Retailers boost sales with increasingly sophisticated recommendations that are tailored to individual shoppers.
As soon as a shopper lands at BuildABear.com, the site's recommendation engine technology has a sense of who the shopper is. It knows how she landed on the site, what device she's viewing the site on, where she's located, whether she's visited the site before and, if she has, what she has looked at and bought. With each click through the site, the engine presents her with suggestions based on its insights into the shopper and what similar consumers typically buy, says Brian Sawyer, the retailer's e-commerce director.
While the retailer had used a product recommendation engine for years, it was only in June, when it began working with RichRelevance Inc.'s recommendation technology to dig deeper into individual shopper's characteristics that the retailer began seeing significant results from its suggestions, Sawyer says. Since then Build-A-Bear Workshop Inc. has posted strong "double-digit" sales growth online thanks largely to BuildABear.com doing a better job upselling and cross-selling merchandise, he says. And since June the retailer's web site has outperformed its stores in terms of the ratio of shoes sold per stuffed animal sold, a key metric the retailer regularly tracks.
"Our store employees can make a compelling case for how and why a shopper should accessorize a bear," Sawyer says. "We can't offer that same experience online, but we can use what we know about the customer to recommend accessories."
Because of its success with RichRelevance's recommendations, Build-A-Bear is in the midst of testing whether it can further bolster its results by moving the recommendations to different spots on its product pages and checkout page, and if shoppers will click if it offers suggestions on other pages currently without recommendations, like category pages, Sawyer says.
While 702 of the largest 1,000 North American retailers feature product recommendations, according to Internet Retailer's Top500Guide.com, some like Build-A-Bear are finding that after digging deeper into who its shoppers are and what they're looking for that they are able to derive results far beyond their previous recommendations efforts. There's a clear value to offering up relevant suggestions; 39% of online shoppers say they buy more from retailers that present personalized product suggestions based on their on-site browsing and past purchases, according to a recent survey conducted by e-commerce consultancy The E-tailing Group Inc. That's why, in addition to deploying increasingly sophisticated recommendation engines many retailers, like Build-A-Bear, are regularly testing their sites to find more ways to leverage suggestions.
One challenge retailers face with product recommendations is that there are no firm guidelines about where on a site to suggest products, says David Selinger, RichRelevance CEO and co-founder. "Every retailer is different in terms of its assortment and its customer base, and how that customer base responds to recommendations also varies," he says. While that's true throughout e-commerce for most merchandising tools, it is particularly pronounced when it comes to recommendations, he says.
Each retailer must decide how to use recommendations based on objectives. For instance, pet supplies retailer Pet360 Inc., which bills itself as a hub where "pet parents" can read pet-related news, find resources and buy products, uses recommendations to help shoppers discover the breadth of content on the site. A consumer who lands on the site after searching for "How to train with a dog whistle" might not know about the site's nearly 50,000 dog-related SKUs, while the shopper who comes to the site looking for dog treats might not know about the site's rich content, says Rose Hamilton, the retailer's executive vice president and chief marketing officer.
"When people come to our site they might not understand everything we have to offer," she says. "We look at recommendations as a way to curate the site to the visitor."
The retailer tags every piece of content, including articles, videos, products and community threads, using an in-house-developed system. That system works with Certona's personalization tools that leverage which pages a shopper has looked at, what he has clicked on and how he arrived at the site to offer product and article recommendations.
Together the systems attempt to offer shoppers "smart" suggestions based on the retailer's insights about pets. For example, if a shopper arrives after searching for "arthritis in dogs" the two systems working together will likely recommend articles and products related to older, large dogs because those are the most common traits among dogs with arthritis.
Nearly every page on the site features product and article suggestions above the fold, which Hamilton says is the result of its nearly non-stop testing to see what leads shoppers to engage with the site. Testing also shows that every time the retailer adds suggestions to a different section of the site, the time on site goes up, she says.
For shoppers who share information about their pets via Pet360.com's registration process, the retailer personalizes the site to alert shoppers whether a specific product is "right for their pet" with a note that appears just below the product image, and the articles and products it recommends are tailored to the shopper based on her pet information as well as her previous visits and purchases. Those personalized recommendations generate a 117% higher conversion rate than more generic suggestions it serves non-registered consumers, she says. The retailer declined to share the specific conversion rate.
Personalized recommendations that leverage what the retailer knows about the consumer help Pet360 make shoppers feel special by giving them the sense that they're part of a larger "pet parent community," Hamilton says. "Little words make a big difference," she says. "It may sound small to say 'Pepper may also like,' but those touches draw people in and make them feel special."
Personalized recommendations make for a better customer experience, says Brandon Finch, director of e-business at Jelly Belly Candy Co. "You want to make them comfortable with your offering or your site experience," he says. "If the algorithm is doing its job it should show people products they're interested in, which makes the customer experience smoother and better."
Jelly Belly uses MyBuys Inc.'s recommendations engine that builds a profile based on the consumer's clicks throughout the site. The system then offers suggestions based on what other shoppers who have taken similar paths throughout the site viewed. The algorithm the candy retailer and manufacturer uses focuses on driving shoppers to click throughout the site because Jelly Belly uses its commerce site for branding and marketing and wants high engagement rates, as well as to drive sales. MyBuys also has algorithms that aim to drive conversions.
That means that a shopper who arrives at the site and navigates directly to berry-flavored Sport Beans, which are primarily marketed to endurance athletes like marathoners, will likely find other flavored Sport Beans recommended to him. But a shopper who looks at those Sport Beans, then clicks to look at its Cold Stone Ice Cream Parlor jelly bean collection will likely see a broader assortment of products suggested—perhaps one type of Sport Beans, a couple Cold Stone varieties and perhaps another product.
"We are known for our jelly beans but not as many people know about our gummy worms, chocolate products or any of the other brands we have," Finch says. "Recommendations are an opportunity to show people that we make more than just jelly beans."
The suggestions have worked. Finch says testing shows that JellyBelly.com sees between a 2% and 10% increase in its conversion rates after adding recommendations to product pages, search results pages and category pages throughout the site. It also saw a "slight" uptick after recently adding product reviews, in the form of stars, to those suggestions.
Those gains are why Finch says he is constantly looking for new places to show recommendations. For instance, he plans to run a test to see whether a consumer will click on recommendations that appear in the pop-up window shown when an item is added to the cart. He's also planning tests to see whether displaying four or five suggestions produce a better response. "We want to make them as useful as possible," he says.
That's the idea, too, at Build-A-Bear.
"You have to be relevant," Sawyer says. "You have to show the guest that you understand what she's looking for." By using what it knows about its customers—what the shopper has looked at, where she is located, what similar shoppers have bought—Build-A-Bear is showing shoppers what they're interested in. And that's leading more of them to click and buy.