compared with the 3.2% bump that more traditional models yielded.

Commerce + Communities Today contributing editor Joe Gose talked with Tong Wang, co-author of Evaluating the Effectiveness of Marketing Campaigns for Malls Using a Novel Interpretable Machine Learning Model, which soon appear in the peer-reviewed journal Information Systems Research. She’s also an assistant professor of business analytics at the University of Iowa’s Tippie College of Business.

What sparked your interest in shopping mall marketing campaigns, particularly given your background in electrical engineering and machine learning research?

I’m generally interested in applying machine learning to real-world problems. Unlike department stores, which care about sales, the only source of mall revenue comes from the rents that mall owners collect, so the more customers malls can attract, the higher the rental rates they can charge. But with the increasing competition from online businesses, malls have seen declines in customer traffic. In addition to that, mall owners must arrange and coordinate different campaigns and promotional events to create a better shopping experience for customers, especially during holidays when people want to go out and shop. But the problem is deciding on what campaigns to host and how much to spend. Today, those decisions are made by humans based on human experience. Because malls have accumulated all this traffic data from sensors over the years, we decided to study it to see if we could come up with a better plan.

What were your findings?

Sales-based campaigns encourage tenants to provide discounts, while others create experience incentives for shoppers. We’ve found that it is more beneficial for shopping malls to spend money on the experience-only incentive campaigns. Plus, mall owners invest a lot of money in campaigns during holidays, but traffic is already high in these peak periods so if malls spend a lot of money during this time, the return is small. On the other hand, if they spend a little bit of money during off-peak periods where there is large room for improvement, it will be more beneficial. It especially makes sense because nowadays customers find it very easy to shop online, so if consumers want to go to a shopping mall, there has to be something to attract them that online shopping does not offer. That is the experience.

The malls you studied were in China. Are there any cultural differences that may not translate to U.S. malls?

We trained our machine learning model in medium to large enclosed shopping malls with 100 to 200 tenants, movie theaters and restaurants, which would be comparable to big shopping malls here. China’s peak shopping period is during the Spring Festival [the Chinese New Year]. We also included Christmas, but it’s not as big as it is in the U.S. While the timing of peak periods is different and some holidays in China don’t exist in the U.S., our findings are based on the general definition of a holiday, which is when people have time to go to a mall. We also looked at online promotional campaigns in June and November, when platforms like Alibaba and others give big discounts on many products in China. In the U.S., we have similar online sales promotions. These events are big competitors with shopping malls, and if malls really want to compete during these periods, they have to spend a lot of money to attract customers. Still, we found that malls can actually piggyback on the online events because people are in the mood to go shopping. Malls just have to be more strategic about how they design their campaigns.

Have you received any feedback regarding your results and model?

We haven’t employed the exact model anywhere, but we have talked to the owners of malls who said that our findings are consistent with some of their own results. Over the years and especially as a result of the pandemic, they have invested more on experience-driven campaigns, just like we have recommended. During the summer when kids are out of school and have a lot of time, for example, some malls have designed activities and campaigns to attract that specific customer subset.