How to normalize impression bias when calculate the ctr?

Question

when we calculate the ctr from the data, like this

             #(click)
ctr   =  ----------------
          #(impressions)

if the number of impressions is too small, the calculted ctr is higher then the bigger impressions. Is there any way to calculate normalized-ctr for evaluating each ads's performance? How can I normalize?

Answer 1

As you know, it is impossible the number of clicks would be higher than impressions! Because in theory, each click is accompanied by an impression all the time! Hence, ctr <= 1 unless you have a technical problem in counting impressions or clicks.