Online Video Advertisers: Enough Double Stuf, Time to Get Targeted
What a wonderful world: Thanks to Fox, Comedy Central, and the other TV networks that are generous enough to post the latest episodes from their hit shows online, I can watch all of my favorite episodes on the Web on demand without paying a dime to Comcast (oops, “Xfinity”) for premium cable channels or DVR rentals. It’s hardly a case of altruism, of course. The networks make money on online video by selling so-called “pre-roll,” “post-roll,” and “in-stream” ads. In exchange for all that free content, I’m happy to sit through the ads—pretty much.
There’s just one thing that puzzles me. Having watched quite a few online episodes of 24, Glee, and Fringe (does it say something disturbing about me that most of my favorite shows are on Fox?), I’m getting pretty tired of watching the same ads over and over again. Literally the same ads. During a single episode of 24, Fox will show me a single 30-second ad eight or 10 times—at the beginning, at the end, and at every commercial break. Lately it’s been Red Bull, Double-Stuf Oreos, and Bioshock 2 ads. Evidently the network has me pegged as a twenty-something software developer with a cookie and video game habit.
I don’t mean to second-guess video advertisers at their own business, but it strikes me that showing the same ad over and over to someone who’s never going to buy the product is not an optimal use of all of the advertising time that goes along with a 43-minute TV episode. In fact, it can be counterproductive. By the time I’ve watched the same Red Bull ad 10 times, I’m so sick of the company that I’m even less likely to buy their sugar-and-caffeine concoction, even if it does promise to give me wings.
Other networks seem to have slightly more on the ball when it comes to online ads. I was intrigued the last time I logged onto Hulu, and it asked me before 30 Rock which of two Southern Comfort ads I preferred to watch. But after that one spot, the rest of the ads during the show were non-interactive and were chosen, as far as I could tell, at random. Not only that, but the 30 Rock ads looked to me like the same 30-second spots that the ads the networks show on broadcast TV, just repurposed for the PC screen.
So, here we are in 2010. I’ve got a 30-megabit-per-second Internet pipe into my house, and my browser is loaded up with glitzy interactive video software like Adobe’s Flash Player, and the cookies on my computer probably know me well enough to guess my toothpaste brand. But to the network advertising executives, it’s still 1975. Whatever happened to all the talk about targeted advertising? I could use a little bit of it right now.
“The dirty secret is that most pre-roll is not targeted,” says Bill Day, the CEO of ScanScout, a tech startup that buys ad slots from a network of 1,000 online video publishers and sells them to advertisers. “The ‘targeting’ is literally as simple as, ‘You’re a premium brand, so whatever you give us, we’ll assume it’s of interest.’” No wonder Hulu keeps showing me a Dove soap ad, even though I prefer Dial.
Day says ScanScout is working to make online video advertising work better for everyone—advertisers and viewers alike. “Our clients suffer from this incredible problem,” he told me when I visited the company’s headquarters in Boston’s Leather District last week. “People are becoming better and better at tuning ads out, while at the same time ad frequency keeps going up and up. So the question becomes, how do you create advertising that is interesting to users?”
Google solved this problem in the realm of text ads long ago. First, it realized that people searching terms like “kitchen remodeling” are more likely to be interested in ads for Home Depot than for Travelocity. Second, it introduced a payment system so that Home Depot only pays for the kitchen-equipment ads when people actually click on them. Everybody wins.
“We see the potential for a similar model evolving” in online video ads, Day says. How would that work in practice? You can see for yourself in ScanScout’s online ad gallery. The most advanced example, right now, is what the company calls “Super Pre-Roll,” in which designers at ScanScout take standard ads supplied by its ad-agency clients and soup them up with interactive Flash-based features such as real-time polls. ScanScout believes that if it can engage viewers through such features, it’s much more likely to leave them with a strong impression of the brand supporting the more engaging content, and maybe coax them to take an action such as clicking through to an advertisers’ website.
It’s so confident about this theory that it’s proposing charging advertisers under a completely new system—CPE or “cost per engagement,” the video equivalent of Google’s cost-per-click pricing, rather than the traditional CPM or cost per thousand impressions.
“When you make an engagement model work, it powerfully aligns forces that haven’t been aligned,” says Day, who co-founded ScanScout with president Waikit Lau and chief technology officer Steven Lee in 2005. “Clients only want to produce really great creative, and they only want to show it to the right people. We only get paid for showing it to the right people—so if you are not inclined [to watch], we are not going to waste your time.”
But ScanScout doesn’t rely solely on the allure of its Flash features to boost engagement rates. Before its network even delivers an ad to a viewer, its software goes to work indexing the video programming itself, sifting through both visual and audio cues and associated information on the Web, such as comment streams, in order to divine what the show is actually about. That allows it to target ads alongside related content—and keep ads away from controversial content. (“Tiger Woods was a very salable name to brand marketers six months ago, and now his name is radioactive—nobody wants to be next to him,” Lau told me. While Tiger’s travails are obviously common knowledge, ScanScout’s system crawls the Web all day long for clusters of negative concepts that it wants to help advertisers avoid.)
ScanScout also tracks and categorizes what people watch on its clients’ websites. That way, when Procter & Gamble comes along with a body-lotion campaign aimed at 18-to-34-year-old females, ScanScout knows which viewers to show the ads to. It even knows whether viewers are more likely to engage with an ad in the morning or the evening, and serve up something appropriate based on the time of day.
Together with other Boston-area video analytics companies like RAMP (formerly Everyzing) and Visible Measures and video delivery companies like Extend Media and Brightcove, ScanScout is gradually dragging video publishers—the companies we used to call broadcasters—into the 2010s. I say the faster ScanScout can spread the idea of cost-per-engagement ads, the better. Because if Fox shows me one more “The Donalds vs. the Mannings” Oreo commercial, I’m going to rub Double Stuf all over my computer screen.
“It’s not lost on us that in an ad-driven economy, ads are the price for getting to content; they are a means, not an end,” says Day. “If it’s a choice between paying for content or viewing ads, you will choose ads. The key is to make them as good and as infrequent as possible—but pre-roll, the way it was invented around 2007, is an exact copy of how TV ads were run for the last 40 years. If that’s all online video becomes, then we’ve way undershot the mark.”
For a full list of my columns, check out the World Wide Wade Archive. You can also subscribe to the column via RSS or e-mail, and you can download Pixel Nation, an e-book version of the first 80 columns, as a free PDF file or a $4.99 Kindle edition.