roodle - 1:42 pm on Mar 3, 2010 (gmt 0)
Thanks for your reply. I have 3 or 4 ads in each adgroup. Although the style of ads in the same group is the same, each ad is subtly different from the others. So in effect I'm comparing ads within the same adgroup AND overall adgroup performance for all adgroups. The only common elements are the keywords. Is this not a valid experiment?