CUSTOMER STORY | TELECOM
MTN raises churn detection accuracy from 45% to 57% across 21.4 million subscribers
MTN analyzed 6 months of data, 12.1 billion records, to identify and curb churn. TIMi’s analytics team processed the full dataset in under 5 days using social network analysis.
Churn detection accuracy, up from 45%
57%
Revenue generated through targeted retention actions
+26%
To process 12.1 billion records
<5 days
OVERVIEW
The challenge.
A very large subscriber base, and 6 months of raw data to analyze.
21.4 million subscribers
A base of more than 21.4 million subscribers and more than 95 million phone numbers.
12.1 billion records
6 months of data to analyze: 718 GB of text files and 12.1 billion records.
SOLUTION
What was done.
TIMi combined social network analysis with a predictive model.
Social network analysis
The full dataset was processed in under 5 days using social network analysis techniques.
Modeled dataset
The result was a modeled dataset of 21.4 million rows and 294 variables.
Predictive model
A predictive model was built on 657 variables, fine-tuned by a TIMi expert.
RESULTS
What changed.
Better churn detection and revenue generated by targeted retention actions.
Churn detection
Accuracy improved from 45% to 57%.
Retention revenue
Targeted retention actions generated 26% more revenue.
Speed
The entire dataset was processed in less than 5 days.
USE CASES BEHIND THIS STORY
Read the method.
Churn Prediction
Score every customer on the risk of leaving and see the variables behind it.
SEE FOR YOURSELF
Test it on your own data.
Run TIMi on your own data with a free trial, or talk to us about your case.
