Hi everyone, hope your summer is going well. I know, it's a non-standard way to start an article talking about statistics, but I wanted to get off on the right foot. Also in that vein, I wanted to give you some information that I've received from comScore about their Video Metrix numbers, to sort of clear the air. See how cool I can be?

Now this information was in my possession for a while but I was unable to check it out due to time and moving etc. Now that things are a bit settled I've had time to look through some of it. There was no mention of any of this being confidential. If it were, I'm probably the last person they would have told it to anyway. Am I right?!

Now a word from our 'sponscore'

  • A video view is counted after three seconds (which I find really poor metrics),  claim to be  reporting ALL videos in the universe [sic] that are streamed longer than 3 seconds.
  • They do use panels, census-based data and claim 100% accuracy in video volumes [sic]
  • 75% of their top 50 properties are calculated using census-based tracking. Which I assume means direct tracking.
  • They have webinars telling their clients all about the methodology etc, but fail to inform journalists to the same level... Perhaps they might want to make that a priority at some point? Well, they probably don't care as most people just blindly accept everything.
  • Claim a far more comprehensive tracking of a company's sub properties which accounts for their higher numbers. Example: They roll CNN into Turner's numbers.
  • Claim major statistics differences (AOL in particular) is because Nielsen didn't roll in 5Min. (see note above)
  • They only release one press release a day, "so as not to overinundate the media." Sounds sort of silly to me really. Their reason, makes the media sound like a bunch of morons... hey wait a second!
  • Apparently, Netflix should show up in the reports at some point. However, their reasons for it not being in the reports directly conflicts with their all videos in the universe statement, BUSTED!
  • They report sessions to mesh better with more of the industry. Still doesn't really clear up their numbers though, does it.

Call to action!

Who out there would rather have actual, valid video views reported than sessions and viewings that last 3 seconds? I mean 3 seconds?! That could certainly account for the really high numbers. How a three second view of a 5 or 10 minute video is a valid metric, let alone if it were say two hours like a film, is beyond me. I have accidentally watched three seconds of a video before getting the browser to return to the previous page...often at Hulu in fact. Do you really want that counted as a video view? If it were my content, I would much rather have a more valid metric for a view. Much like the discussion I tried to start with you in my previous article about what constitutes a view.

A viewing session is defined as a period of time with continuous video viewing 
followed by a 30-minute period of video inactivity.

So really, that could be multiple actual views. It could also include this scenario.

I start a video on Hulu and watch it, then go make something to eat for 20 minutes before beginning another. That's just one session with at least two video views. Alternatively, it could also be that I watch a 100 videos in a row without a half-hour break and still be one session. See why I say it's a bad metric? It doesn't really account for the difference between a webpage or display ad and video...which is the fact that the content changes over time. If you watched the opening credits of a film or TV show, would you say you've seen the show or episode? NO! You'd have no idea as to what the plot was, who was in it, etc. Hence why I presented the time-based view definition.

June 2011 Online Video Sites by Viewers

They made sure to tell me that they track 21 metrics, but only put several in the press releases, so they can make money by charging others to get the full report. It's a standard research method, but begs for some further transparency if you ask me. No, I'm not picking a fight, I just believe that if they want all of us, all of you, to truly believe their numbers, they'd be a little more open in general. I do appreciate the time taken to get me all the information I've shared with you today, but I'm still skeptical, after all, that's my job is to question, not to blindly believe and obey.

So top ten, Google on top, VEVO still #2, Yahoo! alright it's all sort of boring now isn't it. Let's just skip to the chart and then on to video ad networks. Oh wait, Facebook dropped two spots, Viacom/Microsoft flip-flopped and Amazon knocked off NBCU. That must be their Prime program kicking in.

Top U.S. Online Video Properties Ranked by Unique Video Viewers
June 2011
Total U.S. – Home/Work/University Locations
Source: comScore Video Metrix
Property Total Unique Viewers (000) Viewing Sessions* (000) Minutes per Viewer
Total Internet : Total Audience 178,447 6,255,493 1,008.3
Google Sites 149,281 2,311,116 324.1
VEVO 63,003 399,503 112.0
Yahoo! Sites 52,665 247,834 34.8
Microsoft Sites 50,663 286,892 32.8
Viacom Digital 49,493 274,933 76.8
Facebook.com 47,687 167,137 20.7
AOL, Inc. 43,915 251,987 49.3
Turner Digital 30,063 121,301 46.2
Hulu 26,701 156,939 184.8
Amazon Sites 21,247 43,193 8.3

*A viewing session is defined as a period of time with continuous video viewing 
followed by a 30-minute period of video inactivity.

Hulu dropped two million viewers this month NBCU at least 9 to fall off the chart. Probably to Google who ramped up 2M new or perhaps VEVO who got 3M more or Viacom who got the same number.

Video Ads Viewed

Hurray for online video ads and marketing. Wait, what am I saying? Well, I am proud the industry is doing well, I just hate sitting through all of them.

Along with Hulu's 2M lost viewers, went 300M lost video ads this month. Actually some really big changes this month in the chart.

Top U.S. Online Video Properties by Video Ads* Viewed
Ranked by Video Ads Viewed
June 2011
Total U.S. – Home/Work/University Locations
Source: comScore Video Metrix
Property Video Ads (000) Total Ad Minutes (MM) Frequency (Ads per Viewer) % Reach Total U.S. Population
Total Internet : Total Audience 5,286,917 2,286 35.6 49.2%
Hulu 1,001,736 424 38.8 8.6%
Tremor Media Video Network** 753,034 429 12.1 20.7%
Adap.tv† 677,708 386 11.0 20.5%
BrightRoll Video Network** 628,600 369 9.5 21.9%
Specific Media** 421,722 214 6.8 20.4%
Undertone** 332,597 171 13.2 8.3%
SpotXchange Video Ad Network** 281,859 171 7.8 11.9%
Viacom Digital 275,230 134 10.4 8.8%
Microsoft Sites 226,951 125 9.2 8.2%
AOL, Inc. 217,347 85 7.3 9.9%

*Video ads include streaming-video advertising only and do not include other 
types of video monetization, such as overlays, branded players, 
matching banner ads, homepage ads, etc.
**Indicates video ad network
†Indicates video ad exchange

OK so, Tremor is up 53M, Adap.tv is up 36M, BR a whopping 63M. Specific came out of nowhere after not reaching the chart which was 181.7M at its lowest last month. Undertone up 42M, SpotX up 25M, Viacom up 10M, Microsoft down 42M and ABC fell off the chart as did CBS making room for AOL.

See, big changes. Last month the tenth spot was, as I said, 181.7 while this month it's 217M.

Best Places to put your video ads (according to me)

According to these numbers, here's where I suggest you put video ads:

Top Spot: Specific Media: High reach (4th), LOWEST ads per viewer (6.8), good amount of ads shown (meaning a wide network)

Runners up:

  • Brightroll - Highest reach, 5th lowest ads per viewer, 11.9% of all video ads shown
  • Tremor Media - 2nd Highest reach, but 3rd highest ads/viewer,  14.2% of all ads shown
  • Adap.tv - 3rd Highest reach, 4th highest ads/viewer

Really, Tremor and Adap.tv are quite close in reach and ads per viewer as well as percentage of all ads shown. One might fit better for your needs than the other. But it seems that Specific is the place you seriously need to check out. It's a great combination of very low repetition of ads per viewer and a pretty large reach.