Viewability – whether an ad is viewable or not – is an issue that continues to plague the video advertising industry. And make no mistake – somewhere out there there’s a parallel universe where the industry cracked the problem and saw significantly more big brand money flooding in. One of the tech vendors working to crack the problem is Veenome
, a video ad tech company that also offers brand safety and video categorisation solutions. Here Kevin Lenane, CEO of Veenome
, discusses some of the worst examples of non-viewable ads he has seen, his company’s new RTB solution, and how the industry is struggling to defining and solve viewability.
There are numerous definitions of video viewability. What is Veenome’s definition?
We see video viewability as just a piece in our greater mission of transparency in video advertising analytics. The viewability component consists of the position of the player on the page usually represented by x and y coordinates, the size of the player in pixels, the muting state and the auto-play state. Looking at each of those in more detail, I’d break it down as follows:
Traditional Video Viewability Definition
Tracking a user session enables the measurement of the the amount of video that is actually on a user’s screen; that is, how much of the ad was viewed. However this measurement on it own is incomplete because it fails to take into account several important attributes listed above. So why are these other attributes important?
First, the size of the player is of critical importance. A large percentages of videos we see as part of our automated scanning (non customer centric) are at the 300X250 size. These videos which are often premium content, can play in view and be easily ignored. However, based on the traditional definition they are equal to a full-view NBC video, for example, regardless of the diminished attention level of the consumer.
The same holds for position position – videos playing on the right hand side in the mix with other animated rich media ads are not garnering nearly the same attention as something large and in center screen. So while it adheres to the traditional definition and may be technically playing on the screen during a real user session, it simply gets ignored. In addition video playing below the fold on auto-play can be detected through position and auto-play attributes.
So the traditional takes care of one of the most flagrant viewability violations which is videos playing deep below the fold on auto-play which is great. However the presence of auto-play certainly indicates user intention and should be detected and reported regardless of whether the video is above the fold or below the fold. For instance a 300 X 250 px video on the right hand column of a page on auto-play can be simply ignored like any rich media ad and hence represents lower value. However a user–initiated impression on the right hand side represents a lot of intention since the user actually requested the content despite the small player size and fringe player position.
Finally, the detection of multiple videos completes the picture. The importance here is that it eliminates a new form of viewability evasion that we see now that below the fold auto-play videos are being detected. In the traditional definition, you could place two videos on auto-play above the fold. So you might have one large size in center and then one off to the right hand side at a 300 X 250px size. They’ll both play on auto-play and, although the user is clearly not able to consume both ad impressions, both are in-view and both count as impressions according the traditional definition .
Knowing all these attributes just paints a fuller picture of the content landscape and how things are being presented to the user. If you have player position, player size, auto-play state, muting state and a video count then you have a total picture of how impressions are relayed to users – this data can be utilized for tracking and targeting at the individual impression level and at the domain level as publisher averages.
Beyond viewability, the data on the content of the videos that come after the pre-roll is the final piece of reporting needed to provide complete transparency to the video advertiser or agency.
What are some of the worst examples of poor view ability you’ve encountered?
Oh man, well we see everything, especially in our new Real Time Publisher Index, where we are scanning the top ~1000 video publishers daily to gather domain data. Through that product’s QA terminal, I’ve seen pages with four 300X250 videos on auto-play on the right hand side. These videos would actually stopped and then started again continuously such that if you left the page open, it would generate 4 impressions in view every 30 seconds – that was kind of incredible.
You’re launching an RTB product soon. How will that work?
We’re very excited about this one. The Veenome Real Time Publisher Index is a subscription-based service that delivers ‘publisher indices’ in real time based on a series of individual page URL impressions, delivering actionable data about the website’s video content. Every day we automatically scan thousands of the most viewed video publishers, many of these publisher pages contain unique video which is then indexed by our video categorization and viewability products. The data from each URL is compiled into ‘publisher indices.’
The Real-Time Publisher Index attributes include the following domain-level averages: player size, player position, auto-play percentage, multi-video page percentages and the categorical breakdown of the video content included brand safety factors. Subscribers to this data can call and store this data in real-time via the API or in daily batches for use in RTB environments.
Is the industry doing enough to tackle viewability? What’s the solution?
I think its moving along at a decent pace but its also tricky because its truly a moving target. Even now as the IAB starts to settle around a traditional definition – we see folks starting to skirt that definition even before its been ratified by fiddling with multiple videos on auto-play above the fold and stuffing more tiny players in the right hand column. The best way to approach these kind of truly dynamic issues is to simply understand that there is no magic bullet definition or detection and that providing as much transparency as possible is always what makes the picture the most clear.
You recently announced a partnership with Innovid. Could you provide a little background on how you’ll be working with them?
It’s an interesting partnership for us. Basically we are providing our viewability and content products through their ad server product to produce reporting for agencies and brands. We know that brands are very interested in the information that we provide but we haven’t had a method of providing them the data in way that make sense. We don’t necessarily want to introduce another data stream that the advertiser will have to manage so the partnership with Innovid is ideal as the advertiser already trusts them to create and serve their ads. We can simply pass our reporting to Innovid who has vetted our products and then Innovid can provide them to their brand and agency customers along with the myriad other kinds of reporting they already provide. This just makes things cleaner for the end customer with a single source of analytics. Their organisation is chock full of super smart developers and product managers and they’re great to work with.