How much is a video advertisement view worth? Do we really know? It’s hard to put a price on something when we can’t even agree on what a view really is. Is it, as some say, just 3 seconds of video streaming to a player? Is it, as someone else says, a segment of video content that ends with an ad pod, even if there is more of that actual content to be seen? These are serious issues that are plaguing the online video industry and hampering the growth of online video advertising adoption, some say.
In this post, I walk through some of the current issues and opportunities surrounding online video ad metrics and propose a new measurement standard that I’m calling the “EVV (Effective View Value).” Essentially, this is an attempt to create a metric that provides deeper insight into video advertising ROI.
In a recent interview, with Will Richmond at VideoNuze, Discovery’s Chief Digital Officer JB Perrette, talked about the problem with online video advertising not being one of measuring, but being one of comparison and compatibility in that there is no standard currency that allows one to compare its value across video ad networks, across media, or even in general.
For video ads, I think a view has to be the ad being played to completion, and some agree with me and don’t bill a customer until the video ad has played all the way through to completion. That seems a proper way to go and a good place to begin.
For decades, TV has done its value measurement based on how many eyes are on a show or ad, well really, how many devices are tuned into that content actually. That also seems like a good place to start.
So let’s start.
Building a Universal Online Video Advertising Currency
So we’ve got a baseline for a view but I want to add on to it because of some newer technology that allows us to get far more complex than even a standard TV gross rating point thing.
What is a video view?
A video view is:
A user-initiated play through, to completion, of a piece of digital video content while on the active tab of a browser, in the active window (not in the background/minimized/etc) and in the view port of the browser (not below the fold) or in a stand alone video player that is active during the course of the video play.
This I think is a good start at defining a video view. It does some things, as I mentioned, that TV cannot.
- First, you have a better chance of actually having the viewer looking at the content. Built into the definition is the fact that whatever is playing the video, browser, video player, etc, is the active application as in, on top of the pile, in front of all other applications.
- On top of that, in a tabbed browsing environment, as in most browsers these days, the tab that the video is playing in, must be the active tab while the video plays.
Oddly, while writing this, I am playing a video in another tab and just listening for the keywords in the back of my mind. It’s the actual JB Perrette interview from The Cable Show that Will did. Would that then technically count as a view? No, probably not in my definition. However, the video content in question in this situation is nothing that couldn’t have been done in audio alone in my mind (an interview with little to no B-roll). If there were ads playing on this content would I be able to recall them? Probably not, therefore it would be an ad view with little value because I neither saw nor heard the brand message even though I might have been physically listening.
So you see why I say it needs to be in the active application, on the active tab in my definition. Sure, all of their analytics packages will register my playing the video as a view, but it’s not really a view per se and from an advertising standpoint it would be a near zero value view or at the very least an extremely low value view.
In the interview they talk about measurement being great, but currency being missing. How do we translate data into currency? Everyone needs to adopt the currency, much like Nielsen is, on TV. So let’s begin to build a currency. As with all major endeavors, this is but a first stab at it. People will disagree, state I’ve missed much and want to tweak, tune and more readily standardize it. I’m all for that. This is just my immediate thoughts on the topic (well, perhaps it’s been percolating in the back of my head for awhile and the brewing has just now got to a first taste test state).
Translating Data to Currency
Since I now have a standard basic unit of measuring what a view is, I can now start adding to it in order to create a common currency to place value on that view.
There are certain things that we need to take into account because of the nature of online video which makes it unique when directly compared to TV. Things like:
- Interacting with the video or, in essence, the brand through video overlays or calls to action.
- Being able to scrub through a video to get to the interesting parts.
- Watching a video ad in order to get our video content for free.
So let’s start with that last one. If a consumer is willing to watch several ads, pre or mid-roll, in order to get the video content in question then we can divide the value of the content by the number of episodes they watched. Say a show retails on iTunes for $1.99. If I go and watch that content online and view two video ads in order to get it for free, each video ad has a basic value of $1 (for simplicity). That’s the clearly defined value per ad from the consumer’s standpoint because, in my hypothetical situation, the viewer would surf away at the beginning of a third ad. Clearly they feel that the $2 to buy it instantly is offset by the time it takes to watch two video ads.
So now we have:
The value of video content is based on the perceived value of the content in the eyes of the viewer. Therefore an advertisement is worth
video retail price / number of ads to view video in its entirety for free
Example: A TV episode is available online for $1.99. A viewer is willing to watch a total of two advertisements in order to see that video totally free. $1.99 / 2 = $1 (rounded).
For a place like Hulu, a single TV episode can have 10 ads attached to it and therefore each ad would then be worth $1.99/10 = 20 cents. If you were to put that into a CPM it would be extremely high. $200 CPM is crazy expensive! Who knows, perhaps that’s actually Hulu’s rate (they don’t release that info unless you’re actually going to advertise with them).
The Museum of Broadcast Communications has some awesome math on TV CPP vs. CPM. Basically, they say that if you bought 10 ad spots on a show with a GRP (gross rating point) of 11 (11% of the number of TV viewing households, or roughly 10,300,000 in their example,) at $150,000 per spot it would be $1.5 million with a CPM of $14.56. It’s pretty low compared to my formula above.
The great thing about online video is that we have far more data, as JB said, we’re drowning in data, so we can tweak all sorts of things.
Learning to Swim
I have to admit, even I’ve been buried in data from my web analytics at times, who hasn’t, right? So a way to boil it all down to a universal currency, or rating, is the Holy Grail of online video advertising and perhaps even advertising in general.
