Today’s column is in response to “Second-Guessing the Second-Price Auction Model” and written by Jonathan Wolf, Chief Buying Officer at Criteo, a buy-side, display ad tech company.
Esco Strong at Microsoft wrote an interesting piece in a personal capacity on this site this week. While I am and remain a fan of Esco, and his piece was elegantly argued, I strongly disagree with it. As I see it, there are two options in building a long-term business: by pricing transparently, or by taking unfair advantage of your customers. Only the first seems sustainable to me.
I am sure that 99% of readers’ eyes glaze over as the discussion moves to auction theory, dynamic floors, and other arcana. So let’s take two simple, real-world examples:
Example 1: “Dynamic Pricing”
Imagine going to the grocery store and buying what you thought were $85 worth of groceries but instead getting a bill for $133. “Why?” you might ask … “Because”, the grocery store replies, “we deduced you’d be willing to pay $140 for those groceries.” Meanwhile the same basket of groceries is a different price for someone at the next aisle, because they were only willing to pay $100. You’d rapidly stop buying from that store.
Example 2: “First Price Auctions”
Imagine you decide to bid for something on an eBay auction. When you choose your bid, you expect that the price you’ll pay will be the 2nd highest bid +$1. Therefore you are happy to bid your maximum price. Continuing the example, if you bid $100 for a tennis racket and the second bid is $50 you owe $51. Instead, imagine if you now had to pay $100 for that racket? You’d be angry, and soon you’ll figure out that you should bid much less than your “true value”.
Laughable you might think, but actually these are both examples of how some publishers and exchanges are selling their inventory. We believe there is a simple rule: if you wouldn’t like it in normal life, you can bet that we buyers of display won’t like it either. Sure, in both these pricing examples, there’s short-term money to be made by the publisher, but it’s at the expense of long-term revenue. And when buyers wise up to being gamed, inevitably you end up worse off than when you started.
It’s my strong belief that dynamic pricing and first price auction models will lead to fewer impressions bought and a reduction in bid prices, reducing growth and revenues for both publishers and advertisers. Yes, publishers are in an environment where competition is less developed than it will be in 12-24 months time, but these are terrible solutions that will rapidly destroy value for those same publishers. This is exactly why most premium publishers use a floor price in situations where demand is not yet mature enough.
As Mike Baker at DataXu wrote recently about these engineered pricing models, “Who wants to have ‘winner’s curse’ and find themselves buying something they could have gotten cheaper by underbidding the true value, but high enough to beat the competition? Perhaps sellers should think twice before they take this approach. This is not a zero-sum game.” I agree. An auction with a fixed floor, as seen with eBay, Google search, or in offline auctions, is the mechanism that creates the most value for everyone.
So … What Should We Do?
I believe two things: (1) a buyer needs transparency about how the publisher determines price (e.g. first price, second price, or some complex blend based on past bidding), and (2) a fixed floor price + second-price bidding maximizes long-term value for the publisher.
As buyers if we get both these things, then we bid full value – and as a result we maximize the size of the business we do – the clicks we send to our advertisers, and the money we spend with our publishers. Because the majority of our over 2,000 advertisers have uncapped budgets this creates a positive feedback loop – and explains why we are the largest performance buyer for many publishers across the world, paying CPMs that have meaningfully lifted their yield.
Esco’s conclusion that we should move to first price seems highly counter-intuitive. The immediate feedback from our business intelligence team, who would have to live with the repercussions of first price, is informative:
First price auctions are a huge pain to bid on, as each bidder needs to spend effort guessing what the second price is, in order to get the best price. If we think an ad is worth $10 to us, we still only want to pay just what is needed to win the auction, whether this is in a First Price or a Second Price auction. If we’re smart we’ll guess that the next bidder is only willing to pay $5, and so we’ll bid $5.01.
This is exactly the same outcome that we would have got in a Second Price auction, except now with all the risk and potential inaccuracy in guessing the Second Price. And if we were wrong and the second price was $6 we fail to show the ad and the publisher makes less money because we should have bid more than $6.
Conclusion: we waste lots of effort building technology to figure out the bidding landscape, that we could have used to improve volume of clicks and conversion rates for our advertisers.
Of course in the real world, one of two things happen:
1. We just significantly reduce all our bids with publishers who behave like this. It takes away a lot of complexity and removes any risk. We will lose some bids, but make more money on everything we still win. The result is that we make a little less margin, while the publisher makes a lot less money.
2. Buyers with very large budgets move to a situation where we negotiate fixed prices with publishers to run our ads ahead of RTB. Dealing with a fixed price buy is much simpler for us to deal with.
To conclude, as an industry we have a choice. We can consider the size of the display industry as a fixed pie, and fight to the death between publishers and advertisers on who captures the most value. Alternatively, we could agree to industry “Principles of RTB” that incentivize buyers to bid fair value and publishers to support this, and therefore focus on applying more technology to display so we can dramatically increase the total size of the display industry.
I vote for the second.
Follow Jonathan Wolf (@jwolf), Criteo (@criteo) and AdExchanger (@adexchanger) on Twitter.
