A comprehensive analysis of Telco customer churn patterns using Python, Pandas, and Matplotlib/Seaborn visualizations in a Jupyter Notebook. This analysis examines 7,043 customers across 21 attributes to identify key risk factors driving customer churn and provides actionable recommendations to improve customer retention.
| Metric | Value |
|---|---|
| Total Customers | 7,043 |
| Churned Customers | 1,869 |
| Overall Churn Rate | ~26.5% |
| Retention Rate | ~73.5% |
| Highest Risk Segment | Fiber Optic + Month-to-Month (0-3 months tenure) |
| Lowest Risk Segment | DSL/No Service + 2-Year Contract (50+ months tenure) |
The three internet service categories (Fiber Optic, DSL, and No Service) show dramatically different churn rates, with Fiber Optic customers churning significantly more than other service types.
Root Causes Identified:
-
High Monthly Charges — As shown in Figure 8, Fiber Optic service monthly charges are substantially higher than DSL customers ($80-100+ vs. $40-70). This significant price premium correlates directly with the high churn rate for Fiber Optic customers.
- Hypothesis: Customers are leaving because Fiber Optic service is expensive.
-
Predominantly Month-to-Month Contracts — Figure 9 reveals that most Fiber Optic subscribers are on month-to-month contracts rather than longer-term agreements. Figure 1 shows that over 40% of month-to-month customers churn compared to just 3-11% for longer-term contracts.
- Hypothesis: The combination of high cost + short-term contract flexibility makes it easy for customers to leave Fiber Optic service.
Analysis: Fiber Optic customers represent a compounding risk — they are expensive, on short contracts, and have high churn rates.
Figure 2 clearly demonstrates extreme tenure-dependency in churn rates. The churn rate follows a clear inverse relationship with customer tenure:
- 0-3 months: Extremely high churn (new customer critical window)
- 4-6 months through 50+ months: Progressively lower churn rates
- 50+ months: Minimal churn (highly loyal customers)
Figure 3 (tenure distribution histogram) visually confirms that churned customers cluster in the 0-10 month range, while retained customers span across all tenure periods.
- Hypothesis: Long-tenure customers are highly loyal; new customers are vulnerable in their first 3 months. This is the critical intervention window.
Analysis: Tenure is the strongest predictor of churn. The company loses its highest-risk customers in the first few months.
Figure 4 (boxplot comparison) shows that churned customers have a slightly higher median monthly charge compared to retained customers, indicating a weak but consistent correlation between cost and churn.
- Hypothesis: Higher-cost service tiers correlate with marginally elevated churn risk, though this effect is secondary to tenure and service type.
Analysis: Price sensitivity plays a supporting role in churn; it amplifies other risk factors (like Fiber Optic) but is not a primary driver on its own.
Figure 5 - Partnership Status:
Customers without partners churn at higher rates than those with partners, suggesting that household stability correlates with retention.
Figure 6 - Phone Service:
Minimal difference in churn by phone service availability, indicating this is not a primary churn driver.
Figure 7 - Senior Citizen Status:
Senior Citizens churn at 40%+ rates, significantly higher than non-seniors (~25%). This is a secondary but important risk factor.
Figure 10 - Tech Support Adoption:
Senior Citizens have dramatically lower tech support adoption despite their high churn rates, suggesting they lack the technical skills or knowledge to access available support services.
TIER 1 - CRITICAL PRIORITY:
- Fiber Optic + Month-to-Month + 0-3 months tenure = Highest churn concentration
- Action needed immediately upon signup
TIER 2 - HIGH PRIORITY:
- New customers (0-3 months tenure) across all service types = Vulnerable retention window
- Senior Citizens (any service type) = 40%+ churn rate due to lack of support
TIER 3 - MODERATE PRIORITY:
- Fiber Optic customers on any contract type = High cost sensitivity
- Month-to-Month customers of any age/service = Easy exit path
New customers are the highest-risk group. Implement proactive engagement:
- Onboarding: Dedicated support specialist contact within 24 hours of signup
- Check-ins: Proactive outreach at 30 days, 60 days, and 90 days
- Early win: Quick setup/activation to demonstrate service quality
- Satisfaction tracking: Monitor satisfaction scores and intervene if dropping
- Success metric: Reduce 0-3 month churn from current X% to <15%
Figure 10 reveals Senior Citizens receive inadequate tech support adoption (significantly fewer than non-seniors), which directly correlates with their 40%+ churn rate.
