What is RFM Analysis?#
RFM (Recency, Frequency, Monetary) Analysis is a customer segmentation technique that uses three key metrics:
- Recency: How recently a customer made a purchase
- Frequency: How often they purchase
- Monetary: Total amount spent
Statistical Approach to Binning in RFM#
Binning is critical in RFM to categorize customers:
- Divide each metric into quartiles (4 equal segments)
- Assign scores 1-4 for each dimension
- Create customer segments based on combined scores
Python Example:#
import pandas as pd
import numpy as np
def rfm_segmentation(df):
# Calculate RFM metrics
rfm = df.groupby('customer_id').agg({
'order_date': lambda x: (df['order_date'].max() - x.max()).days,
'order_id': 'count',
'total_amount': 'sum'
})
# Rename columns
rfm.columns = ['Recency', 'Frequency', 'Monetary']
# Create quartile-based scoring
rfm['R_rank'] = pd.qcut(rfm['Recency'], q=4, labels=[4,3,2,1])
rfm['F_rank'] = pd.qcut(rfm['Frequency'], q=4, labels=[1,2,3,4])
rfm['M_rank'] = pd.qcut(rfm['Monetary'], q=4, labels=[1,2,3,4])
return rfmR Example#
library(dplyr)
rfm_analysis <- function(data) {
rfm_result <- data %>%
group_by(customer_id) %>%
summarise(
Recency = as.numeric(max(max_date) - max(order_date)),
Frequency = n_distinct(order_id),
Monetary = sum(total_amount)
) %>%
mutate(
R_rank = ntile(Recency, 4),
F_rank = ntile(Frequency, 4),
M_rank = ntile(Monetary, 4)
)
return(rfm_result)
}Customer Segments Created#
- Champions: High scores across R, F, M
- At Risk: High historical value but haven’t purchased recently
- Hibernating: Low scores in all dimensions
- New Customers: High recency, low frequency/monetary
References#
- Specifically referenced from:
- Blattberg, R. C., & Deighton, J. (1996). Customer Equity Framework
- Kumar, V., & Reinartz, W. (2016). Customer Engagement Marketing
