If you used the decision tree in the beginning of this chapter (see Exhibit 5.1), you were directed to this section because you have knowledge about what your customers purchase and are in a market where you can sell more products and services to them at the same time. In order to do this analysis, you must know what is being purchased together (generally speaking, what can be found in the same shopping basket per customer visit). Sometimes you also know who holds the basket, but even if you do not, you still can learn about what products should be co-promoted, placed together in stores, or bundled and presented as a new product.

Bundling Using Cluster Analysis

If you use a cluster analysis for bundling your products, you will get not only a series of product combinations that goes well together but also a segmentation model that divides your customers into distinct groups according to their user needs. To do this sort of analysis, you should have knowledge on a customer level: which customers purchase what, and some sociodemographic data for all customers in order to be able to describe the clusters afterward. You can read more about how to make a cluster analysis on DW data in Chapter 4 in the section “Needs-Based Segmentation Created from Data Warehouse Data.” Here is a brief summary of the four-step process.

1. Recode the data so that there is one variable per product or service and one line per customer.

2. Set the new variables so that if a customer ...

Get Business Analytics for Sales and Marketing Managers: How to Compete in the Information Age now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.