Contextual ad models evolve for online video

The growing popularity of Web video, combined with the relatively small share of online ad dollars being funneled to online video, present significant opportunities for publishers and advertisers. Some are looking to capitalize through contextual advertising solutions that improve advertisers’ ability to target ads to video content.

Nielsen reports that 11.2 billion video streams were served in July, up more than 30 percent from the previous year. But video advertising accounts for just 4.3% on online advertising spend, according to eMarketer.

Safe to say there’s a lot of interest in this space. And several vendors are developing advanced video search and indexing technology to help publishers and advertisers deliver more targeted, relevant ads to online video consumers.

For example, Affine Systems has developed a contextual advertising platform for online video that utilizes computer vision technology to scan and index videos across the Web. The technology can assign ratings to videos and tag specific items, such as objects, people, product logos and scenes (a soccer field, for example, or the beach), according to Michael Sullivan, CEO of the San Francisco-based startup. The company then uses third-party data to match videos with prospective advertisers.

"Contextual targeting is a massive opportunity for online video," Sullivan said. "The mechanism for buying online video is nascent. There are lots of integration issues. Advertisers are mostly buying inventory against premium publishers like Hulu or NBC. But that’s only about only 10% of the audience."

In addition to offering contextual video platforms that can scale, vendors are getting more granular with their targeting. ESPN is using technology from an Australian company called Demand Sport that will launch pop-up ads during a specific segment of a video. As MediaPost describes the program, a Denny's ad for its "Grand Slam" breakfast would appear during an ESPN.com recap of a baseball player hitting a grand slam.

"The whole concept is synchronizing brand messages with a moment in time," Luke Reinehr, Demand Sport's CEO, told MediaPost.

Another startup, Eyealike, also uses computer vision technology to scan and index images – and potentially videos – with its VisualAd platform. The company is currently testing the platform with a large social networking site, exploring ways to monetize the many images that community members post.

"Our product lets us identify gender, age and ethnicity," said Greg Heuss, the company’s president. If VisualAd identifies babies in images, for instance, the user who posted the image could be served ads for Gerber or Playschool. People in Facebook wearing glasses could be served Lasix or Lenscrafters ads, Heuss explained.

Heuss declined to name the social networking site Eyealike is working with, but said the additional targeting has the potential to increase advertising CPMs on the site from 15-20 cents to $10-$12.

Eyealike isn’t testing its product for video yet, but Heuss sees the potential. "Video is a series of still images," he said. "We’ve built a scalable product. You can see how publishers could monetize some of that video."

Heuss does caution that the technology is still evolving in this space. "It’s not ‘Minority Report’," he said. "We’re years away from being able to do what everyone thinks these solutions can do."

But the prospects certainly are intriguing.