Inside Intent

American Demographics, March 1, 2004

Max Kilger

Director of Statistical Sciences Simmons Integrated Marketing Solutions

It used to be that if you wanted to find out about linen stores in L.A., you'd take data on linen consumption and information from the census tracts. Market researchers would take database A and database B, and smack them together. Unfortunately, data sets have different people in them. Nielsen, for instance, has incredible data on what people are viewing, but it doesn't have anything on the products or the psychographics of those viewers. Other databases have thousands of products and hundreds of psychographics, but they don't have data on viewers. Database integration creates a behavioral system that spans the two database systems when the people aren't the same. As a social psychologist, I'm a big believer in behavior. We have been doing this for about five years, but it's still very new. Here's how it works: Let's suppose you have two databases: Nielsen's viewers and a consumer product database. First, you create behavioral segments. Nielsen's has 20 and product consumer product one has 20 as well. Let's say John Smith is in behavioral segment 7 and Nielsen says he's in the late-night junky segment. He watches Conan O'Brien and Letterman. Then you go to database B and find a behavioral segment, one that is more likely to eat Cheese Whiz. What do they have in common? That's the key. What you do is produce a behavioral system that spans database A and B. You look at the driver variables on the Nielsen side and driver variables on the consumer product side. If late-night junkies in the Nielsen database index 235 for Conan, then you know things are consistent. You want behavioral segments to act different. Next you test non-driver variables. You want to know if the people who watch Conan also eat Cheese Whiz. If late-night junkies index 195 for Cheese Whiz, then you know that that consumer product segment 7 is not only a discriminating driver, but a non-driver variable. Then you can go to the client and say, 'Give us an ad buy for Cheez Whiz on Conan.' Using data integration you can find your consumers and find them at a bargain. Or, if you're a network and you're trying to hawk a new program to Cheese Whiz advertisers, it's hard without ratings. Now you can say that that show is similar to others in that behavioral system: late-night talk shows. Then you can say their viewers index 195 for Cheese Whiz and you've made your sell. We're making refinements to make more powerful drivers to produce integrated systems with more pull, more predictive power. We're bringing the consumer into sharper focus and solving the problem of no single source database.

Thom Mocarsky

VP, Communications Arbitron

The Portable People Meter, or PPM, is a new rating system that's been under development for the past decade. This spring we expect to test it with Houston's radio, TV and cable stations. Any electronic media can have an encoder, which sends inaudible codes - to the PPM. It's multimedia. Right now, measuring radio audiences is a manual process. With the PPM, people who participate in our ratings survey carry a device the size of a little cell phone. They don't have to write anything down, or push any buttons; they just carry the PPM. It passively picks up the encoded frequencies from TV, radio and cable broadcasts and logs them. All the people have to do at the end of day is put the PPM in the bay station, which sucks out the information and sends it to a hub at Arbitron. Everyone 6 and older will have a PPM. There's also a motion sensor inside the PPM, so it records when the people are moving and for how long. With a diary, you have no idea what people should have written down. But with PPM, we know it's being carried and if it's being carried, we know it's doing its job. It's better than what we do now. We're measuring exposure to encoded audio. So far, we're finding individual radio stations have much greater reach than the diaries indicate. Advertisers tend to use radio as a frequency buy, but they could use it as a reach buy. PPM uncovered the truth. Also, electronic measurements mean that we'll get information a lot more quickly. Currently, it takes about 21 days for people to mail the dairies to us and for us to transfer the information into a database. At some point, we may be able to do that overnight. And advertisers, who like electronic measurements, will have more confidence in radio. We've been testing this puppy in Philadelphia since mid-2000. Radio clients wanted to see it tested in a market with more Hispanics. The Hispanic market is difficult to survey because of language and culture. That's what brings us to Houston. Philly is 5 percent Hispanic; Houston is about 28 percent. Philly is 19 percent African American; Houston is about 16 percent. The two most difficult groups to capture, we have in abundance.