34%Churn Reduced
91%Model Accuracy
$2.4MRevenue Retained
30 daysEarly Warning

The Challenge

A regional telecom provider was losing approximately 4.2% of its subscriber base monthly to competitors. Their retention strategy was entirely reactive — calling churned customers post-cancellation with win-back offers that had a less than 8% success rate.

The business had rich transaction, usage, and support data but no analytical capability to turn it into predictive insight. Leadership challenged us to build a system that would identify at-risk customers 30+ days before their likely cancellation date.

Our Approach

1

Data Audit & Preparation

Ingested 3 years of customer data from billing, CRM, network usage, and support ticket systems. Cleaned and normalised 2.1M records, handling missing values and encoding categorical features.

2

Feature Engineering

Created 47 predictive features including usage trend scores, support escalation frequency, contract tenure ratios, and competitive market exposure indicators.

3

Model Development & Selection

Trained and evaluated 6 model types. XGBoost outperformed others with 91% precision and 88% recall on the hold-out test set. Explainability via SHAP values was preserved for business users.

4

Salesforce Integration

Deployed model via AWS SageMaker endpoint. Daily batch scoring pipeline pushes churn probability scores and top risk factors for each customer directly into Salesforce CRM fields.

5

Retention Playbook Design

Collaborated with the retention team to map churn probability score bands to specific intervention playbooks — personalised offers, service enhancements, or proactive support calls.

The Results

In the six months following deployment, the client's monthly churn rate fell from 4.2% to 2.8% — a 34% reduction. The retention team's outreach became highly targeted allowing them to focus efforts on the highest-value at-risk accounts.

  • Monthly churn rate reduced from 4.2% to 2.8% within 6 months of deployment
  • Model identifies 78% of actual churners 30+ days before cancellation
  • $2.4M in ARR protected in first year through targeted retention interventions
  • Retention team efficiency improved by 3x — fewer calls, higher success rate
  • Tableau dashboard gives management real-time churn risk visibility by segment

“Before this model, we were chasing customers out the door. Now we can have the right conversation with the right customer at the right time. It has changed our retention strategy fundamentally.”

— Head of Customer Retention, Regional Telecom Provider

Ready for Results Like These?

Let’s discuss how we can solve your unique challenges and deliver measurable impact.

Start a Conversation View All Case Studies