Where and Why Customers Churn: Revenue, Tenure & Behavior Analysis
Interactive Tableau dashboard analyzing 7,043 customers across contract types, revenue segments, and behavioral patterns
- Overview
- Dashboard Highlights
- Problem Statement
- Dataset
- Key Metrics
- Dashboard Structure
- Key Insights
- Business Implications
- Tools & Technologies
- Project Structure
- How to Use This Repository
- Limitations
- Future Improvements
- Author
This project analyzes customer churn behavior using a structured business approach, focusing on tenure, pricing, contract type, and customer behavior patterns.
The goal is to identify high-risk customer segments, understand key churn drivers, and derive actionable insights that can support retention strategies.
🎯 Quick Links:
- 📊 Interactive Tableau Dashboard — Explore live visualizations
- 📖 Data Dictionary — Complete column definitions and metadata
🔗 View Interactive Dashboard on Tableau Public
The interactive Tableau dashboard provides a comprehensive view of churn patterns across multiple dimensions:
Top-Level KPIs:
- Overall churn rate: 26.54%
- Total customer base: 7,043 customers
- Average revenue per customer: ₹1,384
- Average support tickets: 0.60 per customer
Key Visual Insights:
- Contract type analysis showing month-to-month contracts driving 42.7% churn
- Revenue segment analysis revealing ₹70-₹90 band as highest risk zone (40% churn)
- Payment method comparison with electronic check users at 45% churn rate
- Tenure-based churn curve demonstrating 47% churn in first 12 months
- Contract × Tenure heatmap identifying specific high-risk customer segments
Customer churn directly impacts revenue and long-term growth. However, churn is often driven by multiple interacting factors such as:
- Customer lifecycle stage
- Pricing sensitivity
- Contract structure
- Payment behavior
This project aims to answer:
- Where is churn highest?
- Why are customers leaving?
- Which segments require immediate attention?
The dataset includes customer-level information such as:
- Demographics (e.g., Senior Citizen, Dependents)
- Account details (Contract Type, Tenure)
- Financial data (Monthly Charges, Total Charges)
- Behavioral indicators (Support Tickets, Payment Method)
- Target variable: Churn Value (0 / 1)
Source Files (available in /data/raw):
Telco_customer_churn.xlsx— Customer demographics and churn dataOnline Retail.xlsx— Transaction and purchase behavior datasupport.csv— Synthetically generated support ticket data
For detailed column definitions and data sources, see data_dictionary.md.
| Metric | Value | Insight |
|---|---|---|
| Churn Rate | 26.54% | 🔴 Above industry average (20%) |
| Total Customers | 7,043 | 📊 Analyzed customer base |
| Average Revenue per Customer | ₹1,384 | 💰 Per customer contribution |
| Average Support Tickets per Customer | 0.60 | 📞 Low engagement with support |
The dashboard is designed to move from high-level KPIs → detailed drivers → actionable insights.
Provides a quick snapshot of overall business health and churn level.
Compares churn across contract categories to evaluate retention stability.
Analyzes churn across monthly charge bands to identify pricing sensitivity.
Highlights behavioral differences across payment types.
Tracks how churn evolves across the customer lifecycle.
Identifies high-risk zones by combining contract type and tenure.
Reveals concentration of churn across pricing segments and supports revenue-based decisions.
Early-stage customers (0–12 months) show the highest churn (~47%), nearly 2× higher than long-term customers, indicating weak onboarding or early dissatisfaction.
Month-to-month contracts are the primary churn driver (42.7%), significantly exceeding fixed-term plans, suggesting lack of commitment increases exit probability.
Customers in the ₹70–₹90 monthly charge segment exhibit peak churn (~40%), indicating a pricing sensitivity zone where value perception may be misaligned.
Electronic check users show the highest churn (45%), nearly 2× above the overall average (24%), highlighting behavioral or friction-related risk.
Support interaction shows negligible impact on churn (~0.15% difference), indicating that support volume alone is not a strong predictor of churn.
Action: Focus on onboarding, first 90-day engagement, and activation strategies.
Impact: Reducing early-stage churn by 10% could retain ~330 customers annually.
Action: Introduce incentives to shift users toward longer-term plans (discounts, bundle offers).
Impact: Every 5% shift to annual contracts could reduce overall churn by ~2%.
Action: Investigate value perception in the ₹70–₹90 segment—consider feature bundling or tier repositioning.
Impact: Aligns product value with customer expectations in highest-risk pricing band.
