How Seethroo uses UGC behavior to profile audiences
Moskowitz: How have you zeroed in on user-generated content as a profiling vehicle? Are you watching what a person produces and targeting them on that basis? Or are you watching what UGC a person views and targeting them on that basis?
Frankovitz: In a word, both. We examine what bookmarks a person has saved for themselves, as well as any bookmarks they've viewed that were found by other people. This is one of the fun things about online bookmarks that people love to do: discover and explore great links other people have found. And it tells us a great deal about that person's interests. So imagine a person who has saved three mortgage-related bookmarks for themselves, and has also looked at another user's online bookmarks about refinancing. We might conclude that this person has an interest in both buying a house and refinancing their current home, but the refi interest may be a bit more short-term or casual. This is just a simple example. There are several factors we use to determine the degree of interest, so there's no one rule that applies to all situations. Seethroo's detailed analysis translates both the contents of a person's bookmarks, and the activity associated with those bookmarks, into solid targeting information.
Moskowitz: How are you categorizing the varieties of this kind of UGC?
Frankovitz: The way we categorize all this information has a lot to do with each ad network we're partnering with: we find out what kinds of ads they're carrying and what campaigns can benefit the most from our enhanced targeting. Actually, we're limited only by their inventory. So, for example, if we're tracking bookmarks related to IBM laptops, we can display a highly targeted, vendor-specific, competitive-upgrade style of ad (for example: "trade in your old Thinkpad and get $500 off the latest HP system"). But if that kind of specialized ad isn't available from our ad network partner, we can fetch a generalized technology-related banner ad (for example: a buy.com ad saying "we're the place for the most popular cool gear").
Some bookmark communities are general interest and store links for every topic under the sun, while other sites are more tightly organized around movies, videogames, or what have you. We work with both non-vertical and vertical bookmark sites to ensure that, whenever possible, an ad is being displayed to each person that matches their individualized interest profile.
Moskowitz: What does tracking UGC give you compared with other behavioral parameters you could track to profile an individual on the internet?
Frankovitz: The nature of bookmarks offers benefits not found in other, more conventional kinds of data. Bookmarks are inherently personal, so they tell a different kind of story than clickstream analysis, time spent on a site, and so forth. A person may surf to 20 websites in a day and generate 100 different pageviews, but of all that activity there's perhaps only one or two links they'll want to bookmark. We care more about those links because that's exactly what each person is telling us: "I care more about these links I've bookmarked." Most technology up to this point has been focused on getting an ad to match the page it's sitting on, or the site's overall content as a whole.
Moskowitz: Or an overall behavioral profile?
Frankovitz: Yes, that too. Our bookmark analysis lets us target ads that match a person, not just a page.
Pretend that your job is to figure out what foods a restaurant should buy. The best way to do that is to find out what the patrons are actually eating, not what they're thinking about ordering. To continue the metaphor, the common practice today is like having consumers open up a menu and see an ad next to every food item. That is a less-efficient "spray-and-pray" approach, and it wastes a lot of marketing dollars on people who aren't really in your audience. Seethroo lets ad spending be more focused and effective, and the benefits include greater value for the advertiser and enhanced revenue for ad networks and publishers.
Moskowitz: When will we see this technology launched?
Frankovitz: We have a few demos and customer trials in the works as I write this, so most likely before summer 2007.
Robert Moskowitz is a consultant and author who speaks and writes frequently in the United States and abroad on such topics as white collar productivity, knowledge management, practical use of the internet, telecommuting, caring for aging parents and business applications of information technologies. Read full bio.