When most people think of pairing ads with search queries, one company invariably springs to mind. But don’t try telling that to Proximic.
Founded in October around a novel contextual-matching technology, the privately funded ad network on Wednesday announced that it has signed ad-syndication deals with Yahoo Shopping Network and eBay’s Shopping.com.
The two agreements bring Proximic’s total ad inventory to nearly 50 million placements. Before the two deals, Proximic’s inventory was, approximately, zero.
“We’re a small company – so we don’t have to replicate what Google has built,” said CEO Philipp Pieper, respectfully referring to Google’s AdWords, the well-oiled machine for serving up ads based on keyword searches.
No, Proximic is taking on Google on a different front, what Pieper refers to as the “long tail” of the Internet. These are the sites that have proven historically difficult to monetize through advertising – blogs, social networks and other content-driven sites.
For advertisers, the Internet’s long tail is a vast, undiscovered country, according to IDC analyst Sue Feldman, who said that the major search engines deliver only a portion of the overall Web traffic.
“Our own studies show that most of the queries on the Web are not going to the major search engines at all,” Feldman said in an interview with InternetNews.com. Most people have a string of favorite sites that they regularly visit, and they don’t use a search engine to get there, Feldman said.
Google’s AdSense is geared for the publishers of these sites, but Pieper believes that Proximic can do it better. Judging from Wednesday’s announcements, he’s not the only one.
“Not only do we use different technology, we do placement differently,” Pieper said in an interview with InternetNews.com. “We don’t take a look at words – the things we identify are character sets.”
Proximic uses a technology called “pattern-proximity,” which is designed to transcend the language-based approach of keyword matching, a method that Feldman said is plagued by semantic ambiguities. (Does a search for “Amazon” refer to the rain forest or the company?)
Searching for patterns of characters within Web pages means that pattern-proximity can deliver semantic results in any language, Pieper said. Proximic claims that its character-based technique yields a result relevance similar to typing 200 words into a search engine.
The architect behind Proximic’s contextual-matching engine is Thomas Nitsche, a German mathematician who won the world microcomputer chess title in 1984. Nitsche serves as Proximic’s chief technical officer.
The computers Nitsche was turning into world-class chess champions in the 1980’s had about 5 kilobytes of memory. He applied the same economical method of coding to Proximic’s pattern-proximity technology, which boasts the ability to handle tens of millions of ads with only a fraction of the servers required in keyword-based systems with much smaller inventories.
Proxomic’s approach to ad-pairing is also much less likely to raise concerns over the specter of privacy-invasive behavioral targeting models, Pieper said. Pattern proximity does not rely on aggregating consumer data to match ads with content, but rather is based on intelligent inferences about the content itself.
A preemptive answer to the privacy question could be a big advantage in chasing down the long tail of the Internet, particularly the social networks. With Google’s AdWords leading the way in search-engine advertising, those endless pages of relatively unstructured data are where Pieper is betting the money is.
IDC’s Feldman agrees. “The digital marketplace – this advertising-driven approach to e-commerce, is really just emerging,” she said. As for the significance of heavyweights Yahoo and eBay joining with Proximic, she said that “in terms of potential for this market, it’s huge.”
Pieper described the dilemma facing Proximic and any other startup ad network as a chicken-and-egg problem: with no publishers on board the company has little to offer advertisers; without ad inventory the company is unattractive to publishers.
“We’ve solved the first part of the chicken-egg problem,” Pieper said. “Now we have an egg problem.”
Proximic expects to solve its egg problem soon enough, though. The company is in the process of finalizing deals with some major publishers, with announcements expected in the coming weeks.