By using A/B and multivariate testing to find the best way to target consumers, Shoebuy.com produced “impressive” increases in the company’s revenue per visit and average order value.
By using A/B and multivariate testing and segmenting to determine the best way to target different types of consumers, online shoe and accessories retailer Shoebuy.com produced “impressive” increases in the company’s revenue per visit and average order value. Shoebuy declined to disclose the specific increases.
Shoebuy.com Shoebuy, No. 90 in the Internet Retailer Top 500 Guide, began using the software-as-a-service testing and segmenting services from vendor Amadesa in September.
The retailer is using A/B and multivariate testing to create several versions of a page-moving around elements such as images and copy-and showing different versions to various consumer segments. It then is evaluating results and serving the page combinations that best resonate with each predetermined segment, the retailer says.
For instance, Shoebuy has used the testing and segmenting programs to target mobile users for its recently launched Apple iPhone and iPod touch applications, to determine the impact of security logos on its web site and to run design and layout tests across its catalog pages.
“We love to try new things and let our customers tell us what they like,” says James Keller, senior vice president of marketing for Shoebuy.com. “We simply tag the page elements that we wish to test and then set up the test versions in the Amadesa console.”
Amadesa captures performance results specific to each test segment and then breaks each test segment down further by various geographic, demographic and behavioral traits, so Shoebuy can see both the overall results of each optimization overall, as well as how each change affects each sub-segment, Keller says.
One test Shoebuy is now experimenting with is the placement of customer ratings on product pages. So far, Keller says tests show giving ratings more prominence has improved conversions and revenue per session.
“We look carefully at the resulting data in terms of how key performance indicators are affected and then continue to fine tune around what works,” Keller says.
Keller says results vary quite a bit from test to test, but overall he has been surprised at the impact that small changes can make. “We now always have multiple tests in the market,” Keller says.