We syndicate videos as editorial content on some of the most popular publisher sites on the web. Many of these videos feature brands as experts in their category. These videos consistently beat industry averages on engagement. Engagement is measured as “the percentage of the entire video that is watched.” Averages in the industry hover around 50%, ours consistently perform right around 80%. That’s true engagement: someone searched for content, found our video and then watched 2:30 of the 3:00 video. It’s happened more than 20 million times for Betty Crocker videos alone. It turns out a lot of people are looking for information about baking cakes online.
So, you’d think we’d be happy with the data about all the videos we distribute. But we aren’t.
When we syndicate videos to the dozens of sites, they leave our tightly measured ecosystem and enter a measurement black hole. Right now, the way that video is measured favors the exhibitor over the distributor. Let’s look at this in a bit more detail.
A publisher decides if and then how it is going to have its video views measured. The standard right now is comScore. Publishers allow comScore tags to run on their players and every time a video fires, a view is counted. Okay, there are some more technical details and rules, but that is the gist of it. This is great for a publisher who wants to charge advertisers for pre/mid/post roll because they can tell how many times a video fired. The challenge comes when your video is content and not advertising.
The individual publisher knows how many videos they serve. You can see it on their rankings. YouTube dominates this space followed by Facebook as a distant second. But what about a video that is distributed across multiple sites NOT on an ad platform, but in the editorial well of the site. See, the video ad networks run their own players and can count views. But the distributed video runs on the publisher’s player, and the view counts for them and not for the video. So, we can know how many times it runs on YouTube and our own sites, but no one can tell you how many times a distributed video runs. Not comScore. Not anyone.
The question is why not? Two guesses here…
First, is that no one distributes this way. There are plenty of views to be had on YouTube–even though it accounts for only 40% of all views on the web–and the other video syndication sites. In fact, AdAge’s recent article about the success of YouTube celebrities lists the amount of views garnered for brand programs these celebs have run. Their success stories range from 1-10MM views on average. Not bad for a single channel, but well shy of the 20MM-plus that can be achieved through a multi-publisher syndication model. Even AOL’s latest acquisition, 5min, barely garners any views for an individual video. Their average view per video is around 50. We are seeing 5,000 view averages from a syndication model, but it is a chore to cobble together the measurements and no reliable third party source for it.
Second reason is that perhaps someone (probably comScore) is already working on a solution and soon will start measuring content that is distributed across multiple publishers and players. This is a technical challenge that we don’t need to belabor here, but enough smart minds see this need that it will come about eventually. Although we have worked with enough of the big video analytics groups to know that this is still a developing space with far too much of it dominated by TV-ish thinking about video advertising running in conjunction with low budget comedy. As the market–as in the consumer audience–continues searching for and viewing video content over video ads, the marketers will follow and true video measurement will exist. None too soon in our book.
Because as much as Web 2.0 dominates the conversation in most marketing circles, it is still the Mad Men era for Online Video measurement. Which means it is a great time to be pioneering.