9 Comments

Summary:

TubeMogul, a video analytics startup, has tried to discern what, exactly, various video hosting sites measure as a “view.” It’s an interesting experiment with interesting conclusions. It turned out Yahoo Video had the most stringent requirements for upticking the view counter, not even registering any views […]

TubeMogul, a video analytics startup, has tried to discern what, exactly, various video hosting sites measure as a “view.” It’s an interesting experiment with interesting conclusions.

tubemogul.jpg

It turned out Yahoo Video had the most stringent requirements for upticking the view counter, not even registering any views in embedded players. On the other hand, Revver was the least picky, registering any initiated play as a view (though apparently, prior to a procedure change in March, the site had been more strict, only counting completed plays).

TubeMogul concludes that the video hosting sites should standardize what constitutes a video view to reduce complexity and better attract advertisers. Indeed, these inconsistencies indicate this is an immature market.

As for methodoloy, TubeMogul made efforts to test the measurements itself by making videos private and controlling their watching, but in some cases had to ask the hosting companies for additional information. For more detailed assessments, check out the TubeMogul blog, where the full report is to be released later today.

TubeMogul, which we first covered in March, got its start winning a UC Berkeley business plan competition. Competitors include Holt Labs’ Vidmetrix, though there’s still quite a bit of opportunity for others to get into this space.

  1. Very interesting analysis. I always used to wonder, how video sharing sites calculate the view.

    Yahoo way to count 1 view per IP has some flaw. Generally, in corporate environment, company expose only one IP address to the outside world. So, if 100 people from one company watch same video, it should not be counted as only 1 view. Better way to handle this scenario is to count one view, if user is logged in to their account.

    3/4 > view may be counted as 1 view as lot of us don’t wait till the complete end of the video.

    Share
  2. Hey, Alex from Revver here. Just wanted to point out that we track several “kinds” of views. For publicly facing viewcounts (the viewcount displayed on video thumbs/watch pages) we do indeed register/display all initiated plays. We also track and report complete views, as well as advertising stats (and more), and this info is available through the account analytics available on revver.com or through our API.

    Share
  3. Thanks for the post, Liz. This is Mark from TubeMogul. Alex makes a good point which we note in our report, which is that there are some great metrics out there like completions, which is in Revver’s API (and unique to them, as far as we know). Alex, would you be interested in assembling with leaders from the other top sharing sites to help determine standards for the industry? Based the different stats available for tracking in your API, you guys have put a good deal of thought into this already. We think putting standards in place is critical to priming the pump for advertising dollars in online video and will be beneficial for everyone involved.
    Mark
    mark at tubemogul.com

    Share
  4. Mark, indeed – standards are always good, and Revver would be happy to work with others in the industry on this.
    alexblack@revver.com

    Share
  5. [...] constitutes a view?” we asked a couple weeks ago, noting disparities between various video hosting sites as researched by [...]

    Share
  6. [...] June, we asked “Hey Guys, What Constitutes a View?” after Emeryville, Calif.-based TubeMogul released the first edition of this report. We were [...]

    Share
  7. I just tried youtube and let it play to end 10 times and did a refresh and the views stayed the same… figure that one out?

    Share
  8. I did the same and it didnt count – is it because I was not logged on I wonder?

    Share

Comments have been disabled for this post