replenishment campaigns, repeat purchase programs, new product introductions, and customer appreciation campaigns. We will also discuss loyalty programs and omni-channel marketing in the age of predictive analytics.
We recommend you focus on dollar value retention. If you don't, you could be retaining customers, but losing money anyway. Also, when measuring customer retention it is important to realize that not all churn is created equal. Losing an unprofitable customer is not nearly as bad as losing one of your best customers. Also, it is a lot easier, cheaper, and more effective to try and prevent a customer from leaving than it is to reactivate that customer after she has already stopped shopping with you. In this chapter we look at different churn management programs, from untargeted, applying equally to all your customers, to targeted, and we will cover proactive retention management and customer reactivation campaigns.
In order to use the predictive marketing techniques discussed in this book you need to acquire both a predictive marketing mind-set as well as certain predictive marketing technical capabilities. You need to evolve your thinking from being focused on campaigns, channels, and one-size-fits-all marketing to being focused on individual customers and their context. From a technology point of view you need to acquire basic capabilities in the areas of customer data integration, predictive intelligence, and campaign automation.
We live in an exciting and somewhat confusing time. A large number of new marketing technologies are becoming available every year. In this chapter, we will give you a high-level overview of the various types of commercially available technologies and describe what it would take to build a predictive marketing solution in-house from the ground up.
There is a huge career opportunity that comes from being an early adopter of new methodologies and technologies, predictive marketing and predictive analytics included. If you are uncomfortable with numbers and math, and fearful of getting started with predictive marketing, there are a couple of things you should know: business understanding trumps math, asking the right questions goes a long way, the best marketers blend the art and science of marketing, and there is a lot you can learn from others.
In general, consumers are willing to share preference information in exchange for apparent benefits, such as convenience, from using personalized products and services. When it comes to personalization, there are different types of customer information that can be used and consumers may feel different about one type of information over the other. Use common sense when considering whether a marketing campaign is delightful or creepy and consider the context of the situation. This chapter will provide some guidelines for dealing with customer data that will engender trust.
Predictive analytics will continue to find new applications inside and beyond marketing. Not only will more algorithms become available, but real-time customer insights will start to shape our physical world, including the store of the future. There are huge benefits for customers, companies, and marketers alike to get started with predictive marketing sooner rather than later. Sooner or later your customers and competitors will force you to adopt a predictive marketing mind-set, so you might as well be an early adopter and derive a huge competitive advantage.
About the Authors
I am a scientist by training; I am an entrepreneur at heart, driven by curiosity of knowledge and challenging status quo. In elementary school, I saw the opportunity to make a profit collecting fruit from mulberry trees from our school backyard and selling it on the street, enlisting my schoolmates to help me run this small business. With some prodding from my engineer parents, I followed in my older brother's footsteps to enter a PhD program in physics at Brown University, studying under Leon Cooper at The Institute for Brain and Neural Systems. Dr. Cooper has received the Nobel Prize in Physics for his work on superconductivity and later decided that the next big problem to solve was in neuroscience, decoding how we learn and adapt. He is a pioneer in learning theory since the early 70s, using both experimental neuroscience as a base as well as statistical techniques for understanding and creating learning systems, now popularly called machine learning. I worked on both biological mechanisms that underlie learning and memory storage as well as construction of artificial neural networks, networks that can learn, associate, and reproduce such higher level cognitive acts as abstraction, computation, and language acquisition. Although these tasks are carried out easily by humans, they have not been easy to embody as conventional computer program.
As I was getting close to graduating from the PhD program at Brown University around 1998, I noticed that the business world was mostly running on simple spreadsheets, and I wanted to apply a data science and machine-learning approach to business. This goal led me to work for McKinsey & Co., the premier strategy consulting firm that helps large companies formulate strategies based on a fact-based problem solving approach.
When I joined McKinsey & Co. in 1999, I was able to test drive some of this data scientific approach in a few studies. My first project was to help a large technology company improve sales coverage, scientifically matching the sales team with the customers based on customer needs, sales team's skill, and experience. The CEO was impressed with the results on paper, but was unable to operationalize the results in real life, in a repeatable way. This is what I call the last mile problem of analytics. I realized that this is a big problem to solve. Analytics is an important enabler in improving commercial efficiency, but can only create value if it becomes part of the day-to-day execution workflow. I saw this theme repeat over and over again in many areas of business, pricing, supply chain, marketing, and sales. Most McKinsey projects I have been part of ended up on a slide deck which had all the right answers but very rarely created any real value. Equipped with McKinsey training, I joined one of my clients, Micro Warehouse as VP of Marketing, in 2002, with the goal to bring data science to everyday operations. I was lucky to be empowered by the CEO Jerry York and President Kirby Myers. Jerry was the most analytically driven person I ever knew in business, still to this day. He was previously CFO of IBM during Gerstner years, and CFO of Chrysler before that. He encouraged me to use data science to help him run the business better.
I knew I had to architect my approach in a way that married data science with execution to solve the last mile problem. I had two important recruits, Dr. Michel Nahon, a brilliant Yale-trained applied mathematician who helped me with machine-learning algorithms, and the hacker extraordinaire Glen Demeraski, who helped me with everything database and application related. I created approaches and systems that used data to more efficiently allocate resources, reduce marketing costs, and uncover new revenue sources. We had significant impact on marketing efficiency, pricing, and discounting patterns as well as salesforce effectiveness. In early 2003 we had real-time systems alerting purchase, pricing, and customer acquisition patterns of the sales team compared to moving averages to take immediate action by the sales leadership. After Micro Warehouse, from 2004 to 2006, I joined Best Buy as Senior Director of Business-to-Business marketing of its newly founded Best Buy for Business division. Best Buy at the time also struggled with the same exact last mile problem, lots of internal resources, tools, many high-flying consultants talking about customer segmentation, and analytics, but when you walked into a store, none of that had any impact at the customer level. This is the true test of analytics; does it impact the customers in a positive way that they can experience it? If not, then you have the wrong setup. Making progress at Best Buy was much more difficult, which I will touch on in Chapter
