In my previous posts, I discussed marketing touch analysis as a way to find out which of your tactics are most involved in the process of getting to closed deals. Now I’ll show you how a little math magic can be used to predict better performance.
Marketing mix modeling started in the consumer-marketing world from the desire to know if and how different marketing techniques drive sales. Organizations executed a series of marketing tactics and looked at the results, perhaps in a region or among a group of stores. Using data mining methods and correlation techniques, including regression analysis, they could associate changes in marketing tactics with increased sales. Should the cereal be high or low on the shelf? Is a television ad in this market associated with higher sales of this product?
In b-to-b marketing, application of this approach has been challenging because of complexity, number of participants, duration of sales cycles and the role of the sales representative in the process. What’s changed, and is driving the growing use of these analytical methods, is the role of marketing in the early stages of the buying cycle. SiriusDecisions research indicates that b-to-b buyers complete 67 percent of their journey online. The sales rep has less of a role in these early stages. As a result, marketing analysis has shifted from modeling the sales rep’s interaction with buyers to measuring marketing’s online interaction with the buyer, making it possible to predict the future behavior of that buyer or similar buyers.
While consumer marketers are largely focused on increasing transaction rates and values, b-to-b marketers are taking a slightly different path. They are using mix modeling techniques to identify the best next tactic to present to a buyer as she or he moves through the sales cycle. These models are used to calculate the likelihood of progression through the sales cycle given an optimal set of tactics.
We estimate that 10 to 15 percent of b-to-b marketers (large companies with the infrastructure and resources to make investments in analytics) are using these methods. However, mix modeling is important to smaller companies as well, for a couple of reasons. First, it’s coming your way. You probably have the data for this kind of analysis, but if you don’t, it’s most likely because you are not collecting it. Second, companies are beginning to see this as an area of competitive differentiation. Now is a good time to take stock of the data available to you and start collecting information for predictive modeling.