As SSPs shifted to first-price auctions, app marketers often end up overspending on inventory. Without visibility into the other bids in the auction, the winner was unable to know the difference between their winning bid and the second-highest bid. In the case of losing an auction, the winning bid was not shared with the other participants. Bid shading has emerged as a way to prevent overspending: app marketers can now optimize for the best bid price instead of bidding without transparency. Marketers unhappy with paying more now don’t have to. But what exactly is it and how does it work?
First things first (and second!): let’s recap first and second-price auctions.
In a first-price auction, the buyer pays exactly what they have bid. For example, there are three advertisers bidding on the same inventory:
Advertiser 3 will win the auction and pay the $2.00 that they bid.
In a second-price auction, using the same scenario:
Advertiser 3 will win the auction, but only pay the second bid price plus one cent. In this case that’s $1.51.
Both methods have their positives and negatives for both the advertiser and the publisher. You can learn more about them in our white paper about the basics of programmatic advertising. Bid shading seeks to find a middle ground and level the playing field for all parties involved in real-time bidding (RTB).
Bid shading is the practice of predicting the most accurate bid price and bidding accordingly. During a first-price auction, bid shading provides clarity on the other bids in RTB and more accurate bidding, which leads to higher ROAS and lower CPM. Once an auction is won, the second price is revealed. When an auction is lost, the winning bid is revealed. Machine learning algorithms can use this data to bid more accurately in the future.
As most SSPs move to first-price auctions, bid shading is becoming more and more necessary for DSPs looking to stay ahead of the curve. Instead of bidding with only user behavior data, internal winning price data, and other contextual information such as geo, ad size, or exchange type, DSPs can also factor in winning price variables, i.e. historic data about previous winning bids. Factoring in this extra information about previous bids, a machine learning algorithm predicts the ideal bid for each auction.
As SSPs moved from second to first-price auctions, DSPs weren’t entirely equipped to make the switch. We discussed the fact that we believe first-price auctions allow for a more level playing field in RTB, and bid shading allows DSPs and other entities participating in RTB to continue to optimize for the most fair and accurate price.
One of the downsides of bid shading for the time being is that not all SSPs are as transparent about winning prices and second prices as they could be. There are a couple on the market at the moment who are able to provide this key information to DSPs in order to help build their machine learning capabilities. However, it’s expected that most if not all SSPs will begin adopting first-price auction transparency in order to avoid DSPs reducing spend.