Beyond Viewability: Video Content Verification

  • by June 20, 2014
Over the past several months there has been quite a bit of press coverage about video viewability, including two very in-depth articles in The New York Times and The Wall Street Journal. At the same time, lots of progress has been made by verification companies and ad networks to combat all kinds of video ad fraud, bot traffic and viewabilty issues. This is a giant measure of progress, and buyers in particular should feel very good about these advancements.

In the shadow of all this tremendous progress and beyond this issue of whether or not the ads are being seen by humans, there are several issues that are now becoming increasingly important in the video advertising industry. I’ll cover video content verification today and more issues in an upcoming post.

Content and Brand Safety

Currently, most of the knowledge of the content of videos in an ad campaign is limited to the domains on which the videos run. More specifically, content is identified by making inferences from a specific domain. So, for example, if we see an impression on, we can infer that the video we ran pre-roll in front of is about politics or news. In this way, we can have some sense of content, and beyond that what the user watching that content is interested in at the moment of a video view and associated ad impression.

However, this method has major limitations, since many times the site or group of domains in a media plan are not as identifiable. So what do we know about the content when a URL is something like$hf8!jfs? In these cases, it really is impossible to tell what the video is about (for example, technology or politics) or whether or not it’s brand-safe (for example, if it contains nudity, refers to financial disasters, etc.)

Beyond this simple lack of knowledge lies another complicating factor in the mass syndication of video we see today. Even with premium domains or sites that seem familiar, syndicated video players that now run on tens of thousands of sites further complicate the issue by running video that is not produced by that publisher. So on a media plan, those sites that look familiar are actually running content from other video producers, and this content can be remarkably different then what might be typical fare for that publisher. Syndicated players can run in very high rotation and on many pages, making it very difficult to track what’s running on a site from one day to the next.

This isn’t to say that video syndication is bad or nefarious. In fact, syndication actually opens up a great deal of inventory by helping promote videos that have been produced with much skill and thought. It is an extremely positive development for online video as long as the right buyers get connected with the right content -- which amplifies the need to understand the content at the video level.

The general lack of knowledge of content creates issues because advertisers can’t target specific types by interest and beyond that they can’t ensure that the content is safe for the parameters of a brand. So a bank can’t ensure that their ads aren’t running on a video about a financial disaster, and a family food brand can’t tell if their ads are running in front of violent or suggestive content.

This challenge is really a result of targeting that is too narrowly focused on demographics or the inferred characteristics of the user. One of the most important aspects of targeting is what the user is interested in at the time of actually watching the content.  So, if I am watching a video about hockey, the most important thing about me is actually the fact that I am interested in hockey at that moment. It is clearly still important how old I am and where I am located (demographics) but a lack of knowledge about the content is both a loss of a targeting opportunity and major brand-safety risk for all sides of the video ad market.

Real knowledge of actual video content is something that everyone should demand. It will benefit the industry as a whole, as more dollars flow into the ecosystem as context and risk are illuminated and priced accordingly.

Next story loading loading..