There is a lot of talk about how predictive models drive marketing decisions. Clients are now asking for agencies to “predict” the channels that will drive the majority of new customer growth and how much ROI a campaign will generate.

For many of us, the first question is often “How the heck are we gonna do that?”

To answer that question, it helps to take a step back and talk about something we might be more familiar with; crying babies. Crying babies are a great example of a predictive model. Figuring out why babies cry can be scary to new parents. But they quickly realize babies cry for mostly three reasons: they are hungry, they are tired, or they are in some sort of discomfort. Based on what you see (or don’t see), you can figure out why your baby is crying.

A predictive model works in much the same way.  Let’s say a consumer packaged goods (CPG) client asks how much of their sales really come from Facebook. Just as we know infants cry, we know Facebook ads influence consumers to buy. So, we need to figure out how Facebook influences consumers.

To do that, we need to understand the motivations that influence consumers to use Facebook (or put another way, the reasons why people use Facebook) and the behaviors they demonstrate as a result. 

Again, if we think about our baby, the behavior is the crying and the reasons are most likely a combination of hunger, discomfort or needing a nap. For Facebook, there is a wealth of behavior and motivation data that already exists, and data scientists use this data to calculate which of the motivations have the strongest relationship to customers who use Facebook, just as you would to determine what was most likely motivating the baby to cry.

Keep in mind – just as there are only a few reasons why babies cry most of the time, there are only a few reasons why consumers who use Facebook buy CPG products. This does not mean that they are the only reasons. It means that these are the most likely reasons.

There may be hundreds of reasons why Facebook users buy a CPG product, but there are only a few that are really closely associated with buying. So, if you could say to a client that 80% of the time customers buy from you, they had seen a Facebook ad, I’m sure you would agree that any client would find value in that “prediction.”

To learn more about predictive models and how to use research and analytics to drive your next CPG campaign, contact us