Commentary
2012: The Year Of The Video View Standard
- by Paul Kontonis , December 13, 2011
How do you define a video view? The answer varies depending on who you ask. Is it once the video starts playing? Does it need to buffer? Does a certain amount of the video need to be played before a view is counted?
There is currently no transparency in the measurement of a video view. Despite the rapid growth of the online video industry, right now you can’t accurately compare video views between two programs, even if they are on the same platform. Your 35 million views may not be any better than someone else's 500,000 views.
Why we need a video view standard
The industry needs to ensure that advertisers get what they pay for and that content creators get credit for their audiences. But the current metrics used to measure video views are confusing and misleading. They’re not helpful to measuring success or determining content value. Currently a view can be any type of “play start.” That means that what people are counting as video views could have absolutely no user intent -- the video could be set to autoplay, with the sound off and the player embedded below the fold in a 300x250 ad unit. We have all seen videos playing like this on the Web.
In addition, the current measurement for a video view has no duration component. That means that watching five seconds of a five-minute video is counted the same as watching 10 minutes of a 12-minute video.
All of this leads to a view metric that can be easily hijacked by black-hat behavior from unscrupulous publishers, creators and video distributors. We owe it to the industry to prevent that sort of foul play, because view counts can lead to real money for content creators and publishers.
The kind of standard we need
There’s several things we need to look at when determining a video view standard. First, we need to recognize that auto play does have a certain value to it. Much of online video advertising is always set to auto play and then pushed in front of the consumer –- that kind of view does have value.
But the second thing we need to recognize is that a user-initiated view is more valuable. This is where things can get tricky, because of the varying definition of user initiated. The most obvious example is when someone presses play on a video player and the video starts to play. But how about when someone clicks on a video thumbnail that then takes them to a page where a video player is set to auto-play (with the sound on)? The prime example would be a YouTube video page. These are video views that should be counted as user-initiated.
So here’s what the industry should do. Start with the current metric, where a video view is defined as a play start of any kind. Then go to the number of views that are user initiated, based on a clear definition of user initiated that does its best to limit foul play. (Note: this will definitely cause quite a stir with any company that urgently offers a penny per user-initiated view -- watch how fast they change their offerings).
I would then layer video duration to know how many user-initiated viewers made it through a meaningful length of the video (the 10 out of 12 minutes versus five seconds out of five minutes). The percentages can be debated, but most advertisers certainly want to know how many completed views a video gets if they are considering brand integration.
So the standard could look something like this:
Series 1: Views = 5MM, User Initiated Views = 1MM, Completed UI Views= 50K
Series 2: Views = 1MM, User Initiated Views = 800K, Completed UI Views= 500K
In this example, let’s assume that the length of each video is the same and reaches the same audience. Which show would you consider better to advertise in? Series 1 may be better for a pre-roll buy, and Series 2 would definitely better for brand integration.
So when I chase the dragon and let my mind travel into 2012, I see a video view standard that gets us closer to identifying value in Web video. And ultimately, doing so will help advertisers move more of their budgets to online video. I’ll see you there!
Couldn't agree more with the need to get good data. In fact, as a DRTV practitioner, I've watched so much web advertising get sold with the idea of measurement, but found that there's a massive gap between "it's possible to know everything" (which it's not) and "here's the web statistics we'll give you".
And, since web video advertising will eventually shift to something more like the TV model (appearing within the content), these statistics will become critical for knowing if you're getting any value for your placement.
An excellent suggestion. Makes perfect sense, and we should all hope that standards like the one you propose are implemented soon.
I love your work Paul but you know we're biased against just buying views.
If your ad is successful, a consumer will engage (literally, interact) with it. If the ad isn't right, the consumer ignores, the advertiser doesn't pay.
Don't buy ad views or content views. Buy ad creative engagement.
It's simple :)
Great article. What's comical to me though is the extreme scrutiny of every view in this interactive, opt in space when the dollars are 1% of TV spend. An industry based on a measurement that is 50 years old and consists of a random sample of boxes ( I've never met anyone who owns one in my life) Everyone seems comfortable there even though I haven't watched an ad in 5 years, outside of live sports and I'm not even some tech guy on the forefront. My parents DVR shows as well.
"Start with the current metric, where a video view is defined as a play start of any kind. "
This is not a metric accepted by the online video community, it is a metric accepted by the ad industry an one that should be totally rejected by the community because it is totally meaningless in the context of the actual video.
The Webshoz Videobar system does separate video views into subsets - the first being what we call "touch plays" where a user has started to view a video but left it at 2 seconds or less. We count video views over 2 seconds as partial plays and provide our clients with the number of seconds on average, on those views. And finally, we measure completed plays. We believe that transparency is the key to success in online video use in advertising and so we built our system that way. Visit www.webshoz.com for more information.
Hi there,
I completely agree with you. Advertisers need to know if their message has been seen. It is true for video as well as for traditional display ads.
Alenty has developed a generic javascript that is able to measure effective ad -exposure. Our script is able to measure real visibility of all kind of format (including video) even if the ad has been served in an iframe. We take into account the active window and the mouse/keyboard activity (we stop counting time after one minute of visibility). Our measurement also take into account the % of the area viewed....You can have a look at our live demo: http://www.alenty.com/en/demo/display-ads.
To measure the ROI of branding campaigns we have partnered with market research companies (TNS, Nielsen...). All the post tests we have done have shown that to get the best recall score uplift the message has to be seen at least 80% of its duration.
We are tracking billions of impressions in France, UK and Germany: less than 30% of impressions are efficient ie seen longer than 80% of the message (for a 15s creative in average).
For branding campaigns visibility are N°1 KPI.... and it can also enhance direct response campaigns as conversion could increase by selecting only efficient impressions