Customer Lifetime Value Optimization Through Proactive Health Engagement

CX Data Analyst Portfolio Project - Insurance Industry

📋 PROJECT OVERVIEW

Business Problem: Insurance companies face high customer acquisition costs and increasing competition. Retaining existing customers while promoting preventive health behaviors can simultaneously reduce churn and lower claim costs.

Analysis Goal: Identify actionable strategies to improve customer lifetime value through targeted health engagement programs, with specific focus on retention drivers and wellness incentives.

Key Questions:

What factors drive customer retention and long-term value?
How does health engagement correlate with customer loyalty and costs?
Which customer segments are at highest risk of churn?
What is the potential ROI of wellness incentive programs?
Where should we focus geographic expansion or retention efforts?

🗺️ PROJECT STRUCTURE

Phase 1: Data Preparation & Exploration

Phase 2: Customer Segmentation & Retention Analysis

Phase 3: Health Engagement Scoring System

Phase 4: Geospatial & Temporal Analysis

Phase 5: A/B Test Simulation & ROI Modeling

Phase 6: Dashboard & Recommendations

📊 PHASE 1: DATA PREPARATION & EXPLORATION

1.1 Data Cleaning

Tasks:

Handle missing values (BMI: 990 missing, Year_last_admitted: 11,881 missing)
Standardize categorical variables (occupation typo: "Salried" → "Salaried")
Create data dictionary documenting each feature
Validate data ranges (age, BMI, insurance_cost)

Deliverables:

Clean dataset saved as insurance_clean.csv
Data quality report (% missing, outliers identified)
Cleaning script with comments

1.2 Exploratory Data Analysis (EDA)

Tasks:

Univariate analysis: Distribution of key variables
Bivariate analysis: Insurance cost vs. health metrics
Correlation matrix: Identify multicollinearity
Outlier detection: Flag anomalous insurance costs

Visualizations:

Distribution plots (age, BMI, insurance cost)
Box plots by categorical variables (occupation, gender, location)
Correlation heatmap
Cost distribution by tenure

Skills Showcased: Data cleaning, pandas, visualization (matplotlib/seaborn)

👥 PHASE 2: CUSTOMER SEGMENTATION & RETENTION ANALYSIS

2.1 Customer Tenure Segmentation

Create 3 cohorts:

New Customers (0-2 years): Churn risk, acquisition focus
Established Customers (3-5 years): Growth opportunities
Loyal Customers (6-8 years): Retention champions

Analysis:

Demographic profile of each cohort
Average insurance cost by cohort
Health engagement levels by cohort
Dual coverage rate by cohort (competitive threat indicator)

2.2 Churn Risk Scoring

Build a churn risk model:

Target variable: Create "At-Risk" flag (0-1 year tenure + dual coverage OR 0-2 years with low engagement)
Features: Health engagement, demographics, cost, lifestyle
Models to test:
- Logistic Regression (interpretability)
- Decision Tree (rule-based insights)
- Random Forest (accuracy benchmark)

Deliverables:

Feature importance plot
Confusion matrix & ROC curve
Top 10 churn risk factors
Customer churn risk scores (exportable list)

2.3 Customer Lifetime Value (CLV) Estimation

Simplified CLV formula:

CLV = (Avg Annual Premium × Avg Tenure) - (Avg Claims Cost*)
*Use insurance_cost as proxy for claims/risk

Segmentation:

High CLV vs. Low CLV customers
Characteristics of each group
Actionable insights for retention

Skills Showcased: Predictive modeling, classification, business metrics

💪 PHASE 3: HEALTH ENGAGEMENT SCORING SYSTEM

3.1 Create Health Engagement Score (0-100)

Components (weighted):

