Predictive Analytics - Customer Churn

Amanda Wright
Jun 23
1 min read

Research Goals

Our goal is to build a model to:

1. Identify factors contributing to customer contract renewal at SY, based on the data from the last telemarketing campaign.

2. Predict the probability of renewal for SY customers in the new campaign to classify customers as “targeted” and “untargeted” whilst covering the advertising-to-sales ratio.

Analyse the success of the strategy by:

3. Comparing the predicted profitability in the new campaign’s targeted approach vs. actual profitability of the previous campaign’s untargeted approach.

4. Using the model to quantify the impact of the cost in upcoming re-negotiations with the call centre.

Last Campaign Summary

In the last telemarketing campaign:

SY cold called all 4140 customers to renew their contract via a call centre, with 1086 customers renewing their contract
Renewal rate = 26.23% last campaign, below the industry average retention of 69% in 2025 in Fig. 1

Renewals Last Campaign

Advantages of Logistic Regression

If we can predict the probability a customer says yes to “Renew”, then we can:

1. “Target” only the customers with a predicted probability to “Renew” above the advertising-to-sales ratio of $7/$50 = 0.14, leaving the remainder “untargeted.”

2. Maximise the efficiency of our marketing efforts and improving profits at SY with less customers for the call centre to call, whilst covering our advertising-to-sales-ratio of $7 a contact for $50 revenue.