Bid shading has returned to the surface as an industry topic recently as more ad exchanges switch to first-price auctions, with Google Ad Manager notably switching to first-price earlier this year. And as more buyers start using it and seeing the rewards, we’ve seen some impressive numbers come out about bid shading’s success. But is it as universally beneficial as it appears to be?
Bid shading is a way for buyers to avoid over-paying in first-price programmatic auctions. But first let’s look at where we’re coming from – second-price auctions. In second-price auctions, which has been the most common auction logic used by exchanges and supply-side platforms (SSPs), each buyer bids the maximum amount they’re willing to pay for an impression given the information available. The buyer which bids the highest amount wins the impression, but they only pay the minimum amount necessary to beat the second highest bidder (e.g, the second highest bid + £0.01). So buyers rarely expect to pay the full amount they bid.
In first price auctions however, the winning bidder does pay the exact amount they bid. So if a buyer bids in exactly the same way in first price auctions as they did in second price auctions, they would expect to end up paying a lot more for the same impressions.
Bid shading is a solution to this, whereby either the demand-side platform (DSP) or exchange/SSP uses information about bidding history to predict the price a given impression might be expected to sell for. The buyer can then adjust their bid so that they beat this expected clearing price, rather than always paying the maximum amount they’d be willing to pay.
Here’s an example of how it works. Take a programmatic auction where bidder A bids £10 for the inventory, and bidder B bids £7. In a second-price auction, bidder A would win, and would pay £7.01. In a first-price auction, bidder A still wins, but would pay £10. If bidder A is buying through a DSP which uses bid shading in first-price auctions though, the algorithm might recognise that the inventory would usually sell for a price of, say, £8. If this is the case, it could submit a bid of £8 (or perhaps higher, depending on how the algorithm is written), meaning that bidder A would win, but would only pay £8.
The most obvious beneficiaries of this are therefore the buyers themselves, and ad tech companies offering bid shading solutions report that they’ve been met with enthusiasm from advertisers. Rubicon Project’s head of global buyer team Adam Soroca said that his company’s bid shading tool, the ‘Estimated Market Rate’ algorithm, was initially conceived as a temporary solution in the transition to first-price auctions. Now however it is being used by over 50 percent of its DSP partners, even in cases where those DSPs do bid shading themselves too.
Socora say’s EMR saves buyers an average of 15-20 percent on their CPMs. IPONWEB, which runs bid shading through its ‘Optimal Price Discovery’ service, says that DSPs using its tool are seeing reductions in media costs of between 15-30 percent. And IPONWEB subsidiary BidSwitch is seeing similar numbers on its ‘True Price’ service, finding average savings for demand partners of around 20 percent across DSPs running the service.
Are there upsides for publishers?
These savings for buyers obviously come at a cost to publishers, but that doesn’t necessarily mean that bid shading is overall bad for media owners. Joe Were, who runs business development at IPONWEB Europe, says that thanks to the Optimal Price Discovery service, demand flowing to SSPs running first price auctions has more than doubled. “So while effective CPMs to those publishers might be lower, total revenue flowing to them is going up.”
Ari Paparo, CEO of Beeswax, said that while publishers may have initially benefited and appeared to be making “excess money” from the shift to first-price auctions where DSPs were still using second-price valuations, this likely would not have lasted, even without bid shading. “It is fairly complex to evaluate this since over time the DSPs algorithms will likely penalise sites that clear at too high prices, resulting in less demand. So an algorithm that tries to bring the demand and supply side into better alignment is probably good for publishers in the long run,” he said.
This net benefit to publishers explains why SSPs, the traditional allies of publishers, are offering bid shading in spite of the fact it helps buyers achieve lower CPMs. “If an SSP’s bid shading technology brings more demand to the publisher then it is good for that publisher, even if the prices decline,” said Paparo.
And alongside buyers and publishers, bid shading obviously has benefits for the ad tech companies themselves. IPONWEB’s product lead Nikita Borisenko believes that DSPs and SSPs which offer bid shading will win out over those which don’t, as platforms without bid shading capabilities “are likely to be optimised out of buys”. And some charge a fee for bid shading too, either through a flat fee, or by taking a cut of savings made by buyers using bid shading tools.
Does bid shading introduce more black box algorithms?
However, that isn’t to say that bid shading has no downsides, at least potentially. One concern is that the practice introduces yet more opacity into programmatic advertising, as it’s hard for advertisers themselves to understand how bid shading algorithms are determining bids made on their behalf.
Paparo said these transparency concerns can largely be avoided if vendors do three things. Firstly the product must be opt-in (demonstrated by the backlash against Index Exchange’s ‘bid caching’ practice last year, where one of the primary complaints was that partners hadn’t been asked whether they wanted it switched on or not). Secondly, vendors should provide stats about the effects of their bid shading algorithm, which ideally would include a control set of bids to give buyers a clear idea of what they’re gaining and what they’re losing.
Lastly, Paparo believes bid shading products must be priced transparently too. “Some DSPs are breaking the third of these criteria, taking a variable fee based on the estimated “savings” of each bid — and I think this does cause a reduction in transparency and trust since they are grading their own homework,” he said.
