Nielsen has released their online video numbers for May 2012 (which directly compete with comScores but are like apples to oranges) in which they say just 163.5 million Americans watched online video. That's some 16.5 million below comScore's estimates (or actually quantified numbers, I'm not sure).

According to Nielsen, YouTube ruled the roost (not a big surprise really, is it?) with 136 million unique viewers or roughly 83% of the total. In terms of streams (whatever that might actually be) Nielsen says there were a total of 26 billion (some 10.5B below comScore's "videos" metric) and again YouTube topped the chart with 16.5B or 63% of them.

Hulu on the other hand was second in the number of streams with just under a billion, but had just 9.5% of the unique viewers showing that Hulu viewers definitely take in a ton of streams each per month. That means Hulu users averaged 62.5 streams each in May 2012 while YouTube viewers cleared around 121.5 streams each. The average streams per viewer according to Nielsen was 160.1 so it must mean that those unique viewers are watching video at multiple sources each month, which is no surprise. My personal viewing spans Hulu, Crackle, YouTube and Amazon.

Surprisingly, the comScore videos number (17.6B) matches quite closely to the Nielsen streams number of 16.5B. Yet, the unique viewer numbers are off by a larger margin 136M to 151.6M.

As far as I'm concerned what this all goes to show is that we, as an industry, need to demand two things from the largest providers of these monthly numbers; transparency and standardization.

The Argument for Transparency

Let's face it, these aren't the most transparent companies when it comes to numbers gathering and if the numbers are supposed to be taken at face value, then it should be readily apparent how they were derived, what they actually involve, who gathered them, who compiled them and who did fact and error-checking on them.

The reason for this is plain and simple, the need for the numbers to be reliable. When they use esoteric ideas like streams as a major metric, clearly there's something wrong and that leads me to the next thing we need to demand. When someone else tallies up video views based on how many advertising breaks there are in a piece of content, there's clearly something that needs doing.

The Argument for Standardization

What is a stream? How does it compare to a view? How many videos are in a stream? How are videos actually tallied? Why are single pieces of video content counting for multiple video views when in fact they're a single program or show?

These are all things that need to be addressed and standardized. A view is not 3 seconds of a video when the video is 3 minutes or more. At three minutes, three seconds is just 1.7% of the whole video. Clearly, that should not be counted as a view. However, it could be counted as a play, as in a user initiated a play of a video. That also means it could not include autoplay videos, so double bonus!

Another major need for standardization comes in the form of how they're collecting unique viewer information. When one says 163.5M and the other says 180.5M it means there's a difference of 16.5M or around 10%. How do sites use that to gauge their actual viewers? Do they average them and say "well, it was probably around 172M," the average of the two? Do we give a range? "Well, it was between 163.5 and 180.5M viewers," which doesn't sound at all like quantitative analysis and more like a guesstimation.

I understand that they've both got their own ultra-hyper-super-top-secret ways of doing things, but really, if they want us all to use their numbers in any consistent and industry-wide useful way, then the numbers have to be more reliable then a 10% margin of error. Any study that put out numbers with that margin would simply be asking for me to tear into them as if I were Ming the Merciless!

Well, for those still reading, here are the actual charts that Nielsen has supplied with their numbers. Here's a link to comScore's May 2012 numbers. I know they hate when I compare them, so I'll expect some email and thou shall know the wrath of Ming the Merciless!

Overall Online Video Usage (U.S.)
Video Brand Unique Viewers
Unique Viewers 163,478,000
Total Streams 26,167,111,000
Streams per Viewer 160.1
Source: Nielsen

 

Top Online Video Destinations by Unique Viewers
Video Brand Unique Viewers
YouTube 136,075,000
Yahoo! 45,336,000
VEVO 42,025,000
AOL Media Network 25,568,000
MSN/WindowsLive/Bing 24,345,000
Facebook 23,159,000
The CollegeHumor Network 22,892,000
Hulu 15,480,000
Perform Group 11,987,000
ESPN Digital Network 11,394,000
Read as: During May 2012, 136mm unique U.S. viewers watched YouTube video. Source: Nielsen

 

Top Online Video Destinations by Total Streams
Video Brand Total Streams
YouTube 16,535,027,000
Hulu 968,007,000
VEVO 727,354,000
Yahoo! 434,329.000
AOL Media Network 330,693,000
Netflix 300,824,000
Dailymotion 229,428,000
ESPN Digital Network 219,172,000
MSN/WindowsLive/Bing 205,403,000
Facebook 120,535,000
Read as: During May 2012, over 16 billion videos were streamed on YouTube Source: Nielsen
Shares
  • Christophor Rick

    keith13k Yeah that we've known for some time and really, to me, it just goes to show that their video "view" numbers are inflated. I addressed that in my articles on the video metrix this month, and in previous months in fact. That's definitely a problem in their reporting, the same goes for whatever the heck a 'stream' is according to Nielsen.

  • keith13k

    So this is from the Comscore press release.  Which means they count "content segments" as multiple videos, which makes little sense, at least to me.  *A video is defined as any streamed segment of audiovisual content,
    including both progressive downloads and live streams. For long-form,
    segmented content, (e.g. television episodes with ad pods in the middle)
    each segment of the content is counted as a distinct video stream.

  • keith13k

    The other funny thing is I believe somewhere Comscore says that a "stream" or "video" or whatever they call it is actually based on the number of content segments.  So if there are 3 ad breaks in a show, that counts as 3 streams.  This may have been fixed but not sure.