RTB Exchanges Are Unusable For Guaranteed Inventory

By Amihai Ulman Do you remember that little interstitial on Sesame Street where a song played in the background with the lyrics "one of these things is not like the other," while a picture appeared challenging the viewer to figure out which one is not the same? Having worked on making all advertising truly tradable, that song repeats more loudly in my head every day.

Over the course of the last few years, I have come to realize that the main reason why advertising has not been made truly tradable is because of a lack of innovation. 'None of these things is any different than the rest.' In short, all of the solutions we have seen to date have all shared a single approach to market and exchange product development: what do we have that works elsewhere that we can apply to advertising?


This is a fundamental misunderstanding of the nature of advertising. One should first ask: does advertising behave like other traded assets? If the answer is yes, than we can ask: what do we have that works elsewhere that we can apply to advertising? The problem is that advertising does not behave like any other traded asset. If you have followed this blog, you may recall that we talked about advertising behaving more like real estate than any other asset. That should be an important clue. Real estate does not trade in an exchange. Real estate does not behave like other tradable assets. So, if we really want to make advertising tradable, we must go back to square-one and build an entirely new exchange framework. While the Sesame reference seemed to be tongue-in-cheek, it was absolutely serious.

Market structures for advertising require a completely new approach to structure the media transaction. It might not be obvious how badly the media trading markets needs this new approach, but it becomes much more obvious when compared to the numerous other transaction markets within our economy. Global media spend will exceed $600 billion in 2013. That is about 1.5% of global GDP! RTB is a wonderful innovation, but billions of dollars in advertising (television and print) have no use for trading real-time impressions. These markets need to transact on a forward basis.

So, let's consider how other markets handle transactions. As I noted in an earlier post, "In nearly all other exchange traded products, all buyers will realize the same value from the transaction." It is this nature of other exchange traded products that translates into relatively tight spreads between bids and offers in liquid markets. Conversely, in an advertising market with many buyers and sellers, it is unlikely that the price of orders to sell will match perfectly with the price of orders to buy. It is this ambiguity of a generally agreed upon value and price that makes advertising so different than any other asset. In other words, a market price is not derived through a consensus of all buyers. Forcing a consensus among buyers is the very reason why second-price (one-sided) auctions have failed as an exchange structure for reserved or guaranteed inventory.

Sequential auctions, the fundamental structure of RTB 'exchanges', have the effect of limiting the information that is available to bidders and of limiting how bidders can respond to information. With sequential auctions, bidders must guess what prices will be in future auctions when determining bids in the current auction. Incorrect guesses may result in an inefficient assignment when ad placement values are interdependent. A sequential auction also eliminates many strategies. Using existing exchange structures for guaranteed inventory, a seller cannot switch back to an earlier item if prices fall too low in a later auction. Bidders are likely to regret having sold early at low prices, or not having sold early at high prices. The guesswork about future auction outcomes makes strategies in sequential auctions complex, and the outcomes less efficient.

Unfortunately, it seems unlikely that the RTB market participants will be able to determine accurate reference prices, particularly given the fact that price discovery is limited only to a knowledge of the clearing prices of what an individual market participant transacted. A secondary market simply does not exist —recall that this is the rationale for the need for MASS Exchange's innovation—and the few transaction prices known (as well as the bids) can be presumed to be highly misleading indicators of value. Inaccurate reference prices give rise to an adverse-selection problem.

The experimental economics literature strongly supports our conclusion that dynamic auctions outperform sealed-bid auctions in terms of efficiency and price discovery. In sealed-bid auctions there is a tendency to consistently overbid (Kagel, Harstad, and Levin 1987; McCabe, Rassenti, and Smith 1990), often resulting in inefficient outcomes. While the second-price methodology seeks to alleviate this problem, the second-price method is only a solution where each impression is auctioned separately. Guaranteed impressions will always be sold on a forward basis and thus can not be priced at the individual level, making current RTB market model unusable for guaranteed inventory exchanges.