Modeling Marketing Data

Course Objective

• Demonstrate a critical understanding of the challenges in research with secondary data, including collection, cleaning, and analyzing the data.
  
• Demonstrate a critical understanding of the concepts and theories used, as well as key topics covered, in the top academic journals using marketing model on secondary data
  
• Provide students with the (software) tools and analytical techniques that will enable them to apply the marketing models and methods taught in the course to get insights in real-life international marketing problems and to design solutions to those problems
  
• Communicate and collaborate effectively with an international cohort of classmates to design and produce case solutions or other presentations, provided in English, to both academic and professional audiences.

Course Content

Companies often have vast amounts of data available (e.g. sales, marketing mix, customer information). Additionally there is an increasing amount of publicly available data (e.g. social media content, Google trends). These forms of data are called secondary data, as they are not collected with a specific purpose in mind. They do however hold valuable information for both researchers as well as managers. For example, for managers these secondary data can be useful for their dashboards monitoring sales, marketing-mix effectiveness, service recovery, etc. For research, the pallet is even wider and a summary will not do justice to the vast diversity of questions that you will be able to answer with secondary data.

To get an idea, the following examples are/were covered in the slides, assignments and exams of this and past years of the course: (i) with a database of IMDB movie ratings, budgets, etc., we consider which factors describe the revenue of those movies, (ii) with a combination of public data sources we consider the link between intended behavior and actual outcomes of sustainable governmental initiatives in multiple ways, such as renewable energy and roll out of electric charging stations, (iii) with a selection of website visitors that did vs did not click on a banner ad, we consider what customer characteristics and browsing behavior make a website visitor more likely to click on a banner ad, (iv) with a multi-source database (e.g. Twitter, Reddit, Google Trends) on the buzz (i.e. online chatter) around several top EA games we look at what aspects explain the customer “buzz” for a game over time, and (v) with a database of songs that made the top charts on Spotify we consider the role of song characteristics on popularity.

To answer these (and similar types of) questions this course covers collecting/acquiring, cleaning, structuring, managing, visualizing and analyzing this secondary data. For example, there will be a tutorial and assignment on how to collect data from the university library, as well as tutorials on how to collect of publicly available data (e.g. Google trends). As these data are often not as clean as the datasets you are used to work with, we also work on how to clean the dataset and how this affects the scope of our research. As examples of structuring and managing data you can think of decisions of aggregation and selection of the data. Regarding the methods this course builds on the methodological foundation of the Consumer Marketing Analytics, and focuses on advanced marketing models, e.g. multivariate regression, logistic regression, time-series analysis, hierarchical models, cluster analysis.

With these tools and methods in hand you will be able to understand the state-of-the art in academia, such as articles in JMR and Marketing Science, as well as get insights in how these tools and methods can be applied to get insights in real-life marketing problems. Some articles and topics will be explicitly covered in the lectures, amongst others, endogeneity, and (unobserved) heterogeneity.

Finally, students will use their growing toolkit in a series of assignments, that will prepare them for the different elements of the thesis, as well as in their future career. In particular, at the end of the course students should have a toolkit that can help them with each challenge concerning secondary data, from code to collect and structure data, to having a syntax for the marketing models discussed in this course.

Additional Information Teaching Methods

Lectures and online tutorials

Method of Assessment

Written examination: 70% (in two parts: dataset part in form of individual assignment and theory and interpretation as exam);
Assignments: 30%;
each to be completed with a minimum score of 5.0

Recommended background knowledge

The course builds on methods discussed in CMA.

This course uses R and RStudio as it is very useful for the cleaning and structuring the data. The course starts with an introduction on using R, so knowing R is not a requirement. However, the learning curve is steep, so students, especially those without any knowledge of R, need to invest time to get acquainted with the software. Next to the lectures and tutorials in the first week, some might need some extra material to jumpstart. Luckily, there is a vast amount of material publicly available, e.g. online blogs and tutorials.

In the end, using R is a vehicle and not a goal, and the use of other software like SPSS or STATA are also allowed. That said, students that do not invest in R might not get the full experience out of the course.

Finally, given that the course is focused on learning by doing, the course will have only a few readings, of which even fewer are mandatory readings. Instead, as said, students are expected to prepare for the assignments and invest time in preparing the basic concepts on which the course builds (e.g., watch and do extra tutorials in R and re-read tutorials/lectures of CMA).