Today, to determine the winner of an in-app programmatic auction, bid shading is becoming an increasingly common way to determine the final price paid of the demand-side platform (DSP) winning the auction.
This model of bidding has arisen as more publishers and supply side platforms (SSPs) are moving to first price auctions - or considering this bidding strategy - as DSPs are not always keen on the highest bidder paying exactly what they bid.
In this latest Whiteboard Wednesday video, Ryan Gauss, InMobi’s Platform Product Manager, explains how bid shading works and how it impacts what the buyers bid, win rates and bid history. The video also highlights how both first price and second price bids affect a DSP’s chance of winning and final payout rate.
Hi, and welcome to another edition of Whiteboard Wednesday. My name is Ryan Gauss, and I’m the Platform Product Manager here at InMobi. Today we’re going to be discussing Bid Shading, why it came about, why it’s needed and what the advantages and disadvantages of it are.
So before we get into Bid Shading, let’s talk a little bit about First Price and Second Price Auctions. And, as we discussed in our previous Whiteboard Wednesdays on First Price and Second Price Auctions, in a First Price Auction what typically happens is the DSP or the buyer or the bidder - they’re all kind of used synonymously - in a First Price Auction the DSP will end up bidding a little bit lower but end up paying higher. That’s because they end up paying what they bid.
So because of this, DSPs weren’t very happy that they ended up having to pay this higher price with First Price Auctions, even though there were benefits of it such as removing a lot of the lack of transparency and allowing them to be able to compete more effectively across multiple auctions.
So what we see with, when we get into Bid Shading, is it came about because A) first off, DSPs had to create a new bidding algorithm to be able to compete in a First Price Auction, and some DSPs have been slow to do that. They’re still trying to use a Second Price bidding algorithm in a First Price Auction. So that’s really to their disadvantage. They end up paying higher than they would have to.
And even for the DSPs that have been able to update their algorithm to a First Price bidding algorithm, they’re not real happy that they still have to pay this higher price.
So because of that, this compromise kind of came about, where bid shading was introduced. And basically what Bid Shading is, it allows a DSP to bid in a First Price Auction and if they win, the final price they pay is actually discounted a certain amount.
Now how much it’s discounted really depends on the auction platform. Two of the most common practices that are used to discount Bid Shading are A) splitting the difference between a First Price and a Second Price Auction or B) looking at the historical win price for that inventory or that user and using that to calculate a discount.
So let’s run through an example right here to show how bid price or Bid Shading could be introduced to a typical auction. So let’s say we have our auction right here, and we have two DSPs that are competing in it. DSP 1 bids with a price of $10 and DSP 2 bids with a price of $8. So in this auction right here, if this was a First Price Auction, DSP 1 would win and they would pay $10. If this was a Second Price Auction, DSP 1 would also win but they would pay one cent higher than the second highest bid, which is $8, so they would pay $8.01. So those are the standard auctions that we typically see.
But now let’s take a look at this and apply Bid Shading and how it could impact the final price. So let’s say we’re going off our First Price Auction again and we’re going to use Bid Shading, and our first example of Bid Shading we’re going to split the difference between the First Price and Second Price Auction. So what would happen here is the First Price would close at $10, the Second Price would close at $8.01, so typically what would happen if Bid Shading was applied is that DSP 1 would win and split the difference, so $9. That’s one of the more common ways that Bid Shading is applied.
The second one is to look at historical data. And when I say look at historical data, basically what’s going to happen is the auction platform is going to look at the price that the winning bidder had paid for that particular inventory, if it’s a site or an app, or maybe for that user, and they’re going to apply that price, some type of discount, to the final price. So let’s say in our second Bid Shading example here that the auction platform looks at this inventory and says historically they’ve been winning at a price of, say $9.20. So what they would do then is DSP 1 would win, and they would charge them a price of $9.20.
So as we look at Bid Shading from a DSP’s perspective, it is a good thing in the sense that they end up paying a lower price, which is good. The disadvantage though is that it’s less transparent. And what I mean by less transparent is that the DSP has no insight into how the discount is being calculated or what data is being used to come about with that percentage. So even though that DSP is getting a discount, they don’t know if that discount is as great as it could be. So there’s really no transparency in there.
Now from a publisher’s perspective, Bid Shading has some pluses and minuses. So first off, the negative for a publisher is that they are going to get a little bit lower price, lower revenue, than what they would earn if it was a straight First Price Auction. So if it was straight First Price, they would receive $10, but if Bid Shading is applied they’re going to get $9 or $9.20. So it’s a little bit less, so that’s a bit of a disadvantage, but it’s still greater than what they would get in a Second Price Auction at $8.01.
The advantage to the publisher is because you are providing some type of a discount, and the DSPs know that, it’s more likely that you will have more DSPs that will now compete in the auction. And those would be DSPs that would either have a First Price bidding algorithm or ones that don’t because they know they’re not going to have to pay that full price that they have bid.
So this is the Bid Shading, how it works, the advantages and disadvantages of it. We see it as really just kind of a Band-Aid or temporary solution. The reason why is because it kind of takes a step backwards in regards to First Price Auction. It introduces a lack of transparency back into the auction mechanics, whereas First Price is trying to make it fully transparent.
And the other reason that we see it as being a temporary solution is that eventually all these DSPs, if they want to continue to compete in this industry, they’re going to build their own First Price algorithm. And once all DSPs are bidding with a First Price algorithm, the prices will start to level out and the DSPs will become more used to paying those price points.
So that’s it for this Whiteboard Wednesday. Thank you very much, and we look forward to seeing you again next time.
Interested in learning more about the mechanics of programmatic auctions, in-app ad servers and other related ad tech topics? Be sure to check out these videos: