In multivariate testing, we begin by identifying which variables on our site might make an impact during the user’s experience that can have a positive effect on leading a buyer closer to a supplier. We start broad—testing many elements on our site at once—and then we drill down to improve the pieces that statistically show promise.
Recently we tested 41 different variables on a page and tested thousands of combinations of those variables. It’s all about finding the optimal combination of variables on a Web page, the combination that delivers the most conversions overall.
TM: What are the advantages of using this two-pronged testing approach?
BM: Multivariate testing helps us to identify areas we may not have considered. Also, it reveals which variables on a Web page can make a difference. For ThomasNet.com, it’s very scalable approach because pages are changed dynamically by a system, so you get to learn quickly what works and doesn’t work.
The biggest advantage of the focused approach is that it is more human driven; we concentrate on the user’s experience. This takes into account user issues and encourages deep thought on our part about the user’s interaction with a Web page in the context of their task at hand. This type of testing can really help identify what users are trying to accomplish, not just what they are interacting with on your site.
TM: What site components are you testing?
BM: We concentrate on the pages that get the most traffic, including our homepage, search results page and the company profile page—where the basic information of a given supplier is displayed. Primarily, we have tested different approaches to navigation, or the best way to organize content on a page, and design, which is the best visual representation of that organized content. We’ve had a wide range of results and have seen increases of 5 percent to 67 percent, depending on the test.