RFM Customer Segmentation Analysis

A Python-based customer segmentation project using RFM (Recency, Frequency, Monetary) modeling on real retail transactional data. The goal was to identify distinct customer groups and translate them into actionable marketing strategies.

---

Project Overview

RFM analysis is a proven marketing technique that scores customers based on three dimensions:

• Recency : How recently they made a purchase
• Frequency : How often they buy
• Monetary : How much they spend in total

By combining these scores, customers are grouped into meaningful segments , each with a tailored retention or growth strategy.

---

Key Numbers

Metric	Value
Transactions Analyzed	800,000+
Customers Segmented	5,860
Segments Discovered	7
Top 5 Customer Revenue	$1.9M+
High-Value Customers at Risk	615

---

Customer Segments

Segment	Customers	Priority	Strategy
Champions	720	🟢 High	VIP rewards, early access, premium offers
Loyal Customers	506	🟡 Medium	Tiered loyalty programs, exclusive discounts
Potential Loyalists	1,430	🟡 Medium	Personalized recommendations, gentle nudges
New Customers	123	🟡 Medium	Welcome series, onboarding journey
At Risk Customers	1,470	🔴 High	Win-back campaigns, 15–20% discounts
Can't Lose Them	615	🔴 Critical	Personal outreach, strong retention offers
Lost Customers	842	⚫ Low	Surveys, last-chance comeback deals

---

Visualizations

Page 1 : Customer Overview

• Customer segment distribution (horizontal bar chart)
• Average RFM scores across all customers
• Key metrics summary
• Top 5 customers by total revenue

Page 2 : RFM Deep Dive

• RFM heatmap: average spending by Recency × Frequency score
• Recency distribution histogram
• Frequency distribution histogram
• Monetary distribution histogram

---

Technical Implementation

RFM Score Calculation

snapshot_date = df_clean['InvoiceDate'].max() + pd.Timedelta(days=1)

rfm = df_clean.groupby('Customer ID').agg({
    'InvoiceDate': lambda x: (snapshot_date - x.max()).days,  # Recency
    'Invoice': 'nunique',                                      # Frequency
    'TotalSales': 'sum'                                        # Monetary
})
rfm.columns = ['Recency', 'Frequency', 'Monetary']

Scoring Logic

• Customers scored 1–4 on each dimension using quartile-based binning
• Scores combined into a 3-digit RFM code (e.g. 444 = Champion)
• Segments assigned using rule-based logic on R, F, M score combinations

---

Tools & Libraries

Tool	Purpose
Python 3.8+	Core language
pandas	Data cleaning and transformation
numpy	Numerical operations
matplotlib	Chart rendering
seaborn	Statistical visualizations

---

Dataset

A retail transactions dataset (Online_Retail.csv) containing 800,000+ records. The dataset is not included in this repository.

---

Output Files

├── RFM_Analysis.py                    # Main analysis script
├── RFM Analysis Visuals (Page 1).png # Overview dashboard
├── RFM Analysis Visuals (Page 2).png # Distribution dashboard
└── README.md                         # Project documentation

---

Business Impact

This segmentation enables marketing teams to:

• Retain high-value customers before they churn
• Re-engage at-risk segments with targeted campaigns
• Allocate marketing budget more precisely
• Manage the full customer lifecycle with data-backed decisions

---