Best Online Marketing Practice - Don't Just Know Your Product, Know Your Numbers

By Gregory Smyth

Testing is a critical part of the website design process - unfortunately it is also one of the most neglected activities of the procedure. Many companies go with their gut instincts, or with what works for them in traditional marketing, forgetting that the internet is a law unto itself. Management can be very hard to convince that testing will yield any real value - short testing campaigns conducted on a black-op basis can sometimes be the easiest way to go. We look at the importance of testing, how it can improve your conversion rates and how to go about testing on a one-person, rather than inter-departmental, basis.

In one recent case, testing only a few small changes to a web page yielded a 1.7% increase in conversions from overall traffic, which represented a 62.9% increase in their conversion rate (up from 2.7% to 4.4%). These results were obtained within four months, testing only several critical factors that did not take much time to implement. Headlines are a critical part of internet marketing, and the headline in this case was changed to a more benefit-oriented, personal statement: from 'Get a toll-free line' to 'Look like a Fortune 500 company'. The person pictured on the page was changed from a somber, business-like person to a smiling young Black woman - the same sex as the target audience. The copy was also slashed by around 50% - there is a wealth of studies that show that the effort spent on search engine ranking optimization, article marketing and link building is wasted when the site copy is long-winded.

Here are some guidelines to test out some of these ideas on your own site and see what sort of results you can achieve. These are oriented to testing operations involving one person that don't necessarily need management consent - as long as you don't mind acting on this basis.

You should only test one page at a time. This way you can be sure of the changes that have had the biggest impact. Changing every page on the site effectively negates the benefit of doing the testing - you may have slightly improved conversion through your internet marketing know-how, but you don't know exactly what you did, or how much that particular activity cost. The page you choose should be one that is either designed to change a visitor into a prospect (for example a white-paper download page, or a newsletter subscription page), or one designed to convert prospects into sales. The latter is even more useful, as you can directly link your website design spend and web marketing strategy spend to the return on investment.

If you are the sole tester or administrator, it is preferable to use an ASP web analytics package rather than an enterprise web analytics package. With this type of online marketing software, the data is stored by the vendor, rather than on your own system. If you want to modify any elements of your testing operations, this is generally easier to do with an ASP package anyway.