Monday, January 01, 2007
Testing, testing...
Recently I've had some questions and comments about testing effectiveness of a website and copy. (1) "I want to test a mailing. First I'll send it to my ezine list..." Are you planning to buy mailing lists? If so, send to a small randomly chosen sample from that list. Your own list may be more or less responsive, depending. (2) "I want to put up a new site with a different domain name and different copy. Then I'll see which works." You're testing too many variables! One site may attract more visitors than the other, or may attract a different type of visitor. (3) "I'll try a new version for a month and see if I notice a difference." You might...but how do you know if your revisions made a difference or if your target market responds to a change in season? The only way to test a web page: Use software (such as 1shoppingcart.com ) that exposes visitors randomly to one page or another. In other words, if ten visitors arrive, visitors 1,3,5,7, and 9 will see Version A. The rest will see Version B. And vary just one element at a time. Headline. Color. Picture. Domain name. Otherwise how will you know what really made the difference? |
# posted by Cathy Goodwin @ 11:37 PM 
