Video viewability is broken - but not for the reasons you think. The way the industry measures viewability does not reflect actual human behavior, and it fails to meet advertisers' real need, which is making sure people actually see their ads. While ad-tech and viewability vendors, publishers, and agencies negotiate what should be considered "viewable" (pixels and time spent on-screen, etc.), actual people are moving on to mobile devices.
For too long, bad actors have profited by running ads in questionable placements, such as pop-unders and in-banner units. Advertisers and publishers have made progress towards viewable-only solutions, but they haven't agreed on a baseline. Late last year, we saw GroupM and Unilever calling for the highest viewability standards in the industry - half the player must be 100 percent in view, half the video must play, the sound must be on for the duration of the video, and the video must be user-initiated. According to Moat, only 22.3 percent of video ads hit that benchmark in Q4.
At the same time, the IAB said that 70 percent viewability should be the goal for 2015 - those viewability goals are based on the Media Ratings Council's (MRC) guidelines of 50 percent in-view for two continuous seconds. The difference between the GroupM standards and MRC standards shows the lack of industry consensus on viewability among advertisers, agencies, and publishers.
Additionally, not all viewability challenges arise from shady vendors and myopic metrics. Ad units and publishers have been slow to keep up with how people use the web. Consumers have become banner-blind and they anticipate and frequently avoid pre-roll video. Ninety-four percent of people skip pre-roll when given the option according to a 2014 MetrixLab, and multi-tasking behavior is commonplace.
Desktop video benchmarks are underwhelming, even against MRC standards. According to analytics provider Moat's Q4 Benchmarks, only 54.6 percent of videos are in-view for one second. For videos fully on-screen, that drops to 43.1 percent. Only 22.3 percent of videos are audible and fully on-screen for half their duration. Television networks have gone so far as to use these numbers to justify their dominant share of the advertising market, saying that they're more reliable than digital providers. Brands have a hard time rationalizing more digital spend when so much of their advertising goes unseen.
Mobile adds more complexity to viewability issues. Consumers spend almost as much time on mobile as they do on desktop according to Kleiner Perkins' Internet Trends Report, and mobile video consumption is growing rapidly. But the technologies that companies use to track viewability on desktop, such as VPAID, don’t work on mobile. So even though mobile plays a primary role in consumers lives, many brands are struggling to move advertising dollars to mobile because their verification vendors cannot measure viewability on phones and tablets.
This is another area where the push for viewability diverges from actual consumer behavior. Most mobile video runs full-screen, rendering viewability issues irrelevant. Facebook and Instagram have introduced mobile auto-play ads, but even these expand to full screen when a consumer engages. With these units, advertisers should be confident that their ads will be seen and that video will perform better than on desktop. But instead of buying based on practical experience, some advertisers will only run where their ads can be verified - even if performance would be better elsewhere.
Brands and their agencies need to focus on getting ads seen by real people. Media professionals should start by thinking about how their consumers behave and where they will actually experience and engage with ads - then find the partners to get the ads there. Standards for viewability will hold the industry accountable on desktop, but ultimately they're backwards-looking - they solve for legacy problems, not the new world of mobile media.
Corey Weiner is chief operating officer at Jun Group, which provides an ad platform for video and branded content.