Well, the results are in. Latest CPC and conversion costs have climbed, and now are not much different from where I started out. CTR and conversion percentages are still down from where I started with my manual bids, which I am not totally comfortable with. Average cost per conversion is down, but I think that is primarily attributable to a number of keywords that I was overbidding for. Quality of conversions also seems to have suffered, with many more of the users who registered in the last couple of weeks already having agents they were working with, and/or providing more bogus contact info. I don't know how that happened, or if it was even related to the different ad positions or other AdWords factors at all, but it just seems to have been the case.

This simply tells me I had a pretty good handle on my keyword bids to begin with, and not that the Bid Optimizer doesn't work. Bid Optimizer did reveal quite a few keywords that I was over-bidding significantly on, so that is a plus.

Final decision: It was an interesting and educational exercise, and good for brand new AdWords campaigns, but not for my more mature and stable campaigns that I had already optimized manually. I have now reverted back to manual bid management, keeping many of the Bid Optimizer recommended bids, but reverting most of the others to what they were when I started. This allows me to better focus my ads and bid priorities on my top communities, which are high-end, and lower bids and priorities on other low-end communities. Google would not know about these priorities, so gave all of them equal treatment.