As a skeptic, I've watched the ascendancy of mix models for the last 20 years and I have been consistently puzzled by the ad industry's willingness to be measured by ad weight alone. Creative agencies, media buying groups, networks, copy researchers and advertisers all seemingly agreed to having their efforts evaluated by GRPs alone. This is because GRPs were and continue to be THE advertising input to mix models. In fairness, a few modelers have played around with other measures in the last few years, but for most of the last two decades GRPs alone were the sole model input for advertising.
Does that mean we could air 30 seconds of a poster that reads "compliments of a friend" and get the same value as the most effective 30 second commercial ever made? Or, that 50 GRPs in network prime have the same value as 50 GRPs in post-midnight syndication? From the model's point of view, it all collapses to ad weight, measured by GRPs.
From a formal measurement standpoint, the GRP is just plain flawed. In the days that people concerned themselves with the failed concept of "effective frequency" it sort of made sense. But in today's world where we know that reach is the drive wheel, why should we multiply the two measures to arrive at a confounded one?
It's as if we're describing a man as being "big." A basketball player at 7' 9'' is big, but at only 250 pounds, he would be a stick figure. A sumo wrestler at 5' 10'' and 350 pounds is also big. The problem here is that the word "big" is an indeterminate concept. You can get to "big" in lots of ways. The concept of GRPs is also indeterminate; you can get to 100 GRPs in dozens of ways, and the underlying value assigned to GRPs assumes that all of those ways are equally valuable.
GRP Counts Reward Inefficient Buying Practices
Accumulating reach while minimizing wasted frequency is not an easy thing to do. Consider two plans, each delivering 40% reach. Plan #1 was assembled very carefully in order to minimize frequency, while plan #2 allowed frequency to accumulate at a faster rate.
Plan # 1 = 40% reach X 2.0 frequency = 80 GRPs
Plan # 2 = 40% reach X 2.5 frequency = 100 GRPs
Should plan # 2 be valued at 25% more ad weight than Plan # 1? I don't think so, and I think most media professionals would agree with me. But that's the way mix models would treat ad weight in this case.
The GRP calculation rewards buying practices that are "reach inefficient." Assume a really good planner could buy 50 reach points at a frequency of 2.0. A chimpanzee throwing darts might get to a 25% reach with a frequency of 4.0. Both plans work out to 100 GRPs, but would any reasonable media pro consider these two plans to be equal? The mix models do.
Reach and Frequency are NOT Independent Measures
You can't buy more reach without increasing frequency. Erwin Ephron used to say, "Frequency is the by-product of buying enough reach to make a difference." If frequency is truly the by-product of accumulating reach, doesn't the GRP calculation "double count" those naturally occurring (and unavoidable) frequency counts? If a savvy media planner can't avoid a growing frequency count as the plan accumulates reach, why should the GRP calculation reward that natural and unavoidable accumulation of frequency? That's exactly what happens when you multiply reach by frequency. So, why not simply use both as separate inputs?
Plus, the frequency effect of adding reach isn't linear -- adding 10 reach points to a base of 20% reach might increase frequency by "X." Moving up by those same 10 reach points, this time from 60% reach to 70% reach, you might easily triple that "X" increase in frequency. The higher the reach level, the more frequency you add with each step upwards. And then, you multiply the two numbers to get to GRPs! That can't be a good thing.
While you can't buy more reach without increasing frequency, the same isn't true in reverse. You can buy more frequency without increasing reach. Multiple spots in the same program would increase frequency without any real effect on reach. And, because of the way that GRPs are calculated, and treated in a mix model, you make it look like there's more "ad weight."
Into the Brave New World
Today there's a movement in advertising research to develop a "common GRP across screens." Why would any reasonable researcher want to apply the troubled GRP metric across TV, mobile and computer screens? Cross screen measurement is going to be hard enough; why would we choose to compound the difficulty with a unit of measure as flawed as GRPs?
Bottom line, the value of each contributor to advertising effects is grossly misrepresented when the only input into the models is GRPs. And every one should be upset with this and should demand better.
In our next column, we'll look at what the models do with/to GRPs in order to estimate the sales value of advertising. And that discussion leads us to a battle of the media planning heavyweights -- Simon Broadbent versus Erwin Ephron. Stay tuned.
The opinions and points of view expressed in this commentary are exclusively the views of the author and do not necessarily represent the views of MediaVillage.com management or associated bloggers. Image at top courtesy of freedigitalphotos.net.