It's the End-User, Stupid
Behavioral Insider: GoTo at this point has probably been around longer than any other major mobile content provider in existence. How would you compare the actualities of mobile targeting today to online targeting?
Lee Hancock: It's still quite early on. What we don't have at this point is the ability to track specific user history across cast arrays of sites and content the way it's done online. Or let's say readily accessible. The universe of theoretically available information for targeting is probably greater than in any other medium. Right now there are already billions of pages of incredibly rich and detailed information, but most of it remains buried in logs. Mining it all would be expensive and time-consuming, but feasible. But the fact is, there's far more data than advertisers know how to use wisely yet.
BI: So you sound a little more cautious about the possibilities of behavioral targeting than others in the industry,
Hancock: I'm very gung-ho about what's possible in mobile targeting, but I'm also very cautious. The nature of this medium is so personal that the biggest danger now is over-targeting prematurely in a heavy-handed and intrusive way.
BI: Can you elaborate?
Hancock: Our goal has always been to raise the bar for mobile in terms of end-user experience first, on the premise that the more user-centric the approach, the more opportunity for relevant targeted advertising there will be. The targeting we do will always need to be integrated seamlessly into a positive end-user experience. If it detracts from the user experience, if it's even something the user is aware of in the sense that it's extraneous to how they enjoy their experience, then it's counter-productive. Our strategy is to evolve deeper and more customized content channels organized around specific interests and even sub-segments within an area of interest.
BI: What types of behavioral data would you say are unique to the mobile space?
Hancock: What's completely unique to the medium is that we know the location of each user, based on not only where they live but where they physically are located. That's incredibly powerful, but also not information that can be abused. The potential for privacy violation -- or just being perceived as a nuisance -- are too great to just jump in without being very careful. We're also a medium that's menu-driven, as opposed to open search. We allow users to customize their own menus. However most users find it easier for us to customize their menus based on what their demonstrated behavior and interests are. Thus if a person listens often to a certain ringtone that's been in the number-eight position on the menu, we'll move that up automatically to number two or three. So we start with fairly generic content channel placements on a menu, but those choices and sub-choices over time become customized to specific interests and behavior.
BI: How do you see that information being applied?
Hancock: As data sources converge from the carrier side, where you have demographic information, and then from our side, where we have information about what users are interested in and where they are. amazing potential marketing synergies become possible. One way interest, behavior and location could work seamlessly together is, say someone has downloaded Madonna song ringtones and goes to the movie menu frequently. An entertainment company promoting a new Madonna movie can target based on that content and behavior, but also focus on the specific movie theater near where that user is. Then a Starbucks nearby having a special Frappuccino offer can target their promotion or a Barnes and Noble nearby the theater can promote a new Madonna coffee-table book. Very powerful stuff, but obviously something that needs to be done responsibly.
BI: What's your sense of some of the targeting opportunities for mobile advertising in the near to mid-term?
Hancock: We still don't have the pieces all together. What we are at a point of developing are end-user tools which allow us to learn a great deal more about how interests and locations intersect in the real world. For example, we have tools that enable users to locate stores near where they are for particular products they're looking for. One interesting aspect of mobile behavioral data that's barely begun to be explored at this point are geo-based speed polls, What we do is run polls where we ask questions on various topics. where users opt in with their opinions and preferences on all sorts of topics of interest to a marketer. Obviously this kind of information is potentially useful on many levels -- but is one potential application of great interest, not just to direct response advertisers but to brand advertisers looking to customize their messaging by geo-segments.
So there are no shortages of highly targeting pieces of behavioral data which will gradually come together. The question still remains, though, does the targeting you do fit in ultimately with your overall value proposition to end-users.