The Perfect Match: Personalised Shopping Recommendations for Mobile Devices

LEE, Dongwon | GOPAL, Anandasivam | PARK, Sung-Hyuk

Few people leave home without their smartphones or other mobile devices, which now dominate almost every aspect of our personal livesincluding shopping. We already know that purchase behavior differs between mobile- and PC-based shopping, but the exciting potential to optimise recommendation systems for mobile channels has been largely untapped—until now. In a timely new study, HKUST’s Dongwon Lee and colleagues reveal the unique effects of personalised shopping recommendations on the browsing and buying behavior of mobile phone and PC users.

In the competitive world of e-commerce, shopping suggestions that match users’ interests can be the “nudge” they need to buy a product. In short, online recommendation systems boost sales. “Indeed,” note the authors, “76% of retail websites list personalised product recommendations as a priority for their online sales strategy.” However, most related research has focused on computer desktop users. As more and more online shoppers swap their PCs for their mobile phones, retailers need to know how to design recommendation systems for mobile channels.

Rising to this challenge, the researchers examined the shopping behavior of a cohort of 11,623 mobile users and 2,567 PC users who visited an online shop selling cosmetics, accessories, and clothes. On the shop’s landing page, half of the shoppers were shown personalised recommendations, while the other half saw only a generic list of best-selling items. This allowed the authors to explore how recommendations affected product views and sales, as well as the sales of niche products, for both mobile and PC users.

Overall, the recommended products were more popular than the generic products in terms of both views and sales. Interestingly, the recommendation system also made the shoppers more likely to view—but not to purchase—niche products. Armed with this information, say the researchers, retailers can better devise strategies to convert views into purchases, such as “the use of discount coupons or sales promotions specifically aimed at niche, recommended products.”

Strikingly, the sales-enhancing effect of the recommendation system was more apparent among the mobile users than the PC users. “Mobile users appear to be more responsive to recommendations, show more product views, and higher rates of clickthrough and conversion as a result,” explain the authors. Mobile users were also more likely to purchase niche or new products. Compared with PC users, “mobile users tend to be more diverse in their purchase behaviour—a contingent channel-specific result new to the literature,” say the researchers.

In an era of smartphones, these novel findings could help retailers reap the benefits of personalised recommendations by designing algorithms that are customised for mobile devices. Recommendation systems should exploit the “any time, anywhere” usability of mobile devices while addressing their shortcomings, such as small, low-resolution screens, connectivity problems, and limited processing power. There is also tremendous scope to apply these results to other mobile ecosystems in the world of e-commerce, such as advertising, promotion, and targeted sales.

LEE, Dongwon

Assistant Professor
Information Systems, Business Statistics & Operations Management