Don't test more than one hypothesis simultaneously. Test with variations that significantly differ. Too small details might be too less to give you a precise result.. It isn't only about adding or editing elements - sometimes it's also about removing them!