• by Steve Smith , Staff Writer @popeyesm, February 16, 2007

Advertisers continue to be shy about putting their brands into the mosh pit of user-generated media and social networking, because you never really know where (and next to what) your ad will show up. By applying behavioral targeting techniques to the unique activities of social media, a brand can segment audience more precisely and advertise to them in social settings with greater confidence, according to Andy Monfried, president/founder, Lotame. But UGC requires a different kind of profiling and segmentation that is based on deliberate actions users take in their communities. The publisher-oriented solution launched in December with one video-oriented site and will be part of an upcoming large social media launch

BI: What problem is Lotame trying to solve?

Monfried: The fastest growing source of publishers' inventory was becoming social media and still is. But there is a monetization problem around social media. The CPMs are not keeping up with the rest of the Internet. The growth is there but the data in terms of the targeting is not. We set out to develop data collection and a data platform that is focused for social media.

BI: How are you profiling and segmenting differently from familiar BT engines?

Monfried: BT is really navigationally focused. You are on site X and we're going to show you ad Y because of your previous navigational history. In social media, we believe what is more important and heavily weighted in our targeting and data collection algorithms is the activity: how you interact and what physical actions you take inside social media. .

BI: What kinds of activity can you define as valuable?

Monfried: Things like upload a video, search a video, comment on a video, comment on a picture upload, form a group, join a club. These are things you physically have to do to interact with the media. That for us is far more important than what context you are looking at. So the first thing we did for social media publishers was to identify the physical action. We assign a value to that physical action and then we overlay it with the navigation, which we call the interest targeting. We call it activity vs. interest.

BI: What would some of these segments actually look like, then?

Monfried: One advertiser is now working with the publisher to build this subset out: users who have read an e-mail, who have sent an email, who have created a blog, users who have just registered to the site in the last 48 hours, users who have one or more friends [in the network]. That's the activity side. We assign a value to those actions in our algorithm. The next one is the interest side -- and it is all music -- so [the question is,] how do all those activities match up against a list of bands. That's the navigational side.

BI: How are segments like this of value to specific marketing goals?

Monfired: The one we've seen big success with are recency and relevancy ads. The big thing for advertisers right now is getting everyone who has taken the action of X in the last 24 to 48 hours. We allow targeting on a recency level, i.e., the last 24 hours.

BI: Translate this into an actual campaign goal.

Monfried: A campaign for a cell phone and service was targeted at a unique level and the results were excellent. They wanted to target anyone who had uploaded a camera phone video, who viewed a phone cam video, commented on one -- and attributes like that -- in the last 24 hours. You can see how the CPMs quickly go up.

BI: BT is always struggling with scale. How small do these segments get?

Monfried: We encourage the advertisers using the tool to broaden their audience in their first couple of campaigns and include testing protocols; A/B tests of larger audiences vs. smaller ones. We're seeing campaigns now compared to run of site with 2-times and 3-times [CPM] lift just based on this level of targeting.

BI: What size site do you need?

Monfried: If I say I want to show this ad to anyone who has uploaded a camera phone video, if you don't have between 50,000 to 100,000 people in that group, it's not going to makes sense in social media. Once it gets over 100,000 the real question is more about the frequency of seeing that person. Some social media publishers see much greater frequency than others.

BI: This is a site-specific rather than networked solution?

Monfried: Yes, but it's all self-serve. We created a system where the publishers can build out their own tags and segmentation system. My system will be like Google's AdSense or AdWords. It will all be self-contained at both the advertiser and publisher level.

BI: A constant concern of advertisers in social media is that people don't' responds to ads when they are more engaged with one another in these settings.

Monfried: As the industry matures and the targeting gets more efficient, that [concern] will naturally dissipate. If you see a Classmates or LowerMyBills ad for the 400th time in a day, I agree that response rates will go down the toilet. As you see publishers removing the numerous ads they have on their site and sticking with creative that works and targeting that works, that will naturally disappear and you will see CPMs rise and the relevant advertiser will go along with it.

BI: And how about advertisers being unsure where their ads show up in UGC?

Monfried: We believe that once you segment this [audience] out you address the adjacency issue...not 100%. That is not going to go away overnight. However, if the brands pick where they are advertising, the users, and the inventory they are advertising to, they would make the decision, as opposed to the publisher. We're putting control in the hands of the advertisers and the brands. We're saying to the advertiser, this is where most of the people are spending most of their time online. This takes the attributes of social media and makes them available to advertisers.