Publishers, Advertisers, and Modified Second-Price Auctions
http://www.adexchanger.com/the-sell-sider/beating-the-second-price-auction/
Our in-class analysis of second-price auctions was limited to the sealed-bid variety, in which all the participants submitted their bids at once and the winner was determined by reading off these bids. A variation of this classic Vickrey auction model, however, occurs when buyers bid in successive rounds, much like in first-price auctions, and when there can be multiple levels of winners. This round-based second-price auction is actually a common tool in the advertisement publishing industry, where the sellers are the publishers and their products are various spaces on billboards, websites, and the like.
There are multiple forces motivating the interaction between publishers and advertisers. Publishers wish to maximize their CPM’s, or costs per mile, which represent how much they make for every thousand impressions (ad views) for a given space. To calculate CPM, you first find the profit per view (amount the advertiser pays minus the cost of maintaining the ad space), then multiply it by 1,000. CPM is a standard within the publishing industry for determining a company’s success. The advertisers, on the other hand, wish to obtain ad space for as cheap a price as possible. They utilize DSP’s (Demand-size platforms) to help them accomplish this goal by establishing a maximum bid that any one advertiser can submit for a given ad space. This constrains the profit margin of the publishers, who cannot use an identical strategy (setting a minimum sell price often results in an individual publisher being excluded from the market).
It seems, then, that this would be an extremely unbalanced auction model, and that the buyers (the advertisers) would constantly be taking advantage of the sellers (the publishers). This is to a certain extent true, as attested to by the author of the article linked above – many publishers do not know the true value of their inventory and simply run with whatever price they think is reasonable. With a good knowledge of one’s inventory, however, it is possible to turn the tables on the buyers.
The author calls the sellers’ dominant strategy “controlled inflation,” and it relies on undermining the information possessed by a single DSP. By throwing multiple remnant advertisers (ad companies that search only for the cheapest ad spaces and are the opposite of “managed advertisers,” or consistent subscribers to one publisher) from different DSP’s into one exchange, the seller makes it so that the advertisers have to compete amongst each other for the space. This artificially created demand causes higher profits for the sellers across the board.
The other component of “controlled inflation” is the raised baseline price. By taking advantage of the fact that the advertisers involved in the exchange do not know the prices that managed advertisers are paying for a company’s ad spaces, publishers can increase their profit. They simply keep raising the price in each round of the auction, starting from the current baseline (the amount paid by the managed advertisers). Once one bidder becomes resolute in their bid ie. they won’t raise their bid, the auction settles in a predictable way: starting from the drop-out point, the other bidders pay one-cent increments of that amount for their own ad spaces ( a quirk of these advertisement auctions is that instead of paying the true second price, the winner pays the second price + one cent.
For example, let’s say a managed advertiser is paying a publisher $1.00 for an ad space. The publisher decides to auction out 3 more identical ad spaces to remnant advertisers. He raises the baseline by 50% for a starting price of $1.50, then begins the auction. The first bidder drops out after a few rounds at $1.98. The other two bidders, being rational people, would both bid $1.98 in the next round. One of these bidders wins and gets the second ad space for the second highest price $1.98 + $0.01 = $1.99. The last bidder, then, gets the last space for $1.99 + $0.01 = $2.00.
The multiple-winner, multiple-round second-price auction has many interesting implications. Since each buyer will be paying for an individual product, it is in the best interest of the seller to raise the “lowest second-bid” as high as possible, since all the other buyers will end up paying similar amounts. In many ways this is analogous to the Vickrey auction, in which the seller has a similar goal of raising the second-highest price to a high a point as possible. The difference, however, is that in the multiple-round scenario, the seller actually has the power to influence this amount (as shown by the tactics described above), whereas in the Vickrey auction, the seller has to depend on the internal convictions of the bidders in relation to the other bidders. We see, then, that contrary to the topical assessment offered at the beginning of this post, where it seemed like the buyers and their DSP’s had an advantage, it is actually the sellers that have an overwhelming advantage in this type of auction – provided, that is, they have a clear knowledge of their inventories and prices.
dhau