Commentary

In The Zone: Geography, Behavior And Cross-Channel Identity

Geographic zones are a key element in traditional direct mail. But they also play a role in digital marketing and attribution — in the form of zones that may have as few as four homes. For insights on this, we talked with Ray Kingman, CEO of Semcasting, a firm that started out compiling data from public records but has since expanded to provide digital marketing services, including cross-channel attribution.  

What are these “zones” that you offer?
We call them Smart Zones. They can be from four to six households, given our design objectives and the ability to personalize. We’ve gone from 35 million zones to over 200 million, 78% one-to-one. Comcast will serve those homes, so they’re more or less static. Frankly, mobile devices now outstrip static, with granularity more refined. Smart Zones and mobile footprints often give us a deterministic link between the CRM file and delivery points to devices in both the home and the business.

How do you identify people across all devices?
We use IP addresses to identify the delivery points of all the devices in homes and businesses. Like everyone else, we saw that cookies were getting increasingly diminished over time. Our ability to link the offline world to online delivery points, is now more deterministic in nature.

But don’t IP addresses change all the time?
Obviously, IP addresses keep shifting. In the static world, meaning the home network, they change once or twice a month, on average. In mobile, they’re going to grab another IP address each time you restart. You need to be able to do this constantly — at least daily. The degradation is about 4% a month.

What can you learn about a consumer, using the IP address and the data you have?
We pick up the device signals, although not as they occur. We know you went to Fenway Park on opening day, and made a request on your device during the game. And we know that you were at a home in North Andover the prior evening and on multiple evenings before. We can probabilistically say that the device belongs to that home. We know the home is worth a certain amount of money, has three residents, and that you drive a certain kind of car and vote Democratic.

That’s quite a profile. But how does that translate into a marketing program?
One of our clients is the Dallas Mavericks. They don’t know the identity of about 30% of some the people who go to their games. They want to know more about them. We look for signals from games in the arena. It’s a simple matchback exercise: We match device signals back to their home and business, then enhance that with demo data — and perhaps their persona characteristics. The Mavericks can then reach out to them for ticket subscription blocks or season tickets.

Any other examples?
We’re also working with almost 2,000 car dealers. These dealers are interested in a couple of things: For one, who’s seeing the ads online? One data point is the audience impressions being served. They have logs of the responses coming in, people who responded to an ad, or maybe went to Web site. We have those logs from dealers, and we’re able to say we served them an ad and they went to Web site — there’s a linkage between impressions Then there’s another component which is CRM data. The dealers have customer databases, which we onboard through our Smart Zones.

How do you handle attribution?
The key challenge of attribution is whether or not it’s a deterministic or a probabilistic reference. The more deterministic the reference, the more likely the attribution is to be correct. You need a common key reference point like an IP address. That IP address has to be for all the touchpoints that you want to attribute. The only alternative to that might be within the walled gardens of Google or Facebook. The IP address is across those walled gardens, into the open net and even to the terrestrial. 

Next story loading loading..