Behavioral Insider: What user behaviors at Gather.com are incorporated into their profile?
Gerace: We look at what information a user is most interested in, whether they looked at food content, wine content, travel content; and we then add to the profile keywords that they have used, that they have published, that describe images they have explored. We use it on the back end to figure out what content we can recommend as the most relevant, as well as figuring out what advertising is most likely to interest them.
BI: How are ads bought and placed?
Gerace: The ads are on the right side. It is a CPC bidding system that just launched in Version 2 with an updated algorithm. Citibank, Brooks Brothers, SmartBargains are using the platform. On Google or Yahoo you bid for just placement adjacent to content; so you are bidding for the keywords on a page or the search words. Gather lets you bid based on the totality of the user's interests. So a user might have searched or explored content on Vegas on travel and on hotels, but if your ad was not bid high enough you might not have appeared on those pages in Gather. But when the user [whose profile shows interest in those topics] goes to look at a political or news article with lower bid prices, your ad for Vegas could suddenly appear. It gives you the ability to reach consumers more cost effectively because we profile the user.
BI: So advertisers are bidding on keywords, but the keywords actually reflect past behaviors, not just search terms.
Gerace: Exactly--the relevancy and frequency of the content.
BI: A persistent complaint in BT is that segmenting an audience too finely provides great targeting but poor reach. Tying fairly general keywords to users' larger behavioral profile would seem to be one solution to the problem of generating a large enough segment.
Gerace: Unlike traditional segmenting systems, which typically classified a user and then targeted ads towards a user group, Gather doesn't have to create specific behavioral segments and just show general ads to the segment. Instead, we can create on-the-fly segments. We are all more complex than any behavioral segmentation system would reflect. The way we think of it is, we create on-the-fly segments based on a user's content interest, and match that with the appropriate advertising based on the user's likelihood of response.
BI: This is different from what MSN AdCenter seems to be proposing, which is to add a BT layer as an option in a general keyword buy.
Gerace: The keyword-only systems like Google and Yahoo have an inherent problem. They create many marketplaces for each keyword. So "Vegas" might cost you 40 cents to buy and "Vegas hotel" might cost 80 cents, and "Vegas hotel Bellagio" might be a $4 click. The advantage here is that instead of creating many marketplaces, we overlap all ad targeting. Someone might be interested in Vegas, in travel, and in day care, and all the ads that overlap for this given user compete for that user's attention. Our system takes into account the totality of the user's interests to create a single unified advertising marketplace that is more efficient for advertisers.
BI: What sort of preliminary ROI and response are you seeing?
Gerace: It is anecdotal from ad customers. I think that we have had very pleased advertisers. The feedback has been good, because in part the power of the Gather audience is that it is engaged, informed, educated, and slightly more affluent. Sell- through rates after people click on Gather ads are good. They are buying typically higher- end merchandise than a standard purchase.
BI: Traditionally, advertisers shied away from user-generated areas not only because of a lack of control, but because the level of engagement in the content resulted in low ad response.
Gerace: On the involvement of users, what we are beginning to do at Gather is look at how we can use the social network as a driver of advertising consumption. So how do you use the word of mouth that drives content consumption to drive advertising consumption? I can talk more about that in a couple of months.
BI: It seems that social networks would allow marketers to track and target a whole new layer of user behaviors, like their level of interactivity, standing in a community, rate of content creation, etc.
Gerace: The way we look at it is more akin to having key influencers in a group that are impacting the behavior of others. Advertisers are already catching on to that, using buzz generation mechanisms by hitting key influencers. One of the things Gather will be able to do is identify the most powerful networkers, the most respected authorities in a social network on a given topic, and create a specialized advertising product that allows advertisers to reach these folks. It allows them to leverage word of mouth, which is far more credible than media-based advertising. It is something we are spending a lot of time on.