Add to Cart: UBC study finds online reviews for some products have a major effect on the sales of others
In the online shopping world, reviews can make or break particular products. Glowing five-star reviews mean big sales, while consumer critiques can make their fortunes fizzle.
When people are browsing online, however, they rarely look at just one focal product: they comparison shop and consider other related items. Instead of looking at a single compact digital camera, for example, shoppers might look at compact cameras by other manufacturers, or other items within the same brand. They might also consider related add-ons such as camera bags, lighting kits and lens cleaners.
All the while they’re reading online reviews of those secondary items — and a new study from the UBC Sauder School of Business shows those reviews can have a major impact when it comes to whether or not consumers hit “Add to Cart” on the primary products. In fact, they can have even more sway than the reviews on the focal products themselves.
For the study, titled On the Spillover Effects of Online Product Reviews on Purchases: Evidence from Clickstream Data, researchers used machine learning to create product pairs between tens of thousands of items viewed in hundreds of thousands of user sessions on a major UK-based online marketplace, similar to Amazon. They used roughly 16,000 pairs of products to first train the AI model; that system then considered things like product names, characteristics and functional similarities as it paired items with near-perfect accuracy. The system then looked at what products shoppers searched for, what they looked at, what they clicked, which reviews they read, what they purchased — and what they didn’t.
The study authors found that positive reviews for similar products — such as queen-size sheet sets or different brands of TVs — could have a significant negative effect on the purchase of the focal product.
“If you look at one product that you're focusing on, and then the other substitute product has a very high rating, it's less likely that you're going to buy the focal product you were looking into,” says study co-author and UBC Sauder Associate Professor Gene Moo Lee (he/him/his). “If there’s a one-star rating increase, it makes roughly a 10-percent difference in your purchase probability. So, a small change actually makes quite a big difference in the economic sense.”
Meanwhile positive reviews of complementary products had the opposite effect on the focal product.
“If I’m looking at an Apple computer, and then I look at complimentary products like an Apple keyboard or headphones, and those complementary products have high ratings, I’m more likely to buy that computer,” explains Associate Professor Lee. “That's the positive spillover effect of a complementary product.”
Interestingly, the researchers found a larger negative effect when people were viewing similar products by different brands, as opposed to similar products within the same brand. The spillover effect was also greatest among consumers who had less experience in a particular product category, and among those shopping on their mobile phones as opposed to PCs.
“In mobile apps, you’re usually on a smaller screen and you don’t want to read all the reviews, so people put more weight on those star ratings,” he says. “As a result, the spillover effects are much stronger if you’re a mobile user.”
The effect varied slightly according to the type of product. For example, a one-star increase on a similar home-related product meant for a 14.5 percent decrease in the likelihood a consumer would purchase the focal item; for a tech product, however, that one-star increase led to a 9.6 percent decrease. A higher star rating on a complementary home product meant a 9.2 percent increase to the likelihood the shopper would buy the focal product, while for tech products it was six percent.
Another influencing factor was “rating variance”: when a product has a wide mix of reviews, from five-star raves to one-star rants, the spillover effect is less powerful than when the reviews are consistently high or consistently low.
Many researchers have looked at the effects of online reviews on consumer purchases, but this study marks the first to look at how reviews of other products influence what shoppers put in their carts.
Associate Professor Lee says as a result of the findings, marketers would be wise to look beyond their own online reviews when devising their sales strategies.
“Marketers usually focus on their own review ratings. But in fact, they have to also consider their competitors’ ratings, and the ratings of complementary products, when they make their marketing strategies,” says Associate Professor Lee, who co-authored the study with experts in econometrics, psychology and information systems from the University of Florida and the University of Houston.
He also adds a suggestion for online retailers: allow consumers to mention other products and brands in their reviews, and to see side-by-side reviews of different items.
“It’s really helpful to let consumers compare ratings and reviews in their interface. It makes their comparison much easier and more informative,” he says. “Consumers rely on reviews, and if you show that in a very accessible way, the consumer will benefit — and that, in turn, will benefit the retailer.”
Interview languages: English, Korean