Sigh. Another day, another huge exaggeration by the media to a small data sample. This time, it's the New York Times. And I guess they have to sell papers, so I understand a certain amount of sensationalism. But I think this goes a little far.
Let's back up for a moment. Ad Age released an article back on September 2 of this year called The Top 10 Viral Ads of All Time. It's an interesting article. Then, on the 13th of this month, they released another article called Book of Tens: Biggest Viral Ads of the Year (Excluding Trailers). It, also, is an interesting read.
But then the New York Times had to go and try their hand at interpreting the data contained in the Ad Age articles, and they spat out this gem: In Online Ads, Celebrities Fall Short. The basic premise is that viral ads starring celebrities don't perform as well as viral ads that have no celebrities in them.
And it's complete hogwash.
Why do I have a problem with this conclusion? Let me count the ways:
1. The data sample is too small.
Online viral video ads have been around for what… about five years? Six or seven if you're generous? Advertising in general has been around for what… hundreds of years? And there's plenty of data to suggest that advertising in general (print, television, radio, etc.) sees plenty of success when using celebrity endorsements. But because the Internet's been around for a couple decades and online video for a handful of years, we should totally draw conclusions based on that tiny data sample (sarcasm hopefully evident).
We need 5-20 years more data before we can say that celebrities don't work in online video. But hey, we've got to sell newspapers today, right?
2. Unfair comparisons.
The New York Times—and Ad Age, where the Times gets their data—is comparing full campaigns against individual ads. The article's shining example of viral advertising success without celebrity involvement is Blentec's series of Will It Blend videos (which, incidentally, is one of my favorite online video series ever). They say, "Look at the top online viral campaign ever… Blentec… and they're not using celebrities." (I'm paraphrasing there, not quoting the article).
But then they compare Blentec's 134 million views (for the entire campaign) against individual commercials like Nike's Write the Future or the famous Pepsi Gladiator ad (starring Britney Spears and other pop stars).
Am I the only one who thinks it's dumb to compare the total view counts of a single ad against a campaign that has more than 120 videos? I hope not.
3. Faulty reasoning.
I'm a bit troubled by the leap in logic at work here. The New York Times is basically saying, "We looked at the top ten videos and didn't see many celebrities, therefore celebrities don't work in viral video ads.” It's just not scientific. It's like waking up every morning and seeing the sun, and then going to bed before the sun goes down every evening… and then concluding that the sun never sets. You can't look at one tiny section of data and draw sweeping conclusions.
4. Manipulated data.
In addition to the article title mentioning that we are "excluding trailers," Ad Age has this disclaimer at the bottom of the piece that also says this:
"This chart does not include movie trailers, video-game campaigns, TV show or media network promotions, or public service announcements.”
Oh, I see. So we have to throw out any of the ad formats that traditionally make use of celebrity endorsements… then we can look at what's left and say that celebrity endorsements don't work. Gotcha.
Also, the New York Times says that the sampling (from Visible Measures, which is where Ad Age got their data) included 180 celebrity-backed campaigns and 270 regular campaigns. What?! So one sample size is twice as large as the other?
5. Biased definitions.
What is a celebrity? Is Isaiah Mustafa (Old Spice Man) a celebrity? He is an actor with other credits on his resume… and even played professional football earlier in his career. Didn't he develop fame with the original Super Bowl spot, thereby making him a celebrity by the time his personalized videos took over the web? How many consumers have to recognize the person before that person counts as a celebrity?
Also, what counts as a "viral ad?” You've already told us you're taking out commercials for movies, TV shows, and video games. So what's left? Is a music video an ad? What about the video made by Jimmy Kimmel's ex-girlfriend (his actual girlfriend at the time) where she jokingly admitted to sleeping with Matt Damon? That got a ton of views… and was clearly intended to promote his show… but is it considered a viral ad in your study? Probably not.
6. Conversions ignored.
Is view-count really the best measure of an online video ad's success? Aren't we really trying to sell products at the end of the day? Isn't it possible that a celebrity endorsement video might have fewer views but a higher rate of conversion?
Audience types are ignored as well. There are different audiences for every company. 100 million views for Pepsi might equal one million views for a small business. Cable TV sensation The Walking Dead is considered a hit for AMC, but it only averaged a few million viewers per episode—numbers that would get American Idol cancelled on Fox. Not every video is chasing the same viewer, so we can't measure one's success against another's by looking at the entire online video audience.
I understand the impulse to look at viral video data and then theorize about what it suggests. I really do. But there are a lot of impulses I don't act on when I know they're not based on logic.
Viral video advertising is just a baby. Can't we let it learn to walk and talk and maybe enroll in kindergarten before we draw conclusions on it? Why can't we include movie trailers and music videos and video game ads, most of which feature celebrities and most of which trounce traditional advertisements in terms of view-counts? Are they not video? Are they not advertisements?
Celebrities can work in online video advertising, and there are plenty of cases to prove that. Do they work as well as other videos? Who knows?! There are approximately a million different kinds of viral clips, with a million different goals and a million different intended audiences. What works for me might not work for you, and vice-versa. The book on what works and what doesn't work with online viral advertising has yet to be written. I'm not sure why we're in such a rush to write it so prematurely.