ACC

Predictive Churn Model

Predictive Churn Model 

Four Steps

1. DEFINE THE PROBLEM

2. PREPARE THE DATA

3. ESTIMATE THE MODEL

4. SELECT CUSTOMER TO TARGET

Source: Blattberg, R. C., Kim, B. D., Neslin, S. A.(2008). Database Marketing. Springer New York.

Predictive Churn Model 

Four Steps

1. DEFINE THE PROBLEM

2. PREPARE THE DATA

3. ESTIMATE THE MODEL

4. SELECT CUSTOMER TO TARGET

Source: Blattberg, R. C., Kim, B. D., Neslin, S. A.(2008). Database Marketing. Springer New York.

Identify customers that will "leave" Hartlauer

Create indicators from Hartlauer sales and customer data

1. Decision Tree estimation

2. Predictive Model

Targeted mailing,

tailor-made offers, etc.

Prepare the Data

Sales Data - What does LEAVE mean?

Average Frequency 

Minimum Recency

Estimate the Model

First Decision Tree on Sales Data

Prepare the Data

Customer Data 

Year of Birth

Gender

is completely

balanced 50/50

Prepare the Data

Indicators... your ideas?!

  • Recency,
  • Frequency,
  • Monetary Value,
  • Pillar Count,
  • Pillar Mode,
  • Dummies of Pillars,
  • Dummies for Bundesländer,
  • Age,
  • Gender,
  • ...

Estimate the Model

Second Decision Tree on Sales and Customer Data

(+ Birth and Gender)

Estimate the Model

Over Time

Predictive Churn Model

By fundf

Predictive Churn Model

  • 609