Demystifying Predictive Analytics

By Jamie Schissler On Apr 26, 2019

What if you had a crystal ball that could tell you which leads are sales-ready and which need some nurturing? Or what type of content will most engage a specific visitor; which customers are most likely to purchase which products and through which channel; or how much demand will increase during peak holiday time?

This is predictive analytics: The statistical analysis and identification of attributes that have a direct correlation to a future event or condition. There’s been so much hype about predictive analytics recently that it’s easy to forget it’s been around for over 50 years — from weather forecasting and flight planning to consumer credit scoring and fraud detection. From a marketing perspective, the accuracy it provides and the predictive opportunities it creates can substantially minimize risk while driving tremendous efficiency and performance gains. What is new is the availability of predictive analytics. Once the luxury and secret sauce for only the largest brands staffed with deep data and analytics teams, expertise, and infrastructure, PA is available to organizations and business of all sizes.

So, how can marketers drive adoption, establish predictive analytics competency, and use it to create better outcomes for businesses and experiences for customers? Let’s break down some of the basics.

Recognize the power of segmentation

Traditional marketing segmentation is about grouping audiences by historical attributes like age, gender, geography, etc. The emphasis of this approach is on understanding what individual consumers have done in the past and/or who they are at this moment in order to better focus and calibrate the brand’s sales and marketing efforts.

Predictive analytics can also be thought of as a segmentation strategy. The difference is it employs algorithmic models against much of the same underlying data in order to identify and group customers by what they are likely to do or prefer. As such, it’s useful to think of and approach predictive analytics as a segmentation strategy to identify customers by future behavior, such as those who are likely to travel soon, renew a lease, upgrade, or refer a friend.

Define specific business cases and develop effective strategies

Thinking of it as a segmentation strategy, it’s helpful to begin by defining those groups that represent the greatest opportunities or needs for your business. Perhaps your primary goal is lead velocity. Or maybe it’s churn reduction or lifetime value. Focus on one or two to start, otherwise, you’ll get overwhelmed quickly.

After you’ve defined your business and use cases, you’ll need to develop a strategy for how you are going to engage that audience. It’s one thing to be able to identify customers who are likely to do something. It’s another to figure out what that means for your business and how you are going to act on it. Both elements – identification and treatment strategy — are essential.

Identify the right data

Before rolling up your sleeves and diving into the data science and algorithms, identify the data required for your model. In some cases, you may find you simply don’t have the attributes or volume to create the segment with the degree of probability required. Focus on what you can do with the data you currently have, while simultaneously working on a longer-term roadmap and action plan, complete with business cases.

Prioritize consumer trust and respect

A few years ago a Target employee infamously used individual purchase data to build a model that would predict, with tremendous accuracy, which of its customers was likely pregnant. The audience was then subtly targeted with specific relevant products via direct mail. This example provides several important lessons when embarking upon predictive analytics. It underscores the potency and accuracy of predictive analytics while demonstrating that it doesn’t take a huge staff to develop and implement (though it does require a lot of data). More importantly, it demonstrates that in all things, trust and respect matter. The creep factor, invasiveness, and breach of privacy and trust in this application was a bridge too far.

To avoid such pitfalls, marketers should ensure there is governance and oversight in how predictive segments are defined and utilized. Socialize internally the segments the organization seeks to create and why, balancing out the needs of the business with customer preference, perception, privacy, and experience.

The power of predictive analytics

The predictive analytics landscape is an incredible opportunity to streamline the customer journey, deliver better experiences and value to your audiences, and boost overall business performance. Brands such as Netflix, through its recommendation engine, and Marriott are using PA to deliver messages at key decision points. Marriott uses it for better insight on the types of travelers likely to stay at their properties; for insight on and understanding of what attracts travelers to it over other brands; and to better understand the unique travel patterns and needs of those who travel for business versus those who travel for pleasure.

What’s really exciting is not just what brands can do with PA, but that the variety of PA technologies and solutions available today make it accessible to every organization. Plan carefully and deliberately, advance incrementally, and implement governance that reflects your organizational values and goals and you’ll be on your way to delivering exceptional customer experiences and accelerating growth.