Just a week ago, I was talking about how great it would be to be able to filter and hypertarget AdWords, and I discussed in some detail a list of changes that I hoped would improve advertising performance.
The results are in. For reasons that are a little complicated, it’s difficult to measure conversions directly with that product, and the next-best-thing is measuring time-on-site. The changes I made last week increased our time-on-site from three minutes to four minutes per visitor, which means we’re getting 33% better clicks for the same money. I still think we could improve that pretty ridiculously if we had per-ISP ad filtering, but in the absence of that, a 33% improvement is nothing to sneeze at.
My fear before running this experiment was that artificially narrowing the market would reduce the flow of clicks below a sustainable level. This appears not to be the case. The optimal marketing strategy seems to be to spend only on the best, most valuable potential customers, and only when you’ve saturated that pool do you start widening the net.
This is extremely nonintuitive to me, as my naive strategy is to build a product and marketing strategy to reach the widest possible pool of people. But that strategy is only optimal if you have the infinite amount of resources required to reach absolutely everyone in that pool. I don’t have that kind of money and you don’t have that kind of money. In the absence of a 12-digit marketing budget, don’t spend one penny on the 1000th-best customer until you’ve already reached the top 200. You’re wasting your money.