Behavioral Insider: What seem to be the fundamental -- or at least, major -- differences between targeting for online and targeting in a mobile context?
Adam Soroca: Mobile is a truly personal medium. The cell phone is a personal device that, unlike a computer, is unlikely to have multiple users. Also, the profiles you get from mobile customers have a substantially longer shelf life than what you'd customarily get from a cookie on a Web site. With mobile you get subscriber data, so the store of data for profiling can be much richer over time -- whereas cookies get deleted constantly.
In practice you've got targetable information that's more locally based and actionable and timely. There's clear data showing, for instance, that mobile user activity is much more geared to immediate specific purchase intent. Mobile users employ search to look for very definite services and content they need right that instant.
BI: Search activity by mobile users remains an unknown commodity for the most part. As one of the early leaders in that space, what do you think is known about it at this stage?
Soroca: There's quite a bit of data we've accumulated on search behavior. One thing that comes out is that mobile users do search a lot. Our estimates are that in aggregate our mobile operators' search traffic will exceed 250 million searches per month by the end of this year. The percentage of searches that result in a direct purchase keeps increasing. One study we did showed more than 11% of searches resulted in a sale.
BI: How did JumpTag get its start in leveraging mobile search data?
Soroca: We started out as a search service provider for carriers. What we did was provide mobile operators with white label search platforms. Based on the subscriber information in their databases, our system provided customized indexing of mobile content for their users to search. Doing this well involves progressively learning more about what an operator's subscribers are interested in based on their profiles, and then providing them the simplest possible interface to access that information on and off portal. Essentially the way it works is that the more functionally relevant the search interface is for consumers, the more targetable consumer search trends and behavior can be for advertisers.
BI: How does it work?
Soroca: What we see opening up is help distilling behavioral profiles that enhance existing profiles of mobile users based on handset type, demography and geography, for the first time. The targeting can work in a variety of advertising models, from pay-per-click, pay-per-call, to more traditional CPM. If we see users searching for BMWs, we can target ads for specific local dealerships.
BI: Beyond retargeting users based on search history, what other dimensions of behavioral targeting do you see on the near-term horizon for mobile?
Soroca: Currently mobile ad solutions are mostly based on either specific application contexts like 'we know someone likes to download ringtones so we can target ringtone ads,' or by profile, as in 'someone uses X sort of handsets.' Adding search behavior to that moves the mobile targeting learning curve ahead dramatically. But there's still another dimension that can make behavioral profile much more powerful. That's aggregating all the historical data related to search and browsing with all the demographic and subscription information in databases including off-network content. There are new technologies we're working with that will enable us very soon to move in that direction.
BI: We've talked about some of the motivations of operators and of advertisers for advancing mobile targseting. What about the content providers and the publishers on the sell side?
Soroca: As the process has gone from expanding information and other content offerings from on-network portals to more off-net as well, I think the universe of behavioral and other data will move further and further beyond current silos. The exact road map isn't clear. But what is clear is that publishers as well as operators are in the process of building deep wells of behavioral data.
One possible trend will be that tier-one mobile content publishers will want to sell inventory by themselves through mobile operators who will extend partnership with them to open up their subscriber data profiles, anonymously, of course, on a case-by-case basis. Tier-two and tier-three publishers, on the other hand, may. and likely will, move more in an ad network model. In any case, there promises very soon to be a dramatically enlarged universe of data from all the buckets I've outlined -- subscriber profiles from the operators, on- and off-network search, purchase and browsing data and publisher profile data.