(Page 2 of 2)
While some retailers rely on consumers' expressed preferences, many web retailers collect data on how online shoppers behave, much as a store salesperson might note the items a shopper is considering, in order to make recommendations. One example is web-only retailer Altrec.com which considers the relevance of the content it displays to each shopper a way to differentiate itself from other retailers of outdoor apparel and accessories, says John Hnanicek, chief information officer and vice president, merchandising.
In Altrec's case, the personalization is based on what a shopper does on his current visit to the e-retail site—and not on past behavior or purchases. Working with personalization technology vendor Baynote Inc. the retailer utilizes cookies to gather 24 metrics, such as the amount of time a shopper spends on a page and whether he scrolls down the page, to pick up strong or weak signals of interest. "If he hits a back button a second after arriving at a product page, obviously that's a weak signal," says Carlos Carvajal, Baynote vice president of marketing. "But if he spends a lot of time on a set of similar products we can determine that he's very interested in a particular type of product."
Layering that individual consumer's behaviors on top of how thousands of previous shoppers acted enables the retailer to recommend to him the products that like-acting consumers bought. The approach works, Hnanicek says. A/B tests show that the conversion rate is 4.5 times higher than the manually entered recommendations the retailer previously offered. And the average order value is about 15% higher.
Key to that success is correctly pinpointing the current shopper's interests and which group of previous consumers, or persona, he matches up with, says Hnanicek. He says making that determination based on current behavior avoids the trap that e-retailers can fall into if they base recommendations on past purchases. For instance, a few months ago Hnanicek bought a children's DVD for his niece. The purchase was extremely atypical. Yet nearly every time he arrived at that retail site for the next two months the site suggested he might be interested in other children's DVDs.
"Our suggestions are based on a consumer's intent or interest rather than looking in a rear-view mirror," he says. "Relying on the wisdom of the invisible crowd works."
An added benefit to the approach is that it avoids any complications that could occur if the site based its recommendations on a cookie that tracked a consumer's past behaviors. In that scenario if a consumer deleted the retailer's cookie, the site's ability to offer relevant recommendations would be lost, he says. A recent study by web measurement firm comScore Inc. estimated 28.5% of online consumers delete their cookies each month.
One crucial data point
While Altrec uses dozens of data points to create its recommendations, online-only military paraphernalia retailer Medals of America largely relies on just one—where the shopper is coming from.
Most visitors see a single version of MedalsofAmerica.com, but about 15% of the retailer's sales come from shoppers who arrive from the online mall of the Army & Air Force Exchange Service, a site where U.S. military personnel and their families shop. And those shoppers see a distinctly different version of MedalsofAmerica.com that is primarily focused on Army and Air Force-related products. That version of the site also offers discounted pricing and accepts the Military Star Card. Similarly, those who arrive at MedalsofAmerica.com via the Navy Exchange's online mall see a site heavily focused on Navy-related products with its own distinct pricing.
To provide this customization the retailer uses CSS3, the latest version of a web design language called cascading style sheets. CSS allow a retailer to present images and product information on its site differently based on the referring site.
"We like the idea of personalizing depending on where someone is coming from," says Lee Foster, Medals of America's owner. "But we didn't have the resources to build an entirely new site each time around. A CSS layer is all we need to give it a unique look and feel." Foster says Medals of America can create each new CSS in-house within about a week using Ability Commerce's SmartSite e-commerce platform, or it can outsource the job to the vendor for about $1,500 per project.
The site also uses a consumer's IP address to highlight climate-specific products. "That way a guy in the North might see a jacket, while a guy in the South will see T-shirts," says Foster. That type of localization increases the likelihood that a shopper will make a purchase, he says.
Since Medals of America launched its first personalized online store for visitors from the Army & Air Force Exchange Service, that online store's sales have risen roughly 3% a month. That's because it offers shoppers a better experience, says Foster. "It makes it easier for someone to find what they're looking for, which ultimately translates into sales," he says.
The trick is figuring out what each shopper is looking for. As more consumers share information about themselves, particularly on social networks like Facebook, that trick should get easier for online retailers to pull off.