I would say that you should test both theories out. If you have the same quality and quantity of traffic going through your sites, then use the same banner for a week and then try different banners for a week and compare results. Once you know what works best, repeat
