Media Planning & Buying

Google Shakes Up the Programmatic Landscape Once Again

March 20, 2019

Google recently announced that its Ad Manager platform would be joining Index Exchange, OpenX, and the rest of its contemporaries by rolling out a first price auction model for its Ad Manager platform. This move away from the traditional second price auction puts programmatic buyers in a tricky position, as they must once again evolve their overall buying and optimization strategy to keep pace with an ever-shifting marketplace.

However, Google’s announcement also delivers certain benefits for the buy side. Now that one of the largest inventory holders in the marketplace has joined the other exchanges in an industry-wide transition to the first price model, buyers are no longer forced to ask the always-thorny question of “what type of auction am I in?” With that issue effectively settled, buyers can now focus on an even more complicated question—one that has lain at the heart of the programmatic marketplace over the past two+ years—“am I paying the right price?”

 

Fall of The Second Price Auction

In the beginning, when programmatic buying was just getting off the ground, the ad exchanges developed a Vickrey-style auction format to govern their exchanges. The Vickrey format is a sealed-bid auction in which the highest bidder only pays the second-highest bid plus $.01. For example, if a you were to submit a winning bid of $7 for an impression in a programmatic exchange, and the next highest bid came in at $5, you would only pay $5.01 for that impression. This approach, also referred to as a second price auction, is similar to the process used in auctions on eBay.

The Vickrey style was adopted by many digital auctioneers in the early days of the internet, including Google’s search auction. This is in part because it theoretically incentivizes bidders to submit their “true value”—rather than over- or underbidding to gain advantage. Of course, theory and practice are often miles apart. Smart buyers quickly realized that they could raise their bids progressively over time to be more competitive in the auction without fearing that they would eventually pay more than market price. This was great for the buyers but created problems for publishers attempting to manage yield and establish consistent value for their inventory.

 

Rise of the First Price Auction

In 2016, SSPs had begun the slow transition to an alternative bidding model known as the first price auction. In a first price auction, the highest bid determines the final selling price. This means that if you submit a winning bid of $7, and the next highest bid comes in at $5, you pay $7. Unfortunately, many of the SSPs were less-than-transparent about this transition at first, which resulted in higher CPMs and created significant disruption across the buying world.

To compensate, Programmatic buyers were forced to modify their entire strategic approach to deal with the relatively new first price auction format. Buyers found they had to constantly monitor their bids and clearing prices to ensure that they were commensurate with the “true value” of the impression. Despite their vigilance, the transition to first price bidding still resulted in higher inventory pricing overall, and increased spend for the buy side and its clients. At the same time, it also removed buyers’ visibility into the prices that competitors were paying, erasing their ability to discern the true value of audiences, and forcing a complete overhaul of buyers’ approach to the marketplace (e.g. bidding strategies, proxies for audience valuation, etc.). This was great for publishers and SSPs—not so much for the buy side.

 

Enter Bid Shading

In 2018, SSPs attempted to address this imbalance on their own by unveiling an entirely new auction dynamic known as bid shading. Bid shading represented something of a compromise between first-price and second-price auctions. For example—if Buyer A bids $5, and Buyer B bids $7, the $7 bid will win but that buyer will pay an amount less than $7 and greater than $5, with the exact amount being determined based on the clearing price value set by the exchange’s algorithm. Bid shading is an attempt to create parity between buyers and sellers by keeping closing prices somewhere between the values that second-price and first-price auctions would have produced.

Bid shading provides buyers with significant savings as compared to first-price auction CPMs—without sacrificing win rates. It may generally benefit buyers more than publishers, but the mid-range bid shading prices are still generally higher than the values that would have cleared in a second-price auction, meaning that bid shade adds value for players on both sides of the field.

Some SSPs have developed proprietary bid shading algorithms, which usually take the form of a free opt-in service that is activated on the buyer’s seat. This allows the DSP’s bidding system to continue to optimize an advertiser’s bids based on their campaign objectives and KPIs. For example, Rubicon’s Estimated Market Rate (EMR) is designed to find the market rate, lower bids algorithmically, and keep the win rate up. However, some SSPs like Index Exchange believe that bid shading, or any other pricing adjustments, should ultimately be performed by buyers, so that they have maximum transparency into bids and clearing prices.

Either way, SSP-supplied bid shading tools are stubbornly opaque, and generally amount to an SSP’s proprietary algorithm setting the price for that SSP’s auction. This means buyers have no ability to verify whether the SSPs are setting the right clearing price, and that the only way to confirm the value of that inventory would be to run a full audit of the SSP. Worse, SSP-side bid shading alters DSP bids in a way that inhibits the DSP’s ability to set the appropriate bid price, as well as its ability to optimize for the campaign’s KPIs. This means that the DSP algorithms can’t effectively learn the actual dynamics of the auction its participating in.

As for the DSPs, many of the major buy side platforms have customized their bidding algorithm to adjust bids in real time, leveraging machine learning to determine whether an auction is first price or second price before making a bid. Google’s DSP (DV360) has a feature called “Optimized Fixed CPM bidding” that applies a discount factor to lower bids where appropriate. Other DSPs like The Trade Desk and Oath created separate tools to identify first price auctions and apply bid multipliers, insuring that buyers are not over-paying compared to the average market rate.

 

The Future

Looking towards the future, the larger ad buying community is optimistic that the first price model will provide greater transparency than what was available with the second price model. Buyers will spend less time trying to understand, and work around, opaque auction systems. Instead, we’ll be able to spend our time focusing on what we do best—accurately measuring, planning, and optimizing for our clients.

Fortunately, because all of the other primary exchanges switched to first price auctions throughout 2018, many buyers—including the team at 360i—have long since adjusted their bidding strategies. Last year, 360i conducted an exhaustive audit of exchange prices that was used as the foundation for our bidding strategies across buying platforms, and a model for the optimizations we make in response to campaign objectives. This means that Google’s transition to first price auctions will have a minimal impact, if not a negligible one, on how we buy.

360i’s Senior Director of Programmatic Noah Porter and SVP, Head of Programmatic Kolin Kleveno contributed to this post.