“The Sell Sider” is a column written by the sell side of the digital media community.
Today’s column is written by Kean Wang, VP, Product & Strategy, Intowow Innovation.
When Google Ad Exchange completed the last mile of the industry’s transition from second-price auctions to first-price auctions in September 2019, publishers were enthusiastic about the change. They believed, intuitively, that this transition guaranteed a price jump in media-consumption cost that benefited the sell-siders – which could be true if demand-side platforms (DSPs) reacted to first-price auctions the same way they did in second-price auctions.
However, the result turned out to be the exact opposite.
The change in auction model completely altered the bidding behavior of DSPs and transformed the DSPs into insatiable cost-saving (bid shading) machines, alongside their already-powerful audience-cherry-picking capabilities. This could turn into the worst nightmare for publishers.
Here’s why such a development was inevitable and what publishers can do to break the ongoing downward spiral.
First-price auctions opened a Pandora’s box.
As a starting point, we must put ourselves in buyers’ shoes and reinterpret the second-price to first-price auction transition.
Media buyers weren’t concerned about auction models at all. Regardless of auction types, they demanded the same set of media inventory and audiences for a given campaign and allocated the predefined amount of media budget to each DSP.
In second-price auctions, bidders (DSPs) were incentivized to bid the true value of impression opportunities and gained an inherent economic surplus (i.e., the monetary gain between value and cost) guaranteed by the system. But in the case of first-price auctions, the surplus has been eliminated, forcing the DSPs to internalize the surplus by deducting it from true values when making bidding decisions – a process known as bid shading.
Bid shading achieved much more than that original goal to control costs.
Previously in second-price auctions, clearing prices were controlled by fellow bidders (or the invisible auction system), making it almost impossible for any DSPs to manipulate clearing prices, except in the case of massive corruption.
But in first-price auctions, clearing prices are administered directly by winning bidders. This opened a Pandora’s box and allowed DSPs to continuously shade the clearing prices (media costs) while retaining the target win rate for their customers.
The well-intended ad exchange price feedback made the situation worse.
To streamline the process toward market equilibrium in first-price auctions, most ad exchanges (SSPs) provide price feedback signals to DSPs, such as the “minimum bid to win” field populated by Google Ad Exchange to its Authorized Buyers. This signal allows DSPs to speed up the price iterations on their bids and converge to an optimal price more quickly.
This is bid shading with turbo boost on a collective scale.
When “minimum bid to win” information is shared, the market value of publishers’ inventory very soon decays. Buyers who are already shading their bids can inch their prices downward even more, closer to the second-highest bid in the market.
Such consolidation is very similar to the equilibrium in second-price auctions – but with a depressing distinction.
In second-price auctions, publishers’ market value was a penny more than the second-highest bid of a “true value,” whereas in first-price auctions the market value drops to the second-highest “shaded true value.”
Over time, as bid shading algorithms iterate themselves with the help of tons of historical data, and as poor-performing DSPs are wiped out from the market, the market value of publishers’ inventory will drop continually to subsequent “better shaded” new equilibriums.
Floor prices are the strongest bargaining chips for publishers.
As a publisher, you should definitely set up floor prices to stop this downward spiral.
Floor pricing works by disrupting the existing DSP win rates in first-price auctions. When DSPs’ win rates decline, it forces them to raise their bid prices in order to secure their win rate targets.
By setting up floor prices, you’re sending the “minimum bid” signal to bargain with the bidders over how much your impressions should be worth. Floor prices soften the impact of bid shading by compelling bidders to reformulate their bid shading algorithms to the new boundaries.
The price feedback gathered by ad exchanges and sent to bidders also takes floor prices into account, and that data is essential in limiting the downward pressure of bid shading.
That means your optimal floor pricing strategy is a range of price points that could alter values of the “minimum bid to win” field. But it should, on the other hand, stay lower than the “true value” distribution of the market’s highest bidders to avoid leaving too many of your impression opportunities unsold.
Publishers are now bargaining with super robots.
Now back to the real world.
Buy-side technologies are composed of dedicated machine-learning modules fueled by billions of historical auction data points. Bid shading, in reality, is a cost-saving module of DSPs’ wider bidding intelligence complex. Using powerful capabilities like dynamic audience clustering and time-series simulation, DSPs compute the optimal bid prices for each audience cohort at any specific moment.
To go up against the DSPs’ bidding intelligence complex, general-level floor pricing strategies will have limited results. When the effect stagnates, publishers may want to try updating their strategies, multiplying them into finer granularities to bargain more symmetrically with buy-side robots.