Suppose we wish to detect a difference of $0.094 (just under a dime) between two different online ads. Suppose the standard deviation of the response (sales) is $103.77 (the standard deviation will be large because most clicks do not produce sales so there are lots of 0's in the data). For an A/B test how many observations do we need in each sample? Use a power of 0.8 and $alpha=0.05$.