A Shift Towards Auction Standardization
The word “auction” typically brings to mind famous auction houses like Christie’s and Sotheby’s that attract many individuals interested in bidding for or selling priceless artifacts and artwork. While these famous auction houses seem to make auctions very distant from everyday life, there is one type of auction whose results the average person interacts with almost daily. That is, the sponsored search industry which was founded by Google / Alphabet in 1998. Through auctioning advertisement slots by charging per “click” on ads, search engines like Google generate a majority of their revenue.
To better understand the rationale behind Google’s advertisement auction structure, we can draw upon economic concepts in auction theory. Specifically, there are two main types of auctions: first price auctions and second price auctions. In first price auctions, the bidder with the highest bid wins and pays exactly their bid for the good. On the other hand, in second price auctions the bidder with the highest bid wins but instead pays a price equal to the second highest bid. A significant result of second price auction structuring is that bidding your true value is always a dominant strategy (Vickrey 1996 Nobel Prize). Thus, the incentive for search engines to know the true price per click value each advertiser has per advertisement slot motivated Google’s implementation of a Generalized Second Price Auction (GSP) structure.
In a Generalized Second Price Auction, advertisers first announce their bids per click. Then, Google matches each advertiser to a slot such that the advertiser with the larger value per click bid gets a slot with a higher clickthrough rate. The price each advertiser is charged is determined as the next highest bid per click. While this generalization of second price auctions sounds nearly identical to the original version, it surprisingly leads to Nash Equilibrium where advertisers may not bid truthfully. While it is important to highlight that bidding truthfully may no longer be a dominant strategy, an interesting outcome of this slight variation of second price auctions is the existence of Nash Equilibriums with market clearing prices where matching maximizes total valuation. This unintended consequence of Google’s attempt to generalize second price auctions enables it to remain an attractive method to sell advertisement slots as it has the potential to reach a solution that is more socially optimal.
Based upon Google’s long history of auctioning advertisements in the search industry outlined above, it may come as a bit of a shock that Google is ready to abandon its Generalized Second Price Auction strategy that has been implemented successfully for quite a while. In Davies’s article “What to Know About Google’s Implementation of First Price Ad Auctions,” we find Google is actively following competitors to a first price auction structure which it argues yields higher levels of transparency for both publishers and advertisement buyers. This is surprising as we know a second price auction structure was more favorable due to an incentive to know advertisers’ true values. But perhaps the consequences of GSP no longer making truthful bidding a dominant strategy and the search industry shifting to embrace first price auction standardization outweighs these original incentives for Google.
Ultimately, it will likely take some time before we can properly analyze the impacts of a shift to a first price auction structure and whether it is optimal for Google, advertisers, and consumers. Davies’s contends the potential for auction standardization and more bid visibility with Google providing a “Bid Data Transfer File” to publishers that contains all bid information they receive may successfully disincentivize any “over-valuing” in bid behaviors. Thus, it is important to consider whether Google’s new implementation is truly a pure first price auction and how any adjustments may impact new optimal bidding strategies and behaviors.