I read Jason Cohen's article about significance testing adwords ads using the Pearson's Chi Square formula. It was very well thought out for the layman.
I am far from being a statistician, so can someone clarify a few points:
Can you apply this using the standard settings in Adwords i.e. For Google to show the more successful ad more often, or should you use the "show ads equally" setting?
Can it be used ongoing? So you've found a winning ad, ditched the loser and created a new contender. The winning ad already has some clicks. Do you count them or start from a zero baseline for both ads the day you put up the new contender?