Header bidding has become the hottest area of ad-tech since first emerging in the market two years ago. But why has it been so slow to play in video?
Header bidding is a programmatic advertising system where publishers offer their inventory up for auction to multiple ad exchanges all at the same time. Opening their inventory to a larger array of bidders, publishers can, in theory, can increase their profits.
In the last couple of years, header bidding technologies have liberated many publishers from seeking ad bids from exchanges sequentially by allowing them to entertain the best bids from each all at the same time. This system has worked magic on display inventory. But the same uplift is not so prevalent when it comes to the fastest-growing segment of ad spending: video.
Part of this blame can be placed on the giants. When the header bidding technique was first brought to the market, the largest advertising companies all commoditized their own header bidding solutions, from Rubicon Project to Google’s Exchange Bidding, which only just launched and is already running into claims of excessive fees.
Of course, the big beasts would want to hold onto their share of the traditional market. But their lethargy, deliberate or not, is not the only reason publishers are being denied the best return. The technology also has to catch up in order to service a video format that is fast becoming the internet’s key driver.
For the uninitiated, a “waterfall sequence” is when publishers seek the best price for their ad impressions by requesting sequential bids at each sell-side platform. You may find the best price from each but selecting the best of all becomes complex at speed and scale.
Header bidding is the solution that publishers most chronically needed in video. Conventional pre-roll prices are high and supply is costly and scarce. New video formats like out-stream units have risen up to compensate, allowing publishers to generate video ad revenue where previously there was no video content at all.
The trouble is that all this growth comes with a speed trade-off. With in-stream pre-roll (i.e., your often contextual video content such as an embedded YouTube video that appears within an article with an ad appearing before the real video starts) viewers tolerate a degree of pre-load buffering. But for an out-stream ad (a standalone video ad that appears on the margins of a page) any delay due to a new piece of technology being tacked on could result in the unit being scrolled past before the ad has even appeared.
No wonder some of the best-known out-stream video vendors get around the problem by treating video advertising not like a marketplace at all but rather more like a private ad network. Rather than subjecting their video ads to the wild west of ad exchanges where any technology goes, they chose to build a direct sales relationship between publishers and buyers.
But that isn’t programmatic, and it doesn’t open bidding to all the benefits publishers can enjoy from a true marketplace. This kind of workaround is understandable. After all, there are genuine technical hurdles to implementing video header bidding – chiefly, the place where the multi-bid processing occurs.
In most current header bidding implementations, that means running header code in the page on a user’s device. Video ads may not have a header per se, but they do have a set of tags (VAST and VPAID) which exist to regulate the display of video ads. The trouble when it comes to video header bidding is that these tags also impose punitive time-out thresholds on ad bids. That means talking to multiple supply-side platforms, a practice that makes header bidding possible, would incur too much latency to make it worth trying.
This is why we are seeing some vendors opt to execute header bidding on the server side. AppNexus, for example, has a server-side switching technology, while Rubicon is also, belatedly, going down the server-to-server route. But since VPAID — whose deployment can help reduce fraud — depends on client-side execution, fully server-side working comes with as much risk as it does benefit. So I think a third way will become more commonplace — a method that combines the best of both approaches.
Publishers should be able to enjoy the protection afforded by VPAID-compliant advertising as well as the speed and choice that comes with server-derived power. I think we will see the emergence of a hybrid approach, which performs the “heavy lifting” on the server side (speeding up the process and reducing latency), all without compromising the client’s resources and page loading time. A hybrid model would still conduct a minimal client-side optimization to guarantee best results.
Video publishers urgently need to step out of the waterfall. Their peers in display are already doing so to great effect. Avenues to deploying header bidding in video may seem less clear. But I think a future is coming into view in which video publishers, too, will get to enjoy higher yields in less time and with continued growth.