When campaigns are running, they should be tested through the keywords entered, bids, population dynamics, image changes and the durability of the post - trial and error
However, you should also test the interests of the focus through keywords, categories of points of interest, Twitter handles and tools such as Facebook Custom Audiences.
In order to obtain good results from the trials carried out, we need to make sure that they are statistically accurate. To make confident optimization decisions based on the tests, you need to aggregate a sufficient sample size for the comparison to be valid. Budget, media costs and market segmentation will dictate the size of the potential sample and the number of tests that can be carried out.