Define Success and Optimize

Web analytics is the perfect tool for pinpointing problems that usability testing and even highly professional design can often miss. Poor integration of rich-media experiences is a common problem on pharmaceutical sites; and the problem, more often than not, is that no one understands how to measure successful integration.

One way to combat this type of parochialism is to use a functional approach to your measurement. The idea behind functional analysis is simple: Each part of your Web site was built for a specific reason, and by tracking how well it serves a function you can measure its success. When you add patient stories to a site, do you only want visitors spending time on that feature, or do you want to influence them to join a community, register for information, or check out how the drug in question works?

Furthermore, too much focus on measures such as time-on-site or views per visit can simply make the problem worse. This underscores a basic truth about Web analytics—that optimizing to the wrong goals is often worse than no optimization at all. It isn't enough to perform Web analytics, you have to do it right, and doing it right means defining success.

Understand the "Key" to Sites

At the opposite end of the spectrum, some sites have focused solely on interactive components like registration or downloads to measure site engagement and success. But "branded drug" sites also have key informational pages that anchor most successful sessions. Remember, if you only focus on interactions, you'll ignore the sessions where visitors seek information about the product but are not motivated to register.

Key pages are somewhat different for every site, but they're usually obvious (like the "how it works" page on most branded sites) both from the standpoint of site design/intention and from a study of actual site behavior. They also have a distinct behavioral signature that will show up in your measurement. Key pages typically have higher exit rates than immediately surrounding pages (a high exit rate is not necessarily bad; it can indicate that the visitor has found what he or she was looking for), a wider range of "next step" behavior than surrounding pages (when visitors finish a task they branch out), and longer average page times than surrounding pages. Most visitors will not be likely, even with the best of sites, to register or self-assess. So it is vital to measure your success by getting visitors to key pages.

Web sites in the pharma industry are becoming more expensive and complicated. Sites that are richer in content, media, and socialization, but still lack the obvious success metrics, are more challenging than ever to measure. But with careful attention to each part of a Web site, its overall structure, and the full range of interesting outcomes, it is possible to judge accurately how well your site works and target ways to improve it. This is the role, and the promise, of Web analytics.

Gary Angel is president of Semiphonic, a web analytics consultancy. He can be reached at gangel@semiphonic.com