Video view numbers are still misleading
We've known for a while that when it comes to counting video views, it's mostly voodoo. Everyone who writes about online video, including me, is guilty of quoting view numbers (e.g. wow, the latest sexual health video from illumistream has 110,000 views in 2 days!) without adding any kind of caveat about their statistical wobbliness.
On Wednesday, online video analytics company TubeMogul will be releasing its study of what counts as a view across a dozen video sites. The basic finding is that with only a few exceptions, all a view means is that a video begins to play. Notice how I didn't say it means "someone began watching" -- watching has nothing to do with it. On YouTube, AOL Video, MySpace Video, Yahoo Video, Veoh and Revver, a view is counted as soon a video begins, meaning tick marks are handed out even if viewers didn't watch a single second. It doesn't matter if your video is two minutes or two hours -- if it started playing, it was viewed.
YouTube just launched its Insight tool, which allows video makers to see certain geographic and demographic information about who's been watching their clips. But conspicuously absent is any information about how much of each video viewers are sitting through. If I were a YouTube regular, I'd want to know if people were watching all, half, or just a few seconds of my stuff, even though they all counted as an equal view. Remember, Nielsen television ratings are based on making people click a remote control button every few minutes to show that they haven't left the room. That way, advertisers know the viewership numbers aren't inflated.
With online videos, though, using the loosest possible definition of a view means viewership numbers are goosed by definition.
Who do you think benefits from inflated view numbers? Viewers? Makers? Advertisers? The video companies? Leave your thoughts in the comments below.
| Bookmark it: |
TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d8341c630a53ef00e55255045c8834
Listed below are links to weblogs that reference Video view numbers are still misleading:
The study also pointed out that two companies - Metacafe and blip.tv - do take into account the amount of time a clip js watched and the IP address before counting a view. These sites are the most conservative on the Internet in this respect according to tubemogul.
Posted by: mark | May 28, 2008 at 04:53 PM