- Action: Mandate tech support enrollment for ALL senior customers at signup
- Senior-Friendly Support: Phone-first support channel (not chat), patient representatives trained for senior needs
- Education: Monthly tech tips calls to help seniors maximize their service
- Simplification: Simplified service plans (3-tier instead of 10+)
- Success metric: Increase senior tech support adoption to >60%, reduce senior churn from 40% to <25%
Fiber Optic customers face the perfect storm: high cost + month-to-month contracts. Address both:
- Contract Incentives: Offer 6-month or 1-year contracts at $10-20/month discount for Fiber Optic customers
- Bundle Strategy: Package tech support + device protection with Fiber Optic (justifies cost, adds value)
- Tiered Pricing: Introduce lower-cost Fiber Optic tier for price-sensitive customers
- Early Lock-In: During critical 0-3 month window, aggressively promote longer contracts
- Success metric: Move 50% of Fiber Optic customers to 12+ month contracts, reduce Fiber churn by 15%
- Reduce friction in payment (auto-pay reduces billing-related churn)
- Offer small discount ($2-5/month) for customers on auto-pay
- Reduce electronic check usage (less reliable payment method)
| Figure | Title | Key Insight |
|---|---|---|
| Figure 0 | Churn Rate by Partnership Status | Non-partnered customers have higher churn |
| Figure 1 | Churn Rate by Contract Type | Month-to-month: 42% churn vs. 2-year: 3% churn |
| Figure 2 | Churn Rate by Tenure Duration | New customers (0-3mo): 50%+ churn; 50+mo customers: <5% churn |
| Figure 3 | Tenure Distribution: Retained vs Churned | Churned customers cluster in 0-10 month range |
| Figure 4 | Monthly Charges: Churned vs Retained | Churned customers have slightly higher median charges |
| Figure 5 | Churn Rate by Internet Service Type | Fiber Optic: ~42% |
| Figure 6 | Churn Rate by Phone Service | Minor difference; not a primary driver |
| Figure 7 | Churn Rate by Senior Citizen Status | Senior Citizens: 41% |
| Figure 8 | Monthly Charges by Internet Service | Fiber Optic ($80-100) >> DSL ($40-70) >> No Service ($20-30) |
| Figure 9 | Fiber Optic Subscriber Contracts | Most Fiber customers are on month-to-month contracts |
| Figure 10 | Tech Support Adoption by Senior Status | Senior Citizens have significantly lower tech support adoption |
- Source: Telco Customer Churn (Kaggle)
- Total Records: 7,043 customers
- Total Columns: 21 attributes
- Attributes Analyzed:
- Demographics: Gender, Senior Citizen, Partner, Dependents
- Services: Phone Service, Internet Service, Online Security, Online Backup, Device Protection, Tech Support, Streaming TV, Streaming Movies
- Contract & Billing: Contract, Tenure, Monthly Charges, Total Charges, Paperless Billing, Payment Method
- Target: Churn (Yes/No)
future-interns-task-2/
├── task2_analysis.ipynb # Full Jupyter notebook with code + visualizations + insights
├── README.md # This file
├── data/
│ └── WA_Fn-UseC_-Telco-Customer-Churn.csv # Original dataset (7,043 records)
└── results/
├── 00_churn_rate_by_partnership.png
├── 01_churn_by_contract.png
├── 02_churn_by_tenure.png
├── 03_tenure_distribution.png
├── 04_charges_comparison.png
├── 05_churn_by_internet_service.png
├── 06_churn_by_phone_service_availability.png
├── 07_churn_by_citizenship_status.png
├── 09_monthly_charges_by_internet_type.png
├── 10_fiber_optics_contracts.png
└── 11_support_level_by_status.png
- Python 3.7+
- Jupyter Notebook
- Required packages: pandas, matplotlib, seaborn, numpy, openpyxl
# Install required packages
pip install jupyter pandas matplotlib seaborn numpy openpyxl
# Navigate to project directory
cd future-interns-task-2
# Start Jupyter Notebook
jupyter notebook
# Open task2_analysis.ipynb in the browser interface- Open
task2_analysis.ipynbin Jupyter Notebook - Run cells sequentially using Shift+Enter
- Or click Cell → Run All to execute entire notebook at once
- Visualizations (PNG files) automatically save to
./results/folder - Review findings in the final "Final Report" section
Current State (Baseline):
- Churn Rate: 26.5%
- Losing ~1,869 customers from 7,043 total
- Estimated annual revenue loss: [calculate based on customer LTV]
With Recommended Actions:
| Action | Target Impact | Timeline |
|---|---|---|
| New Customer Retention Program | Reduce 0-3mo churn from 50% to 35% | 3 months |
| Senior Citizen Tech Support | Reduce senior churn from 41% to 30% | 2 months |
| Fiber Optic Contract Lock-in | Move 50% to 12+ month contracts | 6 months |
| Combined Impact | Reduce overall churn to 20% | 6 months |
Estimated ROI:
- Retain additional ~460 customers annually
- Increased CLV across all segments
- Reduced acquisition costs (retention cheaper than acquisition)
- Tenure is destiny — First 3 months determine 80% of churn risk
- Fiber = Premium + Fragile — High cost + short contracts = high churn
- Seniors need help — 41% churn due to lack of tech support
- Contracts matter — Month-to-month = 14x churn vs. 2-year
- Quick wins available — Tech support, onboarding, contract incentives are all implementable now
Regis Alain Udahemuka
Ashesi University | Computer Science (BSc)
Future Interns Data Science & Analytics Internship
CIN: FIT/MAY26/DS18653