Action: Encourage transition from electronic check to automated payment methods through friction reduction or incentives.
Impact: Payment method optimization could reduce churn by 3–5% in this cohort.
| Category | Tools |
|---|---|
| Data Processing |    |
| Visualization |  |
| Data Sources |  CSV |
Technical Stack:
- Python — Data cleaning and preprocessing (Jupyter Notebooks)
- Pandas — Data manipulation and transformation
- Tableau Public — Dashboard development & visualization
- Excel — Original data sources (XLSX format)
| Strength | Description |
|---|---|
| 🎯 Business-Focused | Focus on business questions, not just visuals |
| 📊 Clear Flow | Data → insight → action pipeline |
| 🔍 Segmentation Analysis | Multi-dimensional analysis (contract × tenure) |
| 💼 Decision-Oriented | Emphasis on decision-making impact, not just reporting |
| 📚 Professional Documentation | Industry-level documentation with data dictionary |
| 🌐 Interactive Dashboard | Publicly accessible Tableau dashboard |
| Limitation | Description |
|---|---|
| 🔮 No Predictive Modeling | Purely descriptive analysis; does not forecast future churn probability |
| 🌍 No External Factors | Market competition, customer feedback, or economic conditions not included |
| ⚫⚪ Binary Churn Treatment | Churn treated as binary (0/1) without probability scoring or risk tiers |
| 🤖 Synthetic Support Data | Support metrics are artificially generated and may not reflect real-world patterns |
Implement ML-based risk scoring (Logistic Regression, Random Forest, XGBoost) to assign churn probability to active customers.
Combine churn risk with revenue forecasting to prioritize high-value retention efforts.
Track retention over time by signup period to identify long-term behavioral trends.
Integrate experimental data for pricing changes, contract incentives, or payment method transitions.
Connect to live data pipelines for operational monitoring and proactive intervention.
The analysis highlights that churn is primarily driven by early lifecycle risk, contract flexibility, and pricing sensitivity, rather than support interactions.
Addressing these factors through targeted onboarding improvements, contract incentive programs, and payment friction reduction can significantly improve retention and stabilize revenue.
🎯 Key Takeaway: Focus retention efforts on the first 12 months, month-to-month contract holders, and customers in the ₹70–₹90 pricing band for maximum impact.
📊 Explore the Full Analysis on Tableau Public
churn/
│
├── data/
│ ├── raw/
│ │ ├── Online Retail.xlsx # Original transaction dataset
│ │ ├── Telco_customer_churn.xlsx # Original customer churn dataset
│ │ └── support.csv # Synthetic support ticket data
│ │
│ └── processed/
│ ├── final_dataset.csv # Complete analytical dataset (~7K customers)
│ └── model_dataset.csv # ML-ready dataset with engineered features
│
├── notebooks/
│ ├── DataCleaning.ipynb # Data preprocessing and transformation
│ └── SupportDataGenration.ipynb # Synthetic support data generation
│
├── tableau/
│ ├── dashboard_preview.png # Dashboard screenshot for README
│ └── sales-customer-analytics-dashboard-2023.twbx # Tableau workbook
│
├── data_dictionary.md # Comprehensive data documentation
└── README.md # Project documentation
-
🌐 View Live Dashboard: Click here to explore the interactive Tableau dashboard
-
📖 Explore the Data Dictionary: Review data_dictionary.md for column definitions and data sources
-
📊 Review Raw Data: Original datasets available in
/data/rawTelco_customer_churn.xlsx— Customer demographicsOnline Retail.xlsx— Transaction datasupport.csv— Support tickets
-
💻 Review Notebooks: Check data cleaning and transformation logic in
/notebooksDataCleaning.ipynb— Data preprocessing pipelineSupportDataGenration.ipynb— Synthetic data generation
-
🔧 Reproduce Analysis: Use
data/processed/final_dataset.csvfor your own exploratory analysis -
🤖 ML Experimentation: Use
data/processed/model_dataset.csvfor machine learning experiments
- 📊 Live Dashboard: View on Tableau Public
- 📂 Complete Code: Available in
/notebooksfor technical review - 📖 Documentation: Comprehensive data dictionary and README for context
Arijit Mondal
Data Analyst | Business Intelligence Specialist
Focus Areas: Customer Analytics, Churn Prevention, Revenue Optimization
🔗 Connect:
This project is available for educational and portfolio purposes. If you use this work, please provide attribution.
Last Updated: March 3, 2026 | Status: ✅ Complete | Version: 1.0
⭐ If you found this project helpful, please consider starring it!