Media Heresy

I have a criticism of media departments at clients and agencies: They like numbers too much. I’m a quantitative person myself, but sometimes I see that we’re beginning to rely on numbers less for the reality they represent and more for the comfort they give us. In the media world, we’ve come to put trust in numbers that turn out to be rather meaningless online.


Back in the 1880s, James Walter Thompson invented a system of making generalizations about people by classifying them by age, gender, and other factors and applying this to media choices. This “demography” that resulted launched his company into one of the top positions among ad agencies. Some say that he even invented the ad agency because of his applied insights.

And this demography worked — or at least it worked better than anything else around. It was true that women tended to hold certain beliefs and subscribe to differing publications that were in direct correlation to their age. Creative could be geared to them. Media choices could be adapted to them.

But all along, the connection was an indirect one. There is always the imperfect assumption that what you are and where you come from defines what you believe and buy. True enough to be useful but not correct enough to be completely true.

We in the media departments of the world have come to assume that demographics are a science that can’t be ignored. I frequently get asked why we did not include demographic information in targeting briefs or media rationales, and here’s why.

The Ghost of a Ghost

In the online environment, we don’t have very good demographic information. Sure, you can get generalizations about who’s appearing on one site versus another site by looking at various sampling studies done by companies like Nielsen//NetRatings. But these are indirect means of measuring. In other words, we are getting only a glimpse at the information that helps us guess what a person is likely to think or do. When you add these two levels of indirectness into the equations, we’re on very shaky ground.

Let’s say that demography gives us a 40 percent chance of being right about an individual’s behavior online and some study will give you a 40 percent chance of knowing the correct demography of that individual. Mathematically (I told you I was quantitative), we have to multiply the .4 likelihood against the other .4 likelihood to receive a very inaccurate 16 percent chance of being correct.

Stop and Smell the Flowers

All of this brings me to the alternative. Online was made for behavioral measurement. We don’t have to guess at who you are to guess at what you might do. We can look and see directly what you’re doing online.

It is my opinion that we media folks try too hard to make the online media look just like the offline media by forcing the same metrics on it. Instead, we should be exploiting new metrics, such as who is buying what and visiting where. Many of the profiling companies, such as Engage, provide media based on these more behavioral measures.

This means that media planners and buyers need to let go of our precious numbers a little bit and open our eyes to some of the new methods of purchasing media online. We need to be much more like our creative cohorts and see things with our “ad gut,” experimenting with our own suppositions of where our clients’ customers are most likely to be and which technologies and behavior information are most likely to show us the specifics we’re looking for.

Related reading