Targeting Offline Behavior Online

by Phil Leggiere, Wednesday, October 31, 2007

ONLINE MARKETERS TEND TO BE fond of the conceit that advanced thinking about targeting emerged "ex nihilo" with new media, leaving behind the dark ages of advertising in the BI (Before Interactive) era. In reality, as Acxiom chief marketing and strategy officer Rich Howe explains below, "old school" direct mail marketing has much to add to the online behavioral targeting mix.

BI: Acxiom of late has been generating a lot of buzz for online marketing services. A few months ago, for instance, your acquisition of behavioral targeting firm Echo Target was featured in Behavioral Insider. Could you give a little more back-story about the firm's roots offline?

Howe: We've been involved in using data to enhance targeting for nearly 30 years, long before there was an online. So our newest work grows out of a long tradition in the world of direct marketing. We've had customers of ours who were interested in doing better with their online direct marketing campaigns. They understand that banner advertising still has a great deal of unrealized potential to be far better targeted than it is. What we learned is that experience we've had for many years in direct mail could be very effectively leveraged in the online world. Specifically relevant was our approach to adding segmentation based on life stages to conventional demographic clustering information.

BI: What have direct marketers known that online targeters are just catching up with?

Howe: Traditional direct marketers have known for many years that the right way to improve response rates was to break down their general mailing list names into smaller, more granular groupings based on deeper segmentation. The insight was that if you begin to look at the Zip code, the starting point for demographic clustering, it's useful for some things but doesn't go deep enough. There are just so much more relevant criteria. If you look at my street, for instance, everyone on it belongs of course, to the same Zip code. But the minute you look around, you see the differences from house to house are more powerful than the similarities. In one house you have a couple with very young children, while across the street the children are in high school thinking about applying to college. And down the street there's a household with grown-up children, and the people living there are thinking about retirement. So you have the same Zip code but radically different life stages, and with them, different consumer behavior patterns and priorities.

BI: How is the data generated and deployed online?

Howe: The data to develop this kind of segmentation online is similar to that offline. We have data we've aggregated from 133 million households, public data that we've clustered into 70 segments and 21 life stages. So the question was how to leverage that online and derive scale. Or how do you target online banners as effectively as we already know how to do with direct mail?

What we've done is run registration-based data compiled online against our offline data base, and then created cookies to anonymously track audiences based on that. The way it works is, we've built out partnership sites with publishers and ad networks where consumers provide personal data on an opt-in basis. It may be entering a contest or doing a survey, or it may be the form someone fills out when they make a product purchase online. When that happens the registration data is run automatically against our 133-million-household offline database to identify that consumer within the national segmentation and clustering system. An anonymous cookie is then generated which codes that segmentation information. So whenever the person goes to a partner site in the future, banner ads can be served based on their life-stage cluster profile.

BI: What is the behavioral component of the clusters?

Howe: Life-stage clusters essentially place every one of America's households into, first, a life-stage segment -- of which there are 21. We identify those segments -- including age, income, location, occupation and key life events, marriage, children, home purchase, job promotions, retirement. Then within those segments are a large number of clusters, currently 70 in all, which further break down households according to consumer preferences, values, opinions, brand affinities, media consumption habits and market behavior patterns.

For instance, we have two clusters within the lifestage segment called 'Gen X Parents.' Both have relatively low net worth and have a mean household age under 40. One we call 'Cartoons and Carpools, the other, 'Kids and Rent.' The first are married and extremely family-oriented, focusing their spending behavior on their children, their homes, and trying to invest for the future. The other are a mix of singles and married. They rent rather than own and spend money primarily on necessities or personal entertainment, with little or no thought to saving or longer-term investments. So though both fall under the GEN X demographic umbrella, and are even technically in the same life-stage with young children, their attitudinal, consumer behavioral and lifestyle profiles are very different.

BI: How are publishing partners taking to this approach?

Howe: For publishers deeper segmentation increases the value of their inventory. The way publishers are currently used to selling their inventory is either contextually or because someone being "tracked" by browsing patterns has visited their site from another which has tagged them as a customer prospect. But if you're able to analyze particular content areas and go beyond that, you can see, 'Oh yes, there are 4 clusters who frequent these pages in the greatest numbers.' This opens up a whole new level of insight about relevant ad placement possibilities.

BI: Do you see interest among brands as well as your traditional direct response sweet spot?

Howe: Because of our experience in direct marketing the base of users might be described still as direct response in orientation. But the lines between brands and marketers are shifting. As brands are evolving their strategies and thinking more in terms of finding a 1:1 conversational relationship with consumers, the need to personalize segmentation becomes much stronger all the time.