how many impressions are required to determine ctr? i mean, 0-100 hits is not enough to determine ctr, because sample size is too small, and fluctuations do occur more than usual. what about 100-150 or 150-200? how many impressions do you consider "big enough" to take the ctr seriously?