John Mracek is CEO of advertising technology company, NetSeer. NetSeer announced a new offering today called “concept-based” advertising. Read the release (PDF).
Concept-based advertising sounds a lot like semantic or contextual. Is the “secret sauce” the real-time identification of concept categories like “Italian food” or “headphones”?
The secret sauce is the ability to algorithmically, and therefore scalably discern concepts on a page as well as determining related concepts “off page.” Categorization is just a crude approximation of what NetSeer’s technology does. Unlike first generation contextual, which is largely keyword-driven based on the page being analyzed, or semantic which tries to leverage Natural Language Processing (NLP), NetSeer’s concept-based approach takes a top down approach to analyzing concepts in the context of how they are used in the entire web. One of the issues with keyword-based classification approaches is that they require meticulous updating and management of keywords and how they map to various categories. Semantic also involves a lot of hand-holding that limits its scale. In fact, some semantic providers boast about the dozens of linguists they use to build out their semantic dictionaries. In addition to making it costly to support, the knowledge about the page is limited to purely what is on the page. NetSeer’s approach is almost purely machine based and therefore scalable and always up to date. Concept-based advertising goes beyond both first-generation contextual and semantic by unambiguously determining the most relevant concepts on the page as well as related concepts that the user is likely to be interested in.
View throughs are a very important metric when evaluating online advertising campaigns. However, our model to date has focused on sending consumers to search ads and tracking both click-throughs and traffic quality as measured by Yahoo, our search partner.
Generally speaking, why is “real-time” important these days for web advertisers and publishers?
We think “real-time” is especially important in an advertising context. The more you know about the user and their context at the time of impression, the more relevant the advertising can be. For instance, one of our health sites deals with information and treatment about cancer. If someone comes to that site with a cookie that says that they searched for “Jumbo Mortgage” two days ago, a typical retargeting system will put a mortgage advertisement in front of them because they score high on the Behavioral scale for in market for a mortgage. However, if that person is on that site investigating different approaches to cancer therapy for a relative, it doesn’t matter how much they want a jumbo mortgage, they are not going to completely break context and click away from their investigation at hand. The signal you get from real-time contextual analysis is always critical for achieving high relevance.
