Making YouTube Pay Off
Since shelling out $1.6 billion for YouTube last year, Google has yet to wring a profit from the video site’s millions of user-generated clips.
Google (nasdaq: GOOG - news - people ) has tried inserting overlay ads into videos, and on Tuesday, it launched a new program to distribute ad-laced, commercially produced YouTube video clips to sites through its AdSense network.
But the key to unlocking profits from the torrent of homegrown online videos may instead be in the hands of a small British video search company called Blinkx.
Blinkx released a tool Wednesday that lets online publishers place targeted text ads in any video embedded on a Web site based on the actual content of the video. That’s a big contrast to Google’s approach: Google figures out what ads to pair with a video based strictly on the video’s title and any keywords attached to the clip. Blinkx software “listens to” and “watches” the video, then inserts text overlay ads based on the spoken words and to some extent, the images in the clip. That technology depends on algorithms developed by a longtime Google competitor, search engine Autonomy.
Here’s how Blinkx’s contextual advertising might work: Imagine a teenage girl doing a podcast about a dress that she just bought. Blinkx software might create a text link to the dress at the bottom of the video player, even if the clip isn’t labeled with the brand of the maker.
Blinkx Chief Executive Suranga Chandratillake says the software can also recognize written words and even a small library of faces. That means that automatic ad targeting isn’t limited to the relatively small amount of video content generated by media companies and savvy bloggers who carefully label all of their content with text tags, he says.
Advertisers will be able to insert targeted ads into the massive number of amateur videos on the Internet, many of which often weren’t intended to generate revenue and so carry no content-related tags.
Blinkx’s new tool comes on the heels of Google’s plan to syndicate YouTube videos on the hundreds of thousands of sites in its AdSense network of publishers. The move, one of the first major steps to monetize YouTube’s millions of hours of video content, allows any publisher in the network to have videos automatically placed on his or her site based on the site’s text content. Video ad revenue will be split among Google, the site’s publisher and the video’s creator. For now, Google plans to circulate only a few hundred videos created by commercial media companies rather than tangling with the mass of user-generated clips that make up the bulk of YouTube content.
“In the short term, Google is obsessed with the head of the market, using a small volume of content and high value ads,” Chandratillake contends. “In the longer term, I think they’ll get interested in the long tail and using their millions of user-generated clips. But their stumbling block will be developing the technology to understand what those videos are really about. The level of targeting really matters.”
Better targeting through audio and image recognition could also improve the click-through rate of videos, or how often an ad is clicked for every time it appears. A report last year from the online ad consultancy eMarketer estimated the click-through rate for video ads in general at 0.75% - tiny, yet still higher than the more prevalent non-video display ads.
Andrew Frank, a media analyst with Gartner Research, is skeptical about advertisers’ ability to grab the attention of an online video audience. A video ad, he argues, simply passes too quickly. “Unlike text ads, you can’t read the text and then click afterwards to get more information,” he says. “Video is a different experience, more about entertainment than information gathering. It’s basically more of a passive medium.”
That hasn’t prevented advertisers from significantly increasing their online video ad spending this year. By eMarketer’s count, video ad spending will account for 3.5% of the $21.7 billion projected to be spent on online ads this year, up from 2.4% of online ad spending last year.
Google is poised to tap into that growing revenue source with or without Blinkx’s image and audio recognition technology, says company spokesperson Brandon McCormick. McCormick argues that culling information from the text surrounding YouTube videos - including comments from users - has generally been enough to allow advanced content targeting.
In fact, contextual targeting - whether based on text or audio and images - is just a small part of realizing online video’s moneymaking potential, says Michael Kelly, a media analyst at PricewaterhouseCooper. Just as crucial, he says, are elements such as increasing video access from mobile devices and targeting users with ads based on demographic details that they reveal through social networking or other personalized mediums.
“There’s no one silver bullet,” Kelly says. “In fact, it’s more like a revolver with six different silver bullets, maybe from six different companies. You’ll have to put all of these pieces together before you can really get the maximum effectiveness out of video advertising.”