Today’s marketers have access to a wealth of information about their customers and prospects, providing an opportunity to communicate and satisfy needs with ever increasing effectiveness. A big obstacle to realizing this opportunity is the overwhelming amount of information, but professionally created predictive models can help organizations leverage this information smartly within a framework of respect for customer privacy and security.
This type of modeling can identify business opportunities, such as who is most likely to purchase, or threats, like who is most likely to terminate a relationship. The benefits of predictive modeling are equally relevant to companies and non-profits; for example, a bank may want to generate new account openings, while a trade association may want to identify who will register for a conference.
In an age of low cost, internet-driven mass marketing (i.e., email is “free”), why should an organization use predictive modeling to drive a targeted approach to engagement?
- Reason # 1: Even when email is “free,” overuse of online communication creates a noisy environment that turns off customers. Targeting with predictive modeling identifies a subset of consumers most receptive to a particular message, which increases the relevance and effectiveness of internet marketing.
- Reason # 2: You realize dollar savings when carefully and scientifically identifying consumers who are most likely to be interested in a message or product, while avoiding those who are unlikely to respond. This is particularly important when relying on more expensive channels such as direct mail or telemarketing. This strategy also gives a clearer picture of the effectiveness of a marketing campaign as consumers who were unlikely to ever engage (regardless of the message or offer) are not counted among those who were not persuaded.
- Reason #3: Predictive modeling is a valuable cross-selling tool. It leverages information on current product usage and customer traits to identify the next best opportunity. And, cross-selling to existing relationships can be a far more efficient way to boost sales than acquiring new customers.
As an example of the power of predictive modeling, a professional association client was interested in maximizing the effectiveness of marketing a major conference and worked with Rockbridge to create a statistical tool using member data. Based on a model that predicted attendance to the previous years’ conference, we were able to score each member by their likelihood to attend the next conference. In fact, members with predictive scores in the top 25% made up nearly 70% of conference attendees, defining a target audience where communications were much more likely to be successful. Equipped with this knowledge, the association could focus on marketing to members who were the most receptive to attending the conference based on factors like geography, past attendance patterns, and professional traits, allowing the client to increase the impact of their marketing budget.
Customer information can effectively identify potential threats and guide efforts to mitigate them. The same methods for identifying likely purchasers can be used to identify customers prone to terminate a relationship, relying on clues as subtle as changing an address or reducing purchasing. To illustrate, a banking client wanted to identify customers with a high risk of terminating the relationship so that it could take proactive steps to retain these customers. Rockbridge developed a predictive model to identify characteristics that foreshadowed closing all accounts 6 months in advance, giving the client a clear targeting strategy for retention efforts.
Predictive modeling using big data is a complementary source of information to traditional market research and much of its value may lie in the ability to narrowly target groups of consumers. Our use of big data to predict future behavior has helped clients gain a deeper understanding of who their most loyal customers are and who may need special attention to fulfill their potential. When used in combination with traditional research methods to uncover the “why” of consumer behavior, predictive analytics can facilitate sending tailored messages to specific segments of consumers in a way that may be difficult or prohibitively expensive to do otherwise. The end result is providing a varied group of consumers with the message most likely to resonate with them.
Written by: Charles Colby, Chief Methodologist