Balance Algorithm in Auctions

Reviving 2 year old blog with a post on Auction Theory. I was looking into existing solutions for the classic adwords auction problem. Natural instinct says, that we can order the ads in auction using a naive greedy strategy of bids from the advertisers. This approach is not optimal, and has got a competitive ration of .50. The competitive ratio is an indicator of revenue capture or monetization efficiency from the advertisers in auction. A slight variation of the positions allocation to advertisers using remaining budget has got a competitive ratio of .8 in majority of use cases. This variation is called as Balance algorithm. If you want to take relevancy of the documents into consideration, you need to model a function with ctr and tfidf score of documents. I am surprised how subtle variations of algorithms can have an impact of millions of dollars on businesses.