• by Seth Rosenblatt , April 5, 2007

It's official. Quantitative, direct marketers have seized the throne of the online business kingdom.

In conducting A/B and multivariable tests for Delta Air Lines, IAC, and many of the other biggest Internet companies over the past six years, we've seen how seemingly minor design tweaks can have major impacts on Web conversion rates.

With Google's announcement yesterday of the launch of its new Website Optimizer, now even the smallest mom and pop e-commerce sites will be able to get into the Web page testing game. It's the strongest signal yet that the quants are here to stay.

We certainly expect to see a lot more interest among all types of online businesses in Web page testing. What can these businesses expect to see when they start testing different combinations of copy, graphics and forms?

For those who want actionable results, rather than a pile of "lies, damn lies, and statistics," it's important to start with a plan. Web page testers need to ask the following questions before starting: (1) what should I be testing, and (2) for whom should I test this?

Clients of ours have tested anything and everything you can put on a web page -- banners, headlines, text, colors, images, logos, pricing, and even more subtle areas like submit buttons and supporting copy. While it is possible to test every nook and cranny of a page, the results don't mean much unless you know what worked, what didn't work, and what didn't make a difference.

Did the red submit button perform better than the green one? How much better? Would it have performed even better with a free trial offer? Being able to spot outliers in the data and correlations between different variables is paramount.

Once you know what works, the next step is to determine who it's working for? Do you want the site to perform best for the highest-paying customers? For people searching particular key words? The analytics gathered in the test can be used to further "segment" a Web page audience and create "personas" based on information we know about visitors.

The end result of this process is often startlingly simple. For example, Time Life recently tested a static "order now" button on products against a flashing "click here" button, and found the latter produced a 7% lift in the number products added to shopping carts. They gained another 8.6% lift by adding a "money back guarantee" button. They always had the guarantee, but the simple act of touting it drove more sales. Ultimately, through a series of tweaks like these, Time Life realized a greater than 70% lift in the number of products added to online shopping carts, which ultimately translated into a seven-figure impact on annual sales.

When it comes to segmenting, a large dating site found that those visiting its page during normal business hours were receptive to a main photo depicting a happy couple, while those visiting in the evening were much more receptive to a photo of three provocative-looking blond women. The psychographic implications of that particular finding opened a treasure trove of marketable data for the company.

In all cases, by following a systematic approach, it is not long before optimization becomes part of the DNA of an organization.

By continually testing the page, Web marketers are able to pinpoint the impacts of seasonality, the psychology of different types of shoppers, and the key drivers of sales.

These data points ultimately become the most important determining factors for creative design, special offers, even inventory decisions. In a nutshell, the data -- and the people who work with it -- become the kings of their domains.

Seth Rosenblatt is vice president/marketing and business development at Optimost, a company that provides technology and services to enable real-time multivariable testing at all points of customer interaction online. Reach him at 650.432.6072.