Hi, What do you guys consider a statistically significant amount of impressions/clicks to find the CTR of your ads?
First of all, test two ads at the same time. Don't test one ad one week and the other another week. That would not be a statistically valid test. Wait until both ads get at least 10 clicks each before deciding which one is the winner. There is no time limit. Ten clicks whether it takes two days or forty. You can do this on the search and content networks. Each should have roughly the same number of impressions too so make sure your campaign setting is at Rotate and not Optimize as that setting favors one ad over another and that skews the results. If you test too many ads, it's also is not a valid test as you are diluting the impressions. I test two ads, sometimes three if there are enough impressions each day. Never test more than three at at time. If you want to use impressions instead of clicks, wait until both get at least 250 impressions on the search network but if you only get one or two clicks, that still won't be statistically valid. If one clearly outperforms the other in 250 impressions, it would have to have 7-8 clicks to 3-4. When testing landing pages for conversions, wait for 10 conversions for each page. Same principle.