New technologies call for new methodologies
Header bidding wrappers are great for Display
Header bidding got off to a great start. If you’ve been hiding under a rock for the past 2 years, header bidding enables publishers to expand their visibility of potential demand by simultaneously auctioning potential impressions to multiple SSPs or other demand sources, thereby increasing competitiveness and in turn, bid rates. Initially the technology involved lots of manual configuration, until it converged into what the industry is now calling “Header bidding wrappers”.
Header bidding wrappers enable users to easily and uniformly manage demand sources and inventory by allowing them to place a single tag on the page or placement, and that tag “wraps” all designated demand sources for that specific placement. This reduces operational costs by managing all demand tags in a single place without the need to modify the code on the page.
There have been several applications to this approach, initially running purely on the client side, and later moving to the server realm, thereby reducing the load on the user’s device. One of the most prominent applications is based on the prebid.js protocol, which standardizes the way demand is “plugged” into a header bidding wrapper.
This is pretty much the status quo for display, and in most cases, it works well. Recent stats indicate that over 70% of publishers have embraced header bidding of this form or another, and most are seeing results improve by 10-50%.
However, standard header bidding wrappers don’t work well with video ads
Video differs from Display in many ways. For one, Video creative files are significantly larger than Display files – as much as 300 times larger: 3-15MB vs. 50KB. And while Display ads can be placed and shown on any browser, Video ads require a player in order to be viewed. Other compatibility prerequisites for video include device and operating system, bandwidth, screen/player resolution, and much more.
But most importantly, the way video ads are transacted is very different from Display. The VAST/VPAID protocols require that advertisers (or, demand sources) commit to an impression only after the video ad starts playing. This means that if an ad failed to load, for any reason, even if the demand source won the bid, but could not send over an ad, the publisher will not get paid.
This makes video very sensitive to technical glitches and to arbitrage plays, since at the end of the day, the publisher ends up losing potential revenue from such lost impressions.
That is why standard wrappers perform poorly with video. In a standard header bidding scenario, as is the case with prebid.js, all associated demand sources are called to make a bid. Then, the decisioning engine chooses the highest bidder as the winner. All of this usually happens on the server side. After the winning source is chosen, its VAST/VPAID tag is transferred to the player to run the ad. However, in many cases, 30% on average, the ad will not play back successfully, and the impression will be lost.
The Hybrid approach for video header bidding works
Cedato created an alternative technology for prebid.js, designed from the ground up for video, and performing dramatically better.
Cedato’s hybrid video header bidding wraps multiple demand sources, much like a standard header bidding solution. This conveys the same simplicity and ease of use, requiring just a single tag to be applied per placement, reducing operational and implementation complexity.
But beyond that, Cedato implements a very different approach. Cedato built much of the optimization elements into the video player itself, working in unison with Cedato’s server-side yield stack. Based on machine learning algorithms that compute dozens of data points beyond just price, the decisioning engine predicts which demand source will deliver and play an ad that will be compatible with and viewed on the user’s device and setup. All this information is stored on a proprietary knowledge graph that enables the optimization engine to predictively pre-qualify the demand per a given impression, without tapping each and every individual demand on the list.
All of this takes place on the server side, to reduce any potential load on the user’s device. However, the chosen ad may indeed opt-out or fail to load. That’s where the hybrid element kicks in. The player itself operates a “safety net” layer that is activated in the event that an ad failed to load, pulling alternate demand sources to make sure that placement will result in a viewable ad.
That combination of applying predictive analytics, taking into account dozens of data points beyond price, optimizing for the best demand, and then applying an additional client-side safety net to recover from VAST/VPAID errors, results in significantly higher results for publishers, increasing their eCPMs and fill rates.
Recapping – why is Cedato’s Hybrid Video header bidding better than standard header bidding for video
– Predictive – practically all header bidding solutions (including prebid.js) are auctioning the impression to all demand sources simultaneously – this requires processing, causing latency, and also incurs excessive QPS-load on the demand sources. With Cedato we only tap demand sources when our algorithms determine they have a high probability of returning a campaign that will result in a viewable impression. This reduces load on both the publisher’s page AND the demand they’re working with, while generating better results.
– Safety net – overcoming VAST/VPAID errors – all generic solutions are server-side, and as a result are missing the point (including prebid.js). VAST/VPAID errors can account for 30% of impressions, and that’s a big number of unfilled opportunities. Cedato’s hybrid solution overcomes these VAST/VPAID errors and can better recover from an opt-out.
– Ease of set up – for video, the header bidding wrapper needs to be compatible with the player and with each demand platform. With Cedato it’s a seamless, single tag process, and everything works in sync.
To conclude – header bidding is great since it generates better rates and results for publishers. Performing header bidding for video is very different than for display – video is more complex, there’s no header, and it requires compatibility between all moving parts. Cedato’s header bidding is hybrid, which enables better optimization (fill rate) and better recovery from video errors, and it is predictive, thus reducing load on both the publisher and the SSPs/Demand partners.
Contact us to see first-hand how hybrid video header bidding would work better for you.