Predictive Analytics for Marketing

Becoming familiar with the three case studies that will be used throughout the course: automotive (manufacturer), travel (cruise lines), and consumer packaged goods (retailer). Exploring sample targeted marketing campaigns, incremental revenue goals relative to advertising costs, and approaches to finding the appropriate audience.

Exploring potential data sources for three case studies. Determining how to define the marketing question for each using frequency of purchase, value of data relative to cost, and data specific to a use case versus across use cases.

Introducing off-the-shelf segmentation. Sketching out rule-based segmentation for each use case: product-based, location-based, and market basket-based. Determining how to apply different contact strategies for each use case: banner ads, email, direct mail, and apps.

Introducing k-means segmentation. Demonstrating a sample segmentation using Excel. Translating from customer segmentation to activation in each use case: product-based, location-based, and market basket-based.

Introducing marginal return on investment. Exploring techniques for appropriate audience sizing. Examining theory on linear versus logistic regression tying back to the three use cases: infrequent purchase, annual purchase possibility, and weekly purchase.

Introducing expansion from previous models to multiple product lines. Sharing techniques for appropriate audience sizing. Determining whether a contact strategy is appropriate at a high level across use cases.

Introducing expansion from campaign focus to lifetime value. Sharing techniques for appropriate audience sizing. Determining whether there is appropriate lifetime value across use cases.

Tying it all together into a coherent story for each use case. Sharing the latest in data ethics and privacy. Considering inherent error in predictive modeling and the associated risks.