I believe your approach is a good one, however, I have found in the past that clickthrough performance != sales performance.
A lower clickthrough banner might actually convert a surfer better than a higher-clickthrough banner. So, to throw a monkey wrench into the works, you might also try to determine a gut feeling about a particular banner's performance.
A really blind banner will have a much higher clickthrough ratio, but, misleading the surfer to a sponsor tour generally doesn't earn the sale.