Regular checkups (30%): 0-5 scale → 0-30 points
Exercise frequency (25%): None=0, Moderate=15, Extreme=25
Smoking status (25%): Smokes=0, Unknown=10, Formerly=15, Never=25
Alcohol consumption (10%): Daily=0, Rare=5, No=10
Doctor visits (10%): Normalized 0-10 scale (fewer = better, preventive care)

Formula example:

engagement_score = (
    (regular_checkup * 6) +
    (exercise_points) +
    (smoking_points) +
    (alcohol_points) +
    (doctor_visit_normalized)
)

3.2 Engagement Analysis

Tasks:

Distribute customers into 4 tiers: Low (0-25), Medium (26-50), High (51-75), Elite (76-100)
Analyze insurance cost by engagement tier
Identify "high-risk, low-engagement" customers (high BMI/cholesterol + low score)
Compare engagement scores across tenure cohorts

Visualizations:

Engagement score distribution
Scatter plot: Engagement Score vs. Insurance Cost
Box plot: Engagement by tenure cohort
Heatmap: Engagement score by demographic segments

3.3 Persona Development

Create 4-5 customer personas:

Health Champion: Elite engagement, long tenure, lower costs
Prevention Avoider: Low engagement, high costs, moderate tenure
New & Uncertain: New customer, mid engagement, exploring options
High-Risk Engaged: Health issues but proactive (regular checkups despite high costs)
Loyal but Declining: Long tenure but engagement dropping

For each persona:

Size of segment
Average metrics (age, cost, tenure, engagement)
CX pain points
Recommended interventions

Skills Showcased: Feature engineering, scoring models, persona development, storytelling

🗺️ PHASE 4: GEOSPATIAL & TEMPORAL ANALYSIS

4.1 Geospatial Heatmap

Insurance Penetration by City:

Customer count by location (15 cities in dataset)
Average insurance cost by city
Engagement score distribution by city
Dual coverage rate by city (competitive pressure indicator)

Visualizations:

Choropleth map (if you can map city to state/region)
Bar chart: Top 5 cities by customer count
Scatter plot: City-level cost vs. engagement
Competitive threat map: Cities with highest dual coverage rates

Business Insights:

Which markets are saturated vs. growth opportunities?
Where to focus retention efforts?
Geographic pricing variations

4.2 Temporal Analysis: Engagement Trends Over Tenure

Analysis:

Average engagement score by years_of_insurance
Track how engagement changes over customer lifecycle
Identify critical drop-off points (e.g., year 2-3?)

Visualizations:

Line chart: Avg engagement score by tenure year
Stacked area chart: Engagement tier distribution by tenure
Trend analysis: Is engagement improving or declining over time?

Hypothesis Testing:

T-test: Do loyal customers (6-8 years) have significantly higher engagement than new customers (0-2 years)?
ANOVA: Is there a significant difference in engagement across tenure groups?

Skills Showcased: Geospatial analysis, time series, hypothesis testing, statistical rigor

🧪 PHASE 5: A/B TEST SIMULATION & ROI MODELING

5.1 A/B Test Simulation Setup

Hypothesis: Offering a 10% premium discount for regular checkups will:

Increase checkup participation by 25-30%
Improve engagement scores
Reduce long-term costs (preventive care)
Increase customer retention

Simulation Approach:

Control Group: Current behavior (randomly select 50% of customers)
Treatment Group: Simulate 25% increase in regular checkup participation
- Recalculate engagement scores
- Apply 10% discount to insurance_cost

5.2 Impact Analysis

Metrics to calculate:

Customer Engagement:

% increase in average engagement score
Number of additional customers moving to "High/Elite" engagement tiers

Financial Impact:

Cost: Total discount given (10% × premium × treatment group size)
Benefit:
- Retained customers (estimate 5-10% churn reduction in treatment group)
- Potential claims cost reduction (proxy: lower insurance_cost for engaged customers)
- Lifetime value increase

ROI Formula:

ROI = (Benefits - Costs) / Costs × 100%

Benefits = (Churn Reduction Value) + (Claims Cost Savings)
Costs = Premium Discounts Given

5.3 Scenario Analysis

Create 3 scenarios:

Conservative: 15% checkup increase, 3% churn reduction
Expected: 25% checkup increase, 5% churn reduction
Optimistic: 35% checkup increase, 8% churn reduction

Deliverables:

Simulation results table (control vs. treatment comparison)
ROI calculation by scenario
Break-even analysis: At what participation rate does the program become profitable?
Sensitivity analysis: How does ROI change with different assumptions?

