Key findings
42.7%
Month-to-month churn rate vs just 2.8% on two-year contracts
0.838
AUC-ROC — model ranks a churner above a non-churner 84% of the time
53%
New customers churn in first 6 months — most critical retention window
45.3%
Electronic check churn rate vs 15% for auto-pay customers
496
High-risk customers identified for immediate retention outreach
$236K
Estimated annual revenue protected at 30% retention success rate
Resources
Live interactive dashboard
krenzanalytics.com — Telco Churn Dashboard
7,043
Customers
26.5%
Churn Rate
0.838
AUC-ROC
$236K
Annual Savings
Churn by contract type
Month-to-month churns 15x the two-year rate
Churn by tenure group
First 6 months: 53% churn rate
Churn by payment method
Electronic check 3x auto-pay churn rate
ROC curve — model performance
AUC = 0.838 · better than random by a wide margin
Customer risk tiers — intervention guide
Based on predicted churn probability from logistic regression model
Risk tierProbabilityCustomers% of baseRecommended action
High Risk>70%49620.6%Immediate outreach — retention offer or account review
Medium Risk40–70%1,52163.1%Proactive check-in — contract upgrade incentive
Low Risk<40%39216.3%Standard service — monitor for changes in behavior
Annual revenue protected (30% retention)~$236,000
Krenz Analytics  |  IBM Telco Customer Churn Dataset  |  Diego Krenz, May 2026
Previous project
Sales & Customer Performance — UK Online Retailer
R · Regression · 541,910 transactions · £10.4M revenue analyzed
View project