• by Brett Wilson , February 10, 2014

Did someone actually see my ad? That's a simple question, but as many marketers are discovering in display, the answer is often complicated. If the industry does not act soon, video may suffer the same fate.

Some have argued that viewability is not a big issue in video given that it is a more captive medium than display. After all, if a viewer made the effort to click on a video, all but the most unengaged will see the in-stream ads during playback, right? Not always. Viewers sometimes avoid video ads by muting them, or by opening another tab or window to wait them out (admit it: you’ve done this). The more pervasive issue, though, are video ads that viewers did not intend to watch, often taking the form of fake pre-roll that autoplays in small players on a page.

Several ad tech companies in the space claim to have proprietary solutions for video viewability, but methodologies vary. Some ad servers, for instance, offer viewability reporting that spans both display and video, but their technology is primarily built for display. The result is reporting that only works on a tiny subset of ads and is often inconsistent. Another example is a large video ad network that offers a viewability solution that uses its own definition of viewability and does not cover muted videos. This begs the question: is a video ad better seen but not heard? In the end, proprietary solutions are just that: proprietary.

Even if you are convinced that you have accurate reporting, it is often difficult to diagnose why a video ad is not getting watched. Currently, most reporting in the market only tracks publisher-level viewability, when what you really need is domain-level visibility. For instance, a broadcaster might have both full episodes and syndicated content among its inventory, with varying levels of viewability, but all an advertiser sees is a large chunk of unviewable inventory. Lack of transparency in the industry makes this problem even worse.

Other metrics like player size, inactive windows and mutes can also help isolate unviewable inventory, but are not always available.

But there’s still he elephant in the room: the majority of video ads on exchanges cannot be measured for viewability at all. There are two main reasons for this: either the technology purporting to track viewability does not cover all of the different video player types, or a video ad is loading in an iFrame. The bulk of unmeasurable impressions come from iFrames, which typically mask where an ad is really running and block domain-level visibility.   

The result is a state of confusion. The solution is not as simple as expanding the display standard to video. Videos are not an image to behold as much as an experience. Should an impression be counted if a video is viewable during the first second of playback -- or when the video completes, for instance? Then there is the question of unmeasurable impressions. Are these inherently unviewable, or should they not be considered in calculating viewability? Approaches vary there, too.

Head spinning yet?