We’ve looked at the cost of putting an ad on TV and we’ve looked at perceived value of content from the viewer. How do we reconcile the two?
Demographics can be used as a modifying factor.
Say you’re looking to reach 18-24-year-old males, as that’s who is mostly interested in your product. Each view that is within that demographic would have a value of one. Each view that is outside of it, would have a value of 0.5 as they are far less likely to be interested in the product, but not 100% uninterested because they perhaps have an 18-24 year old on their shopping list or whatever. So we don’t assign them a value of zero.
Now, with online video advertising, we also have a slew of interactive elements that can be used to modify the value. Obviously anything over effective view value (EVV) of one is positive and anything below is negative. Interactions that lengthen the viewer’s interaction with the brand are positive, actions that include abandonment of evasion (skipping an ad, choosing another) are negative.
Positive modifying examples:
- A viewer clicks the ad (pausing it) and loads the website in another tab or application.
- A viewer requests more information through an interactive video overlay.
- A viewer fills out a survey, form, or signs up for a newsletter.
- A viewer requests to be contacted about a product.
- A viewer purchases a product online, directly from a buy now link in a video advertisement.
Each of these would most likely have a different level of value for advertisers with the conversion being the most valuable. If that were to have a multiplying effect of five then it would even give the view from a non-targeted viewer a positive score of 2.5. The targeted viewer would still have a much higher value, five, which accounts for other things like potential for repeat business (as they’re still most likely to buy more products or additional, complementary products), higher brand engagement, etc. while the viewer outside the demographic might have been just a one-shot deal, or infrequent as in perhaps once per year events.
Negative modifying examples:
- A viewer clicks away from the ad and does not finish it nor return.
- A viewer chooses to skip an ad.
- A viewer chooses another ad over an ad.
- A viewer switches to another application or tab while an ad is running.
- A viewer moves the video player beneath the fold (out of the view port) while the ad is running.
- A viewer interacts with other elements on the page while the ad is running.
These could all be assigned various negative modifiers. Say the viewer abandons the content when the ad is running, that could be a negative multiplier of five. So when your high-value target, the demographic you want, does so, that ad view is rated at minus five. However, if a non-targeted viewer does so, it is only minus 2.5. Again, the value seems appropriate because clearly something is amiss when your main target audience is clicking away. However, it could just as easily be a problem of the content and not the ad or targeted demographic so A/B testing would be required to determine the cause. Or you might then re-target that viewer with a survey to get some feedback on the ad or find out why they abandoned. Those elements could then also modify the final value of the view if you find out that they didn’t want/like the content and the ad was not faulted.
Creating a Reel Effective View Value (EVV) or Universal Currency
To create an Effective View Value (EVV) we could include a variety of metrics and interactions as a base and each advertiser could then customize their EVV formula based on what it is they are trying to accomplish.
Clearly interactions with the brand or product would be valuable and abandonment and evasion would be negative.
Effective View Value (EVV) Formula
Right now we’ve got a formula that looks like this:
EVV = (((target audience value)*(positive interactive modifier)*(negative interactive modifier))
The base CPM might be $25 but because 50% of viewers were outside the target audience, those views are rated at 0.5 and so they are only $12.50 CPM. If 100 of those 500 views ended up in conversions they would be 2.5 and worth $37.50 CPM (or $3.75 total), but since it’s just 1/10th of the thousand it balances out. The other 50% were in the target audience and so have a value of one. 100 of them also ended up in conversions giving them a value of 5 so those 100 views are then work $125 CPM (or $12.50 total).
Just from those numbers we have. $3.75 (for 100 unexpectedly good views), $12.50 (for 100 great views), $5 (400 non-targeted views), $10 (400 targeted views). The EVV of that 1000 views is then $31.25 making the $25 CPM a deal. Multiply that by a million views and you see that you would pay $25,000 but, if the numbers held steady, it would be worth $31,250 – a savings of $6,250.
That would then mean you’ve got an excellent placement there. However, if you were paying $25 CPM and the EVV rolled out into $15, your ad placements there would be hemorrhaging cash to the tune of $10,000 per million ad impressions.
For advertising CPM purposes the views could be sold based on the +/-5 EVV which would allow brands to do conquest or complementary video ad placements with a chosen ‘preferred demographic’ to assign the initial ‘target audience value.’ Then, based on the actions of the viewer, the effectiveness could be determined and price assigned.
More Data Points and Factors
There are numerous other factors that would need to be involved. For example, data tracking purchases at physical retail shops could be done as well as other things like social sharing of information, repeat viewing, etc. Again, because of all the tracking data we’re swimming in, it could be easy to start drowning. The key will be to create a single, cohesive, and easily adaptable score. While it might not include every single factor it could really help us hone in on the true value of online video advertising and help compare it to other advertising forms.
For example, you could take your TV ads and, through several other data services, come to some very similar EVV number that you can then compare. Obviously, TV would be at a slight disadvantage right now because of the lack of interaction in the ads. But they are rapidly moving forward with things like Shazam, QR codes, and connected TV. It could soon come to pass that your TV ads are effective or even more effective than your standard online video advertising.
I think that by trying to build a bridge across all of the media formats advertisers could get a far better glimpse into their ROI and know exactly which formats and ad placements are positive and which are negative. Again, this is the a starting line for a discussion which is going to be a marathon, not a sprint. There are many things to be considered and included and everyone will have a specific tweak here and there they’ll want to include. The major thing would be to get a single formula that the majority can agree upon and then begin working on expanding it from there over time to get a fully standardized solution.