Header bidding rose to prominence following the impressive yield results it delivered for publishers when first introduced almost three years ago. Publishers experienced revenue lifts that often exceeded 40% after integrating just one bidder, so, logically, they continued to add more header bidders to increase competition in hopes of driving even more revenue.
By the end of 2016, the majority of publishers were integrated with at least one, and on average four different header bidding partners. And now, that number has more than doubled -- the average publisher has 10 header bidders running on their site, with some using as many as 19.
While the proliferation of header bidding initially mirrored the technology’s original promise - more revenue through increased competition - this rapid acceleration to adopt multiple headers bidders has introduced a series of unintended consequences.
It’s increasingly clear that the impact of these consequences far outweighs the short-term revenue gains some publishers may experience. This new dynamic, including the entry of several knockoff technology companies, a massive redundancy in ad requests, the mixing of first and second price auctions and a series of strains for buy-side players, is unsustainable. Here’s why:
Entry of ‘Knockoff’ SSPs/Exchanges/HBs: There are more than 20 header bidders in market today, ranging from well-funded mature companies to small self-funded start-ups. These smaller companies cannot deliver the ad quality protections that tier-one companies do, nor do they have the resources to dedicate the same level of service and yield optimization to each partner. This results in massive risk for publishers in the areas of ad quality and beyond.
Working with several header bidders facilitates an environment where DSPs get low cost access to publisher inventory. How? DSPs actively pursue “supply path optimization,” which means their algorithms are constantly searching for the cheapest path to inventory. The cheapest path almost always comes from tier-two header bidding start-ups that have not invested in yield technology. If a DSP can choose between 10 paths to a publisher’s inventory, their technology will always choose the lowest price. To the extent publishers deploy low-tech header bidders, this dynamic will materially impact their yield.
Massive Strain on Buy-side Technology: For buyers, header bidding has been a mixed blessing. Gaining access to 100% of a publisher’s inventory was a major benefit, but as publishers’ simultaneous use of multiple header bidders multiplied, the burden on buyers has grown exponentially. They now regularly see the same impression ten times or more, and with no signal to identify where the other bids are coming from, they often bid in each auction, many times against themselves.
Most DSP buying algorithms were designed as many as five years ago and are simply not able to handle this dynamic. As a result of these dynamics, DSP win rates have plummeted without any insight into why, putting their future spend in jeopardy.
Additionally, the cost to process this influx in requests has put a tremendous financial strain on DSPs, causing buyers to resort to blunt instrument optimization where they simply cut off a percentage of impressions from exchanges. Ntoggle, a third-party filtering company that services DSPs, claims that some DSPs are now filtering up to 90% of requests, which is a direct result of artificial supply flooding, and is a losing proposition for all sides.
In short, the buy side of the header bidding/container ecosystem is nearing a crisis state, and there is no scenario where DSPs will tolerate ten times the number of impressions and impression redundancy going forward.
An Imbalance Between Supply & Demand: Over the past year, overall programmatic supply has grown exponentially, between three to five times more year-over-year, while demand has grown an estimated 28%. This mostly artificial flood of supply--attributed directly to header bidding proliferation--against relatively constant demand, demonstrates a significant market imbalance that has had an obvious impact on publishers.
As the DSPs view of supply has mushroomed, the perceived unit value has diminished, affecting each publisher differently and potentially impacting the value of their brand among buyers. For example, if a publisher’s true supply is 500M impressions per month, but buyers see 5B impressions available across exchanges due to header bidding redundancy, buyers will value those impressions less because they are less scarce.
If the current buying and selling dynamic remains in place, where publishers continue to deploy a multitude of header bidders and buyers continue to process an unreasonable number of requests, both parties stand to lose. It’s simply not sustainable. Ensuring a sustainable programmatic marketplace for the future will require the presence of transparent, non-redundant, safe (both, legitimate and high quality) pools of inventory for buyers. And, in order for this to happen, publishers must take a proactive and selective approach to choosing technology.
by Jason Fairchild, CRO and Co-Founder at OpenX