For many CFOs, investing in a Web site is a lot like a ride through a house of horrors. Web designers dressed in black float by, mouthing strange phrases like deep linking and scripting. The designers, who seem to appear only at night, charge inhuman fees, launch the site late, then leave. Other consultants materialize. These guys claim they can fix the problems caused by the first bunch. Checks are written. The consultants show up three months later, toting data proving what a swell job they did. Unfortunately, the only metric that matters — sales — looks like an EKG in a cardiac arrest unit.
To avoid this scenario, a few companies are turning to an analytic tool called choice modeling. Based on complex mathematics, human choice behavior theory, and scientific design technologies, choice modeling is intended to help companies identify which factors affect decision making. Applied to a Web site, choice modeling enables managers to get a sense of how changes to the site will affect consumer behavior — before the changes are actually made.
Take the case of Australia-based Hilton Healthstream. A $5 million-in-sales direct marketer of health and nutrition products, the company first launched its Web site in 1996. While about 6 percent of shoppers who visited the Hilton site bought some-thing — about twice the rate at most dotcoms — marketing production manager Michael Southall believed the site had design flaws.
Eventually, the company hired enterprise software developer and choice-modeling specialist Memetrics to help with a cyber makeover. Choice modeling turned up some surprises. Example: Site images lauded by the original Web designers did not attract customers nearly as much as the call for specials. “Web analytics don’t tell you why an image is successful,” Southall argues. “I need to know I will increase sales by x percent if I combine certain variables on a page.” Since the redesign, sales at the Healthstream Web site have jumped 120 percent — and the browse-to-buy percentage has doubled.
St. George Bank Ltd., another Australia-based company, is using choice modeling to improve Web performance. The financial institution is analyzing the online strike rate at its private banking business before trying it out with the bank’s 2.5 million retail clients. While the tests are ongoing, Paul Gordon, head of private wealth management at St. George, likes what he sees so far. “Web personalization is too dependent on the operator,” he notes. “Choice modeling goes beyond Web personalization. It’s an enabler to sell.”
Elizabeth Fry is a contributor to eCFO.