With Congress and the Administration stepping up their focus on online advertising, the industry is casting about for solutions to head off increased regulations. One suggestion, proposed by Brooke Aker, CMO of the newly formed ADmantX is semantic targeting, a technology that allows advertisers to dive deeper into content analysis and targeting ads via the emotions and motivations content elicits in users – without cookies or other techniques that elicit privacy advocates concerns.
MarketingVOX caught up with Aker to get a better understanding of this technology and how it could aid in the online privacy debate.
How does semantic targeting work?
Aker: It analyzes page content based on several factors, including:
1.) Role: Whether a word is a noun, verb, adjective, adverb, etc. For example, the sentence "There are 40 rows in the table" uses row as a noun, but "She rows five times a week" uses row as a verb. Ads for rowing machines should not show up next to content that talks about rows in the table.
2.) Relation: How every word relates to other words in sentences, paragraphs or the whole document. For example, the sentence "The buttons on the lousy Motorola cellphone are actually really great" may trigger a Motorola phone ad without considering the connection to the word "lousy." Bad match.
3.) Definition: Every word on the page receives a definition based on the other words around it. This is the true essence of context. For example, in the sentence "I used beef broth for my soup stock" the word "stock" is understood to be in the context of food vs. "The company keeps lots of stock on hand," in which the word "stock" is understood to be in the context of inventory. Ads for inventory management software on a soup stock page are a waste, as are ads for cooking utensils on a page about inventory.
4.) Emotions & Motivations: The totality of a page may be summarized into categories, even when the names (words) of those categories are not directly on the page.
Can you give me an example of how this would work?
Aker: Sure, consider the following page content:
SCORE "Counselors to America's Small Business" will present Kevin McDermott and Steven Smits, owners of Patriot Taxiway Industries in Lomira, Wis., with the SCORE Award for Outstanding Veteran-owned Business. They earned the honor for their contributions to their community, exemplary achievements in their field and overall business success. McDermott and Smits will accept this award, which is sponsored by Administaff, at the second annual SCORE Awards and black-tie gala on Sept. 16 in Washington, D.C.
This content can be categorically understood to include such things as success, joy, prestige, duty, and performance. These are profoundly better concepts to match ads with than the words on the page itself. The vantage point of looking at content as advertisers do when they design ad campaigns, particularly brand campaigns, is clear. If BMW wants to be associated with the emotion of joy, their ads should be matched on far more content than can be contained on the Car & Driver or Cars.com websites.
So how this would satisfy privacy advocates, regulators' potential concerns about behavioral targeting?
Aker: This would satisfy privacy concerns because semantic targeting can go on without ever having to resort to cookie-based click following, IP sniffing or adding "beacons" to a PC. In fact, using semantic targeting means nothing is added to an individual PC. In semantic targeting, all the effort is on the content - a public good, without regard to the individual at all.
Leaving aside the regulatory issue, why should advertisers pay more attention to semantic targeting?
Aker: For a number of reasons. It offers more precise matches without the obvious mistakes keyword systems produce. It offers brands more control over what constitutes brand safety, rather than a static list of "no-fly" key words. For example, for a car company, content that mentions "car crash" may be a brand safety issue, but for an insurance company it is not a brand safety issue and is a natural match. It also offers better segmentation of content. Therefore, content that previously never got an ad match now does, so inventory is cleared faster and fewer ads are brokered into exchanges.
What about pricing?
Aker: That too. With better segmentation and fewer ads being brokered to exchanges, the average price paid per ad rises, a benefit for both publishers and advertisers. Advertisers, who always want premium placement, get the benefit of higher clicks or brand lift with better targeting, which outpaces the higher cost.
