What Online Retailers Get From Recommendation Lists

With online shopping, it’s easy to fall down the rabbit hole of checking out recommended items. And if you shop on your phone, don’t be surprised if you end up buying more, says new research from Maryland Smith.

When it comes to the use of recommendation systems, retailers find that consumer outcomes such as overall sales, click-through rates and conversion increase. And, between mobile and PC-based shopping, the impact of those same systems are much greater for mobile shoppers, says Maryland Smith’s Anandasivam Gopal.

Retailers might see smartphones as just another device and expect similar figures from each platform. But that’s not the case, says Gopal, and a lot of it has to do with shopping on the go.

“What we found was that the recommendation system was actually a lot more influential on mobile sales with people buying more recommended products,” says Gopal, the Dean's Professor of Information Systems at the University of Maryland’s Robert H. Smith School of Business. “A lot of the shopping was being done when people were stuck in traffic, commuting to work or sitting in class when they had their phone with them and it was convenient to shop.”

The research, part of a dissertation project from Dongwon Lee, PhD ’17, and subsequently published in Information Systems Research, analyzed findings from randomized field experiments conducted in partnership with an online retailing firm in South Korea.

In addition to determining whether recommendation systems had a significant impact on mobile retail sales, the researchers also examined whether sales became more diversified by product categories.

While mobile portability and convenience were found to impact sales figures, they were less of a factor in purchase diversity, Gopal says. In fact, a consumer’s surroundings may lead to reductions in that regard, he says.

“In some cases, recommendation systems interestingly have been shown to create these islands where people buy the same things as their friends, peers or colleagues,” says Gopal. “What results is an echo chamber of specific types of products.”

The lower diversity in sales, however, shouldn’t dissuade online retailers from implementing recommendation systems for their mobile-based customers, says Gopal, and neither should perceptions of smaller screen sizes and limited functionality compared to computer browsers. A regular desktop might be able to show five to eight different products at a time compared to four on mobile devices, but those four are still incredibly impactful, he says.

“It’s worth investing in a recommendation system and creating an appropriate mobile browser functionality for your business. There isn’t just a prospect of increased sales”, Gopal says, “mobile data gives retailers the opportunity to understand consumer behavior on a more personalized level since all such browsing behavior can be captured on the phone – with details like time spent looking at a product, product price ranges, as well as how customers got to the page and where they went next. Since the phone is so personal and ubiquitous, the data trail is likely to be highly informative for retailers.”

“We show that even the baseline recommendation system is actually very useful for retailers launching an app because they are providing people access to their product line in a more informative manner,” says Gopal. “The idea that mobile devices are too small to provide recommendations and that people won’t pay attention is misleading because they certainly are paying attention.”

Read the full research article, “Different but Equal? A Field Experiment on the Impact of Recommendation Systems on Mobile and Personal Computer Channels in Retail,” in Information Systems Research.