HEADER BIDDING IS LEVELING THE PLAYING FIELD BETWEEN PUBLISHERS AND EXCHANGES
Header Bidding – the new black, taking the market by storm, the must have, top of the trends.
So why is it such a threat?
Header bidding shifts the power dynamic away from the SSPs and exchanges, moving it back to the publishers. Publishers who manage their own header bidding have created what exchanges promised to create for the last five years: a highly competitive auction for a publisher’s inventory. Not only are publishers making more money with header bidding, they are gaining a bigger advantage, transparency of demand, and inventory control.
Who is winning – and who is losing – thanks to header bidding? Why do I believe it ultimately spells the demise of many exchanges?
THE QUICK HISTORY:
- Digital advertising was born.
- Digital advertising was sold much like TV, print and radio, mostly on a guaranteed basis via phone calls, faxes and emails.
- Digital inventory exploded and “remnant” (unsold) was born.
- Remnant was sold via networks, used for house ads, or went unsold.
- Networks figured out how to create vertical networks and other aggregation methods, providing reach and contextual targeting to buyers.
- Networks started deploying audience targeting tools and were the first to really leverage data (with the exception of Google, who always knew how to leverage data).
- DMPs and other data companies emerged, created huge pools of profiled cookies.
- SSPs, Exchanges, and DSPs emerged to create the ability to “real time target and bid” on remnant inventory.
- DSPs leveraged the audience data to drive performance and the size of the RTB market exploded.
- Publishers were pushed to “be transparent” with their inventory and add higher quality inventory to what was sold programmatically, increasing eCPMs of exchange inventory, but not necessary the overall eCPM of publishers.
- Exchanges, and DSPs starting making some serious money (each taking about 15% of the spend) creating the so called “AdTech Tax.”
- Meanwhile, companies like Criteo realized that being in the header gave them first look for all inventory (not just unsold), driving up eCPMs for publishers while cherry-picking inventory.
- Technology-learning publishers began to experiment with “header bidding,” often leveraging DFPs’ dynamic allocation tools.
- Open source header bidding wrappers and adapters became available and the market exploded.
- Header bidding began driving up CPMs for publishers, allowing them to capture top dollar for high-value inventory.
Header bidding is exploding because publishers reason (often correctly) that they are getting screwed by the second price auction and the intermediaries who seem to be rigging the system. With header bidding, publishers are able to bypass the the waterfall of exchanges, increase their overall eCPMs, gain more control of their inventory and only use their exchange(s) (mostly AdX) to sell what is truly now “remnant”.
Why are the exchanges doomed? First, they’re not all screwed. Some were losing under the “exchange daisy chain market.” They were the third, fourth, or fifth call on the chain and saw lousy inventory that had been picked over by AdX, AppNexus, Rubicon… Certain companies, in particular OpenX and Index Exchange, made early big bets on the header bidding model. All of a sudden, they were seeing 100% of a given publisher’s inventory, and their volumes and eCPMs increased. (Yes, they also incurred a “listening fee” for 100% of the inventory, but top line exploded.) AppNexus jumped on the bandwagon a little later, with their open source solution, prebid.org. However Rubicon was late to the party, as was Pubmatic.
Rubicon has reported header bidding as the primary culprit in their recent earnings calls. One need look no further than the recent stock price collapse of Rubicon. It has been blaming header bidding for its sinking performance for the last few quarters, but was overwhelmed with bad news with its recent reporting (which resulted in hiring Michael Barrett, a consistently successful CEO (successful at selling the companies he walks into).
Most importantly, publishers, particularly the comScore 150, were figuring out how this was all working, and began taking control of the process. Now everyone is talking about server-to-server integration for header bidding. This seems to overcome the real technology problems of header bidding by creating a single header bidding call and having someone else then manage the bidding process. Usually this “someone else” is the exchange. They have the technology infrastructure needed to manage the extraordinary volume created by this system.
Server-to-server also creates a new problem related to cookie matching, but let’s assume this will get resolved over time. The biggest part of the challenge is exchanges’ unwillingness to cookie sync.
Shenanigans, deception, and more of the same-old-same-old.
While server-to-server integration clearly solves most of the problems associated with the speed and load times of header bidding, it leaves publishers exposed to different problems. One traditional complaint of publishers regarding exchanges was that the actual fees charged by such intermediaries were less than transparent. Also, publishers suspected that, in some cases, exchanges were front running their inventory, creating additional spread within auctions, adding or subtracting data for their own benefit. Publishers felt cheated, but were not certain whether it was true.
Over the last few years, as more bid stream data has become available, these suspicions have sometimes been confirmed. Certainly, some deceptive practices have been identified. As with most things, transparency is the greatest disinfectant.
The problem with header bidding managed by the exchange is that it opens the ecosystem back up to the suspicions of self dealing. In one instance recounted by multiple publishers, some exchanges seemed to win a disproportionate amount of inventory when they were the “wrapper” compared to when they were only an “adapter” inside someone else’s wrapper (i.e., when they manage the whole auction they win too much, compared to when they are just another bidder).
Simply put, server-to-server, at least as most people are discussing, is basically a publisher selecting one exchange and giving them all their inventory. Now they are not just seeing the “unsold,” but 100% of the inventory (like all header-bidding). However publishers are opening themselves up to many of the same transparency problems of yesterday’s exchanges.
Why is this a threat to the exchanges? Simple – the biggest publishers with the highest quality inventory don’t need to use an exchange. Companies like Purch have already tackled this themselves, and we are seeing the emergence of other companies stepping into the breach to either support publishers deploying their own server-to-server deployment, or creating new pricing models so they do not have an incentive to play with the auction.
I suspect the biggest and best publishers will migrate to a fully-transparent server-to-server model, which will force the exchanges (who are used to working of a healthy percentage of the media spend) to become more transparent, change their pricing, and provide different levels of service. Publishers will jointly create a cookie-syncing solution, and eventually attract the largest DSPs to bid directly, and not through an exchange at all.
Exchanges will be forced to provide their services to the mid- and long-tail. While some can survive, many will fail. The winners will likely be the exchanges already focussed on the long tail (think SOVRN) versus those competing for the comScore 150.
There are many exchanges in the market, some focused on a more vertical approach (think mobile or video) and others providing higher-quality service and trying to differentiate in other ways. However, the biggest exchanges were and are “display first,” and their business models will come under increasing pressure as header bidding technologies become more ubiquitous, and as the expertise to deploy one’s own solution extends further into the market.
These trends don’t fix many of the other problems with the programmatic marketplace (fraud, walled gardens, viewability, effectiveness…) but this shift is going to return some of the pricing power back to independent publishers. That can only be good for the industry.