Behavioral Targeting Gets Multimodal

Digitally enabled consumer behavior continues to evolve further and further away from the passive couch potato paradigm of yore, as omni-attentional media users become ever more adept at shifting between media and multitasking among them. But till now, as Chirag Patel, CEO of MeMedia, explains in the conversation below, behavioral targeting strategies have lagged far behind actual consumer behavior, stuck in siloed media compartments. The time has come, he argues, for a multimodal targeting across channels.

Behavioral Insider: As a newly launched behaviorally oriented ad network, MeMedia is entering an already pretty crowded space. What distinguishes your approach?

Chirag
Patel: We began with the specialty of delivering advertising targeted at the desktop, working with software publishers of downloadable applications. What we're trying to do is expand the way we leverage these assets in a multimodal manner as we refer to it.

Until now targeting of all kinds -- but especially of behavioral targeting -- has existed in a silo. You can target by desktop application or by Web sites or soon by mobile usage, but if you're an agency you're still forced to view these campaigns in separate pockets or compartments. We are looking at a number of different markets that have been especially underserved, with particular focus on the Hispanic market. There are so many quality publishers who've not been addressed by ad networks.

BI: Can you elaborate on how you bring together the different targeting pieces in this multi-modal model? Start with desktop apps, for instance.

Patel:
For the desktop we have a proprietary ad widget which is designed to aggregate data on application users. For instant messaging, for instance, advertising so far has mostly been at random, but we're focused [on] tracking types of applications downloaded and content preferences based on application usages. If someone has downloaded an IM service and chats about golf, the system will look for gol- related ads.

BI: Would this extend to something like music download software?

Patel:
Yes, but traditionally people have kept their music players on in the background as they perhaps surfed the Web to do other things -- so targeting the desktop strictly for that application never made much sense for an advertiser.

BI: So the idea here is that maybe they could use a consumer's music player preferences and behavior to serve an ad to the Web pages the user was browsing as he was listening to the music?

Patel:
Yes, that's the idea. For Web sites we can serve based on demographics culled from the Web publisher, the content of pages looked at and urls visited. All that is fairly standard fare -- but where we think we're providing a unique vantage point is in aggregating behavioral preferences as manifest in all those areas to predict end-user intentions.

BI: How does the algorithm work?

Patel:
The algorithm classifies all aspects of individual user behavior and prioritizes them by levels of importance. First it looks at how much relative time and attention users take in specific category verticals, both between different verticals and within category verticals. Next it looks at how users are interacting with a specific Web site in terms of which sections they spend more time at and which they spend less at. Finally, at the most granular level it looks at content within specific content pages to look at personal levels of interest and intensity. We bring all these behaviors together to see which available ads and ad campaigns on the network are most relevant to that user.

BI: And what about the mobile component?

Patel:
The mobile component includes applications downloaded, mobile pages visited or content used and most uniquely geographical location data.

BI: What phase is the network launch in right now?

Patel:
The network we have going up live this month will take the first steps at enabling an ad network to look at, for example, a consumer who's downloaded a financial software program, visited banking sites, browsed for information on loans and also linked to real estate sites, and who can be identified geographically by his mobile phone usage. The system can then say ‘I know this consumer's a good prospect for a housing mortgage package' and know the best places to deliver ads in several channels.

BI: Do you foresee any other layers of behavioral targeting to add to the system in the near future?

Patel:
The next big step beyond adding mobile to the multi-modal continuum is aggregating information from set-top boxes on digital IPTV. We're working hard on that now and expect to see a lot more attention paid to this area in 2008.
Recommend (3) Print RSS