Customer analytics is about applying simple models to understand and predict customer behavior, from the level of the individual customer, as in true one-to-one marketing, to the aggregate level of the entire customer base, which may be a good proxy for firm value. Companies have more data about their customers than ever before. Yet they may be “drowning in data” yet “starving for insights,” because they have no way to organize their data within a larger statistical framework. The role of the model is to summarize patterns and generate predictions of customer behavior in the future. |
The models we cover fall into two broad categories: next-period models, and long-term models. Next-period models are good for predicting what happens immediately following some action. For example, customer targeting or selection falls into this category. Long-term models use a long horizon and are for forecasting retention over time, or calculating customer lifetime value or customer equity (the value of the customer base).
We will examine a few types of models to analyze customer behavior, including:
- Recency-Frequency-Monetary (RFM) analysis
- Logistic regression
- Decision trees
- Bernoulli process models
- Empirical Bayes methods
We will use these approaches to answer several substantive questions such as:
- Test marketing: When should a test be conducted, how large should the test be?
- Targeting: which customers should be selected for e.g., acquisition, retention, cross-selling, direct mailing?
- Retention: what is the retention rate, and how does it change when customers are heterogeneous (i.e., different) and may drop out, or churn, over time?
- Customer lifetime value (CLV): how do you calculate the value to the firm of the customer over his or her lifecycle? What is the value of a firm’s entire customer base, i.e., customer equity (CE)?
Managers (as well as consultants, analysts, and investors) are increasingly tasked with providing valid answers to such questions. Yet many of them lack the tools to address these problems. This course is designed to give you the powerful, cutting-edge tools to address these issues.
The course is organized around several case studies that illustrate an important concept with data. All these examples will be “hands-on” and have an emphasis on real-time problem solving. You will develop the necessary skills to estimate these models and evaluate their results in SPSS & Excel.
- For a given model, students should be able to interpret results and explain intuitively what it assumes about customer behavior.
- Students should understand how to validate models and avoid overfitting.
- Students should be able to fit these models to real-world data; they should also be able to evaluate predictions across models and derive the managerial implications.
- Students should be able to calculate customer lifetime value (CLV).
- Understand the role of unobserved heterogeneity in measuring customer loyalty.
- Cluster B courses are OPEN for self-enrollment in OSIRIS for MSc Marketing Analytics students .
- Cluster B courses are NOT open for self-enrollment in OSIRIS for MSc Marketing Management students.
You will get a link to a Google form (via the MSc Marketing General Information Blackboard page) where you can select a maximum of 2 cluster B courses until September 16.
After September 16 students will be transferred into the OSIRIS courses and automatically added to the Blackboard course of the course.
- The courses are NOT available for students outside MSc MM/MA, except for MSc econometrics students, who have to write an email requesting admission before September 16th to TiSEM-MSc-Marketing-Management-Research@uvt.nl.
Type of instructions
Lectures, Computer Labs
Type of exams
Assignments, computer final exam
- Selected articles. The reading list will be available on Blackboard before the start of the course..
- Other material will be announced on Blackboard, no need to purchase any materials in advance..