• by September 23, 2014

One of the lesser-known facts about the video advertising industry is that the ad ecosystem’s knowledge of the content is limited to literally just the page URL that the video is being viewed on. This is in part due to the business model that online video grew up with, which centers on traditional demographic ideas: If I can get the right audience, why do I even care about the content, since I’m getting the right user? This kind of approach might work in display, but it’s actually missing a huge piece of data in video.

The most important thing about that user in his or her “viewing moment’ is simply the fact that he or she has decided to watch that piece of content. That interest data has more bearing on the success of a pre-roll campaign then any generic piece of information.

This makes sense thematically: if I’m interested in a sports video, then it reveals my general receptiveness to sports-centric ad products like Nike or Under Armor. The data illustrates this perfectly. When we use the same inventory and simply route the right ads to the right content (and remove the non-brand safe and non-viewable content), we see all the traditional success metrics like completion rate, click through rate and engagement rate all increase 1.5X to 3X. These kinds of lifts only work when we match the actual video content (not the text content or some approximation or guess about the domain), and it happens consistently.

For example, an A/B test campaign demonstrated the value of true video targeting for the home and garden category:

A - with no video context targeting, the completion rate was approximately 25%

B - with specific DIY and Home & Garden “viewing moment” video identified, the completion rate was 81%, with NO increase in inventory price to the customer. The price per completion was cut by 70%.

In another example, a travel/tourism campaign experienced a 1.9X jump in engagement rate after “viewing moment” targeting, with NO increase in inventory price to the customer.

The importance of this data cannot be understated. We’ve found completion rates on highly targeted campaigns have gone from around 30% to around 80%. As a result, we drastically cut the cost/completion, and that value gets picked up by the customers in terms of broadly increased effectiveness.

The beauty of this approach is that the data is always there, since the content is what the user sees consistently. Demographics are not a guarantee, since we don’t know if we will be able to identify a user before they get to a video -- and furthermore, this kind of targeting can be used in situations where privacy is a premium and/or cookies are blocked. This is due to the fact that the targeting is based on the content and so all we need to know about the user is that they are having a viewing moment of that particular video.

This isn’t to say that demographics aren’t important, as the audience one reaches with a message is of critical importance. Ideally both demo and content data could be taken advantage of to provide maximum engagement and completion results (content) to exactly the right people (demographic).