Frontiers in Dynamic Pricing: Spot Advertising Auctions

Lynne Kiesling

According to this Ars Technica story (and a linked Bloomberg article), Facebook is going to offer a new advertising model to its potential advertisers: a spot auction for real-time ads based on changes in current events or time-sensitive things like sporting event results.

The service, called Facebook Exchange, will use partnerships with other companies to track users as they visit other sites using tracking “cookies” placed on those sites, and allow advertisers to bid “in real time” to display ads based on the interests the browsing history represents. …

The real-time nature of the bidding system means that advertisers can target ads based on both recent behavior of Facebook users and real-world events. For example, people who have a web history related to following a specific Olympic event could get offers based on the outcome of that event.

For the moment, set aside the corresponding privacy issues associated with this use of cookies (although if it does present individuals with ads targeted to sites they’ve visited, that targeting may benefit consumers, and we should not forget to take that into account).

Instead, think about this as a matching or a search problem. Producers want to identify high-value consumers, and whether or not a consumer is high-value or low-value is a function of their context of time and place. Here’s where the Hayekian diffuse knowledge point comes in — the “man on the spot” has private knowledge about how much value he places on, say, buying a Spain jersey to celebrate a victory in a Euro2012 game (yes, I am expecting them to beat Ireland this afternoon!), and that value is itself a function both of whether or not Spain wins the match and of the fan’s perception of the value he attaches to getting a Spain jersey right in that moment. Before digital technology and social media, producers could not identify those high-value consumers in the moments when they are truly high value consumers, so the technology opens up new business models and reduces those search and matching costs in a much more dynamic way. Similarly, from the consumer’s perspective, if I’m exuberant because I’ve just watched a brilliant soccer match and Spain totally dominated (thanks in large part to the outstanding field marshaling and traffic direction of holding midfielder Xabi Alonso), I’m going to be happy to have lower search costs of finding a Spain jersey because of the targeted advertising.

Models of dynamic pricing suggest what we should expect to see in Facebook’s ad pricing — lower prices for time-sensitive products and services at times that are more distant from the event, higher prices for ads closer to and during the events. This advertising price discrimination may also be a better revenue model for Facebook, for whom advertising revenue has not been reliable in the model they are currently pursuing.