The office supplies merchant is deploying Internet-based supply chain software from HighJump Software to connect ...
Office Depot writes its own A/B testing application to conduct tests that deliver more granular results about its site design and navigation paths, says director of global e-commerce strategy.
Office Depot Inc. is a big believer in A/B testing to drive e-commerce sales and make upgrades to its web store.
Specifically Office Depot, No. 3 in the Internet Retailer Top 500 Guide, writes its own A/B testing application to conduct tests that deliver more granular results about its site design and navigation paths.
The application, which was developed using Java script and deployed in early 2007, is capable of creating test pages designed to answer specific questions such as whether a specific feature can increase store traffic or enhance the brand, as opposed to just measuring conversion rates. For example, Office Depot earlier this year began testing pop-up windows on web pages viewed by users of Internet-capable mobile devices such as cell phones. The pages, which must be specifically formatted to the smaller screen size of the mobile device, were intended to build brand awareness by welcoming the visitor to the store or to gather feedback about their shopping experience in that medium by directing them to a survey.
“The aim was to create a testing application that did more than tell us response rates to a design feature or promotion, but that gave us a finer-grain view of results,” says Office Depot director of global e-commerce strategy Noah Maffitt. “We wanted to create features for specific customer sets and writing the application in-house allows us to get to that level of detail.”
The key to A/B testing is to try different triggers that may get the customer to respond, and then analyze the data around their navigation paths to confirm whether the offer or page is generating results, says Maffitt. “You have to be willing to test random ideas,” he says. “There is a different creative cycle throughout the calendar year, so why not apply it to A/B testing? The goal is to find the combination of elements that work.”