Visualizations:

Before/after engagement distribution
Waterfall chart: ROI breakdown
Scenario comparison bar chart

Skills Showcased: A/B testing, simulation, financial modeling, ROI analysis, business case development

📊 PHASE 6: DASHBOARD & STRATEGIC RECOMMENDATIONS

6.1 Interactive Dashboard (Choose: Tableau, Power BI, or Python Plotly/Streamlit)

Dashboard Structure:

Page 1: Executive Summary

KPI cards: Total customers, avg CLV, churn risk %, engagement score avg
Customer distribution by tenure cohort
Top 3 insights (callout boxes)

Page 2: Retention & Segmentation

Tenure cohort comparison table
Churn risk distribution
CLV by segment
Filter by: Location, Occupation, Gender

Page 3: Health Engagement

Engagement score distribution
Engagement vs. Insurance Cost scatter
Persona breakdown (pie/bar chart)
Top/bottom 10 customers by engagement

Page 4: Geographic Insights

Map: Insurance penetration by city
City-level metrics table
Competitive pressure indicators

Page 5: Wellness Program ROI

A/B test results
Scenario comparison
ROI projections
Implementation roadmap

6.2 Strategic Recommendations Document

Structure:

Executive Summary (1 page)
- Business problem
- Key findings
- Recommended actions
- Expected impact
Customer Retention Strategy
- At-risk segment size and characteristics
- Retention tactics by cohort
- Predictive model insights
Health Engagement Program
- Engagement scoring methodology
- Persona-based communication strategies
- Wellness incentive design
Geographic Expansion & Competitive Defense
- High-growth markets
- Competitive threat zones
- Market-specific tactics
Wellness Discount Program
- Program design
- Financial projections (ROI)
- Implementation plan
- Success metrics
Next Steps & Roadmap
- Quick wins (0-3 months)
- Medium-term initiatives (3-6 months)
- Long-term strategy (6-12 months)

Skills Showcased: Data visualization, dashboard design, business communication, strategic thinking

🛠️ TECHNICAL STACK RECOMMENDATIONS

Tools & Libraries:

Data Analysis: Python (pandas, numpy)
Visualization: matplotlib, seaborn, plotly
Machine Learning: scikit-learn
Dashboard:
- Option A: Tableau Public (most common in CX roles)
- Option B: Power BI
- Option C: Streamlit/Plotly Dash (shows coding skills)
Version Control: Git/GitHub
Documentation: Jupyter Notebook + Markdown

Project Structure on GitHub:

insurance-cx-analysis/
│
├── README.md                          # Project overview & key findings
├── requirements.txt                   # Python dependencies
│
├── data/
│   ├── raw/                          # Original Kaggle dataset
│   ├── processed/                    # Cleaned data
│   └── data_dictionary.md
│
├── notebooks/
│   ├── 01_data_cleaning.ipynb
│   ├── 02_eda.ipynb
│   ├── 03_segmentation_retention.ipynb
│   ├── 04_engagement_scoring.ipynb
│   ├── 05_geospatial_temporal.ipynb
│   ├── 06_ab_test_simulation.ipynb
│   └── 07_final_analysis.ipynb
│
├── src/
│   ├── data_cleaning.py              # Reusable functions
│   ├── feature_engineering.py
│   ├── modeling.py
│   └── visualization.py
│
├── reports/
│   ├── figures/                      # All visualizations
│   ├── strategic_recommendations.pdf
│   └── technical_report.pdf
│
└── dashboard/
    └── tableau_workbook.twbx         # Or Streamlit app files

