Video search and monetization has been a hot topic lately, following in the wake of the general hot topic of online video (as advertisers invariably follow the eyeballs). This was accelerated by YouTube’s announcement last week that it will integrate inline video ads.
We’ve explored this nascent field several times in the past, in looking at what a few technology developers are doing to use voice recognition and other technologies to figure out what a piece of video is “all about.” The thought is that as these technologies develop, we end up with better video search and better ability to add contextually relevant ads to videos.
This coverage has centered on YouTube, Blinkx and Adap.tv, following conversations with these companies. But we’ve only given passing references to ScanScout (see past blog comment to this effect). Today, I finally had the chance to speak with CEO Doug McFarland, who has some interesting thoughts on the direction of this developing field.
Like the companies mentioned above, ScanScout employs complex algorithms to determine context that take into account (weighted in no particular order) adjacent content, speech recognition, image recognition, color, sound and other variables.
From there, it inserts ad overlays that allow users to respond by clicking an ad or viewing a separate video window. Given TKG data from User View (Wave IV) showing that online video elicits high percentage response rates, this will continue to be an important area of development.
TKG Survey Data: Online Video Direct Response
There are two parts to this equation though: the ad format and the contextual engine that sits behind it. We’ve examined the experimentation happening with different ad formats, given that pre-roll ads have been largely determined to be undesirable.
But the onset of inline ads as an alternative, though a step forward, is already becoming a commodity, says McFarland. The contextualization engine sitting behind the scenes is more where the value and differentiation among these products and providers will be decided.
No One Said It Was Going to Be Easy
Determining context with video is a much bigger challenge than it is with other online media, according to McFarland, if you consider the complexity of the content, audio and visual components, and the sometimes racy content of today’s universe of online video content. A great deal of the video content out there today simply isn’t monetizable or contains things advertisers want to steer clear of.
“The first job is to determine what a piece of video is about,” says McFarland. “The second is to figure out if it’s monetizable.” The latter task he claims takes into account a host of variables — analogous to but much more complex than the way a spam filter works — to determine if something is pornographic, racy or distasteful in any way.
The secret sauce, according to McFarland, lives inside the algorithms rather than going the ‘folksonomy’ route explored in the past. This concept, explained to me by Yahoo!’s Bradley Horowitz a while back in light of the company’s acquisition of Flickr, relies on humans’ ability to identify and tag rich media content, rather than algorithms. This is behind many of Yahoo!’s social efforts, including social bookmarking, and could be a key part of video search, according to Horowitz.
But this isn’t as reliable, consistent and scalable for video as algorithmic contextualization is, McFarland believes. Add the software’s ability to improve over time by learning the way the human brain operates, and McFarland claims it has the ability to not only decipher meaning or context but also effectively draw myriad and sometimes indirect associations of product advertising and certain types of content.
Implications for Local
The byproduct of all this innovation for local will be better searchability for the growing ranks of small-business video ads that are being created by the likes of TurnHere and Spot Runner and distributed by Citysearch, Superpages, Yellowpages.com and others. These providers will grow in number, as will small-business advertisers that increasingly see video as an affordable and attractive way to advertise locally.
McFarland also agreed with this point that video is a medium small-businesses understand and value. For certain segments interested in exposure and foot traffic, video might be a less abstract concept than some of the other forms of interactive advertising (read:PPC) that have been thrown their way over the past few years. It’s something that they “get” and for the same reason, Yellow Pages sales reps will have an easier time communicating its benefits and that of bundled ads that include print, online and video.
Interestingly, McFarland brought a new fold into this conversation that I hadn’t heard yet — that many large national advertisers are the same way. One question he’s starting to hear from large advertisers when looking at online video advertising is, “Where can I see the ad?” This was an eye-opener for McFarland that could be an important question to ask of all forms of advertising. Sometimes ROI for the advertiser — as illogical as it sounds — is mentally weighed by his or her ability to tune in and see the ad.
This is nothing new, as a great deal of Yellow Pages advertising and small-business video advertising on television has traditionally been driven by this vanity factor. Small-business owners (and large advertisers alike) want to be able to see their ads, smile, show their families and determine that their ad dollars were spent well. Though not a primary driver, this is an interesting concept that shouldn’t be ignored as the innovation and experimentation continues in online video monetization.
Related TKG Reports: Online Video: A New Local Advertising Paradigm.