This is a familiar experience for many of us: A glitzy shopping mall has recently opened up in town, carrying many of your favorite brands. In the hopes of receiving recommendations tailored to your specific needs, you sign up for its membership program and even share personal details like your email, income and occupation. Bad mistake. Before long, your inbox is inundated with a deluge of promotional materials and newsletters of products that you’d never buy.
How is it possible, in this technologically driven age, that customized service is so poorly delivered? The answer, according to Ted Chung, C.E.O. of Hyundai Card, lies with a concept called market segmentation.
“Market segmentation is dividing one million people into five boxes and expecting the same response from each box,” he says, adding that when companies use market segmentation to deliver services, promotional material and recommendations to their target customers, the response rate hovers at around 1 to 3 percent — 5 percent would be considered a “super hit.” “It’s designed for failing less, not success,” he says.
At the IBM Think business technology conference in 2019, Chung introduced an approach called “super customization” that leverages artificial intelligence and credit card data. Hyundai Card’s D-tag is an A.I.-based machine-learning package of thousands of data points. These not only include details like age and occupation, but also consumption preferences and spending habits, information gleaned when customers use their credit cards. Furthermore, the company is striving to optimize data management and operations with the formation of a D-tag Committee led by Chung.
Combined with real-time information — such as weather, location and time — the A.I. is able to ascertain the contexts that surround particular patterns of behavior, which in turn allows it to provide more timely and relevant recommendations.
For instance, the A.I. learns that on a hot day, individual A prefers drinking a refreshing beer, while B would opt for an ice cream. Super customization ensures that on sweltering summer days A would receive discount coupons to use at the nearest pub, while B might get ice-cream stand recommendations.
Recommendations are optimized and served to customers through the channels they use most — whether text message, app alert or email — at the most appropriate time. When customers take up a recommendation and use their credit cards, the data is logged in a prediction calendar, and the optimization cycle repeats.
This method is especially effective in a country like South Korea, where the credit card utilization rate is around 80 percent, and cards are used for transactions small and large. While global giants from Facebook to Netflix are developing new personalized services at an alarming rate, it is Korean financial companies such as Hyundai Card that are creating a new trend with credit card data. This is because credit cards contain offline data, as well as online spending habits, making them the most powerful means of understanding customer behavior and tastes.
“We're using these new techniques in our beta tests, and we are getting amazing results,” says Chung. Hyundai Card has found that super customization leads to results up to five times more often than when the market-segmentation approach was used.
“Super customization does not make you a prisoner of a segmentation box,” Chung explains. “We don't serve the box, we serve you. We know you better, we know your taste, your style.”
But as A.I. gets to know us better and drives services that are increasingly customized, will we actually lose the human touch in personal service?
Not so, according to Chung. “A.I. is replacing human beings’ limits, it's not replacing the human touch,” he says. “Actually, you feel more human care, because we are there when you need us.”
ⓒThe New York Times. All rights reserved.