⏱️ ESTIMATED TIMELINE

Phase 1: 4-6 hours
Phase 2: 8-10 hours
Phase 3: 6-8 hours
Phase 4: 4-6 hours
Phase 5: 6-8 hours
Phase 6: 8-10 hours

Total: 36-48 hours of focused work

📝 PORTFOLIO PRESENTATION TIPS

README.md Structure:

# Customer Lifetime Value Optimization Through Proactive Health Engagement

## Business Context
[1-2 paragraphs explaining the insurance industry challenge]

## Project Overview
- **Objective**: [Clear goal]
- **Dataset**: 25,000 insurance policyholders, 24 features
- **Techniques**: Segmentation, predictive modeling, A/B testing, ROI analysis

## Key Findings
1. [Finding 1 with metric]
2. [Finding 2 with metric]
3. [Finding 3 with metric]

## Recommendations
[3-5 actionable business recommendations]

## Project Structure
[Describe notebook flow]

## Tools & Technologies
Python | Pandas | Scikit-learn | Tableau | Git

## Dashboard
[Link to live dashboard or screenshots]

## How to Run
[Setup instructions]

DataCamp Workspace:

Create a clean, narrative-driven notebook
Use markdown extensively to tell the story
Include business context before each analysis section
Add "So What?" after every finding
End with a summary of recommendations

Customer Lifetime Value Optimization Through Proactive Health Engagement

.mfe-app-workspace-kj242g{position:absolute;top:-8px;}.mfe-app-workspace-11ezf91{display:inline-block;}.mfe-app-workspace-11ezf91:hover .Anchor__copyLink{visibility:visible;}Customer Lifetime Value Optimization Through Proactive Health Engagement

CX Data Analyst Portfolio Project - Insurance Industry

📋 PROJECT OVERVIEW

🗺️ PROJECT STRUCTURE

Phase 1: Data Preparation & Exploration

Phase 2: Customer Segmentation & Retention Analysis

Phase 3: Health Engagement Scoring System

Phase 4: Geospatial & Temporal Analysis

Phase 5: A/B Test Simulation & ROI Modeling

Phase 6: Dashboard & Recommendations

📊 PHASE 1: DATA PREPARATION & EXPLORATION

1.1 Data Cleaning

1.2 Exploratory Data Analysis (EDA)

👥 PHASE 2: CUSTOMER SEGMENTATION & RETENTION ANALYSIS

2.1 Customer Tenure Segmentation

2.2 Churn Risk Scoring

2.3 Customer Lifetime Value (CLV) Estimation

💪 PHASE 3: HEALTH ENGAGEMENT SCORING SYSTEM

3.1 Create Health Engagement Score (0-100)

3.2 Engagement Analysis

3.3 Persona Development

🗺️ PHASE 4: GEOSPATIAL & TEMPORAL ANALYSIS

4.1 Geospatial Heatmap

4.2 Temporal Analysis: Engagement Trends Over Tenure

🧪 PHASE 5: A/B TEST SIMULATION & ROI MODELING

5.1 A/B Test Simulation Setup

5.2 Impact Analysis

5.3 Scenario Analysis

📊 PHASE 6: DASHBOARD & STRATEGIC RECOMMENDATIONS

6.1 Interactive Dashboard (Choose: Tableau, Power BI, or Python Plotly/Streamlit)

6.2 Strategic Recommendations Document

🛠️ TECHNICAL STACK RECOMMENDATIONS

Tools & Libraries:

Project Structure on GitHub:

⏱️ ESTIMATED TIMELINE

📝 PORTFOLIO PRESENTATION TIPS

README.md Structure:

DataCamp Workspace:

Customer Lifetime Value Optimization Through Proactive Health Engagement