Guess is stepping up its digital marketing in hopes of connecting with a younger audience.
The retailer replaced manual product recommendations with an engine from Baynote.
Specialty gifts retailer UncommonGoods LLC was able to increase the number of items it recommends on each of its product pages from four to 20 with software from vendor Baynote Inc. Not only do the recommendations expose customers to more of the thousands of items in UncommonGoods’ catalog than before, but the software shows them a more diverse array of products, too, says director of marketing Brian Hashemi.
Since first testing the technology three years ago, the retailer has rolled it out beyond product pages into e-mail messages as well, he says. Customers who click on one of Baynote’s recommendations in e-mails convert 19.9% more often than when they click on other e-mail links, he says. “They add a good amount of revenue we wouldn’t have seen otherwise,” he says, without giving specifics.
Recommendation engines to promote cross-selling and up-selling on a web site are critical for e-retailers to help shoppers discover products that they might not find using site navigation guides, says Forrester Research Inc. analyst Sucharita Mulpuru-Kodali. While she says recommendations generally account for less than 10% of items shoppers put in their carts, she adds, “the more that recommendations are exposed throughout the site experience, the higher that percent is.”
Previously, UncommonGoods’ buyers had to manually create product recommendations for each of the thousands of items they collectively purchased, Hashemi says. At about five to 10 minutes per item, that process added up to a lot of time, he says. Moreover, with that many items for sale on the site, picking the best pieces to display together wasn’t always easy for the retailer’s buyers. “Our products are so diverse that it’s actually a really complex problem to try and make relevant recommendations,” Hashemi says.
In contrast, Baynote’s algorithms can quickly search through all the options and find items that may not be an obvious match but that share characteristics that might appeal to a particular shopper, Hashemi says. For example, Baynote might suggest a brightly colored beach bag to someone who was looking at a brightly colored vase, even though those two items are otherwise unrelated.
The recommendation engine then goes one step further and tracks customer behavior and interactions with the products it displays. That data include how long customers view particular product pages, how far down they scroll, what they purchase and what other items they click on, among other attributes, Hashemi says. “It’s more than just a binary ‘did they look at this or not,’” he says.
Through Baynote’s web-based dashboard, UncommonGoods can also set up tests for new recommendation algorithms. For instance, the retailer spent about two months testing recommendations within the shopping cart and whether the prices of those items—such as suggesting nothing above $30—would influence a customer’s likelihood of adding an impulse purchase to her order, Hashemi says. The results weren’t astounding enough for UncommonGoods to add the new recommendations to the shopping cart permanently, but it didn’t cost extra or take much time to conduct the test, Hashemi says.
The technology has paid for itself already in the time saved creating recommendations by hand, he says. The retailer pays a monthly fee for the Internet-hosted software, which includes support and service, he says, declining to share exact costs. Baynote pricing begins at $3,000 per month and scales up to $10,000 or more per month for its largest clients, depending on their web site traffic and the services they use, the company says.
UncommonGoods is No. 562 in the Internet Retailer Second 500 Guide.