🛒 Online Retail Customer Analysis

An end-to-end customer retention analysis project using online retail transaction data, from data cleaning and RFM segmentation to actionable insights and dashboard visualization.

👀 Project Overview

This project analyzes online retail transaction data to identify high-priority customer segments and guide retention-focused strategies. The analysis uses RFM-based segmentation to understand customer value concentration, revenue risk, and engagement patterns.

❓ Business Question

To keep sales growing, online retailers need to truly understand how their customers shop. But treating every customer exactly the same is a recipe for wasted marketing efforts and lower retention rates. That leaves us with one question:

“How can transaction data help identify which customers matter most and how they should be retained?”

📂 Project Structure

├── project_da_customer_analysis.py/        → Python script for data cleaning & RFM Analysis
├── Dashboard.pbix/    → Power BI Dashboard
├── Slide Presentation.pdf/   → Project summary and key insights presentation
└── README.md

📑 Data Source

- Public Online Retail Dataset from Kaggle

Check the full dataset here

- Key Variables

• InvoiceNo : transaction identifier
• CustomerID : unique customer identifier
• InvoiceDate : transaction timestamp
• Quantity & UnitPrice : purchase volume and price
• Description : product name
• Country : customer location

📈 Analysis Process

1. Data cleaning and preprocessing (Python)

Cleaned the data from:

• Invalid transaction -> Dropping 2,517 non-positive value in Quantity and UnitPrice
• Missing customer ID -> 132,220 Transactions with missing CustomerID were removed
• Date formatting -> The InvoiceDate column was converted from string format to datetime format

Feature engineering:

• Created total transaction value by multiplying Quantity with UnitPrice
• Created reference date that defined as as one day after the latest transaction date

2. RFM metric calculation at customer level (Python)

Customer behavior was summarized using transaction-based metrics (RFM) to differentiate customer value and engagement

• Recency –> how recently a customer purchased
• Frequency –> how often a customer purchased
• Monetary –> how much a customer spent

3. RFM-based customer segmentation

Customers were grouped into distinct, mutually exclusive segments:

• Loyal -> High recency, frequency, and monetary value
• Potential Loyalist -> Active customers with growth potential
• New Customers -> Recently acquired customers
• At Risk High Value -> High spenders with declining activity
• Lost / Low Engagement -> Inactive or low-value customers

4. Dashboard (Power BI)

An interactive Power BI Dashboard, provide a high-level view of customer value, risk, and revenue drivers to support retention-focused decisions

🔑 Key Insight

A. Revenue Dependency Is Narrow

Even though data shows a lot of lost customers (>50%) of, loyal customer is the no.1 revenue contributor by far amount

Insight: It means that increasing customer volume alone is unlikely to significantly improve revenue without shifting customers into higher-value segments

B. High-Value At Risk Customers Punch Above Their Weight

Although representing only 7.8% of customers, this segment contributes 9.5% of total revenue

Insight: This justifies focused and customized retention strategies rather than broad campaigns

🎯 Business Recommendation

• Prioritize retention efforts on loyal and potential loyal customers to stabilize revenue (e.g. early access to new product)
• Apply targeted and time-sensitive retention for high-value customers showing declining activity to make their comeback (e.g. personalized reactivation email based on the latest purchased)
• Focus retention campaigns on top-performing products and core markets, such as United Kingdom

🧰 Tools

• Pyhton (Pandas)
• Power BI

Contact

• Name: Muhammad Kamil Dipinto
• Email: dipintom3@gmail.com
• Linkedin: linkedin.com/in/muhammadkamildipinto