Open Video Viewability: One Code to Assess Them All?

“We assert that a common interoperable technical approach to measurement is preferable to multiple, incompatible and non-standard technologies.”

Thus reads the charter of the Open Video Viewability (OpenVV) project led by TubeMogul in conjunction with 25-odd ad technology and measurement providers (the latter includes Nielsen and DoubleVerify), most of them household names in the world of digital advertising. It’s a highfalutin way of saying that an open-sourced, agreed-upon technology is a better option than a bunch of quarreling black-box products.

When it comes to viewability, many a publisher can relate to that sentiment. As noted in my recent “How Viable Is Video Viewability?” piece, one of the biggest obstacles for widespread adoption of video viewability is great variability in the numbers reported by the providers (one pub infamously noted a 50-point gap in provider stats).

Variability has also been the most common complaint in display viewability, as we discovered in our “Viewability Litmus Test.” The MRC’s recently completed reconciliation process was designed to bring provider reporting into line by requiring adjustments such as a minimum granularity for viewability measurement snapshots and disclosure of measurement of ad versus ad container (e.g., iframe). While the MRC is reporting that accredited providers numbers are now within a 5%-10% range, some publishers say the gaps are still broad.

But even the MRC’s reconciliation can’t ease publisher indigestion over how viewability has rolled out – a (debated) standard with varying opaque technologies built to meet it. To counter, OpenVV intends to be something similar to the Open RTB protocol that standardized RTB practices across exchanges: a single open-sourced video viewability technology for the entire industry, rather than a host of competing and proprietary technologies spreading confusion and ire. 

The original code was developed by TubeMogul and sits on GitHub, awaiting commentary and updates. The project initially provided data for discussions around the technical specifications for the video viewability standard. This ended up being crucial in the determination of the 50% of pixels in view for two seconds threshold. As it is constantly being updated to reflect the latest environment shifts and trends in measurement technology, OpenVV will not seek MRC accreditation. 

Any video ad platform (e.g., ad server, DSP, SSP) can run the code through VPAID on video campaigns, and some are offering video viewability numbers as part of their reporting. This could prove advantageous as a complaint with campaign-based viewability measurement is small sample size. Since publisher traffic patterns fluctuate greatly by day, OpenVV could allow more holistic measurement of site viewability over time. Of course, reliance on VPAID is a limitation as at best 70% of video inventory incorporates the application; an OpenVV source suggested a code to work with VAST was in development.

OpenVV can work in harmony with third-party viewability providers, thus offering a great research tool to see how different companies’ measurement technologies stack up. As mentioned in “How Viable Is Video Viewability?”, a viewability vendor shakeout is inevitable, with publishers and advertisers siding with the providers consistently reporting favorable numbers. 

Whether advertisers and publishers use the OpenVV code as the measurement tool on which to base viewability guarantees remains to be seen. However, OpenVV is particularly useful for gleaning transparency in programmatic video. The main role of video viewability would seem to be cutting down on a sliver of video ad fraud – not invalid traffic, but bad publisher behavior such as below-the-fold, autoplay, etc. 

Networks and exchanges seem to be the chief accomplices in this fraud, and research from TubeMogul’s video viewability audit powered by OpenVV (To be released in the next few weeks) is already providing insight. According to findings, ad networks on average have far lower viewability rates than ad exchanges, and many ads are being delivered to sites known for auto-play pre-roll. Another interesting discovery: the bigger the video player, the more likely an impression is to be registered as viewable.

In the direct-sold market, OpenVV could play an interesting role considering the current impracticality of guaranteeing against demo and viewability. GRP is the more valuable metric at the moment for advertisers as they can compare them against; so why pay a third-party provider for viewability data if your video ad server or SSP will provide such data through OpenVV, especially if you’re not guaranteeing against it.

(Notable: it does not appear that Vindico, the demand side ad server that claims to serve 40% of all video ads, is a member. Vindico offers its own Adtricity viewability and attribution product.)

OpenVV is another facet in a complicated, contentious space. The video viewability battle will likely rage for many moons, but next-year’s upfronts will be a milemarker in the space’s development – the digital ad ecosystem will be watching.