E-Commerce Customer Churn Analysis is an end-to-end data analytics project aimed at identifying and understanding customer churn drivers in a major online retail store. By applying statistical analysis, customer segmentation, and visual dashboards, this project equips marketing and product teams with actionable recommendations to improve customer retention.
Project Workflow
The analysis is structured around the standard data lifecycle: extraction, cleaning, exploratory analysis, modeling, and visualization.
Data Extraction & Wrangling (SQL & PostgreSQL)
- Queried transaction, user demographic, and web log tables.
- Handled millions of rows of user activity logs to extract features like average order value (AOV), login frequency, and customer lifetime value (CLV).
- Dealt with missing values, duplicate entries, and inconsistent data formats via complex SQL queries and temporary tables.
- Structured cohorts based on the user sign-up date to analyze retention rates over time.
Exploratory Data Analysis (Python)
- Used Pandas and NumPy for exploratory data analysis (EDA) and cohort analysis.
- Calculated key metrics such as:
- Monthly Churn Rate (percentage of active users who did not make a purchase in 30 days).
- Average Customer Lifespan.
- Correlation between discount usage and long-term customer retention.
- Created visualizations using Seaborn and Matplotlib to identify distributions and detect outlier behaviors.
Customer Segmentation & Modeling
- Performed RFM (Recency, Frequency, Monetary) Segmentation to cluster customers into groups such as “Champions”, “At Risk”, and “Lost”.
- Applied a logistic regression model to predict the probability of churn for individual active users, providing risk scores to targeted marketing campaigns.
Dashboard & Visualization (Tableau)
- Built an interactive Tableau Dashboard presenting:
- High-level KPIs (Active Users, Monthly Churn Rate, Average Revenue per User).
- Dynamic cohort analysis grids.
- Interactive filters by region, age group, and acquisition channel.
- High-risk customer watchlists.
Tech Stack & Tools
- Data Processing: Python (Pandas, NumPy, Scikit-Learn)
- Database: PostgreSQL / SQL
- Visualization: Tableau Desktop / Cloud, Seaborn, Matplotlib
- Notebook Environment: Jupyter Notebook
This analytical framework helps businesses shift from reactive customer recovery to proactive churn prevention, highlighting the business value of turning raw historical records into clean datasets and actionable business strategies.