Data Analysis Using SQL and Excel, 2nd Edition

CHAPTER 6 How Long Will Customers Last? Survival Analysis to Understand Customers and Their Value

How long will a lightbulb last? Which factors influence a cancer patient’s prognosis? What is the average time to failure (MTTF) of a disk drive? These questions may seem to have little relationship to each other, but they do have one thing in common: They all involve estimating time to an event, so they can be answered using survival analysis. And, these ideas apply to customers, their tenures, and their value.

The scientific and industrial origins of survival analysis explain the terminology. Its emphasis on “failure” and “risk” and “mortality” and “recidivism” may explain why, once upon a time, survival analysis did not readily catch on in the business and marketing world. That time has passed, and survival analysis is recognized as a powerful set of analytic techniques for understanding customers. And, the combination of SQL and Excel is sufficiently powerful to apply many of these techniques to large customer databases.

Survival analysis estimates how long until a particular event happens. A customer starts; when will that customer stop? By assuming that the future will be similar to the past, historical customer behavior can help us understand what will happen and when.

Subscription relationships have well-defined beginnings and ends. For these, the most important time-to-event question is when customers stop. Examples abound:

• Customers get a mortgage and remain customers ...