Versions Compared

Version Old Version 3 New Version 4
Changes made by

Ernie Veniegas

Ernie Veniegas

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.


Section


Column
width60%


R

ecency, Frequency, Monetary, or RFM analysis is a technique used to determine which customers are likely to respond to a new offer by examining the following:RFM Analysis is a way of grouping & organizing all your customers into segment-based clusters according to customer behaviors: 

  • Recency - How recently did the customer purchase?
  • Frequency - How often do they purchase?
  • Monetary Value - How much do they spend?
For each customer visible in the CRM grid, RFM analysis calculates and assigns a score from 1 to 5 for each RFM measure.
  • what's the monetary value of their purchases?

The goal of RFM analysis is to understand who is more likely to buy from you again and when they will do it. RFM models provide a bird-eye view of your customer base and better understand customer behaviors and purchasing habits. 

RFM uses purchase history. Advanced Search in PSCUM.CRM calculates RFM scores. RFM values range from 1 to 5.  A score of 5 represents the top 20% of the population, while a score of 1 represents the bottom 20%, scores 2, 3, and 4 the 20% ranges in between.  A score of one (1) means the least recent, least frequent, or lowest spendmonetary value, while a score of five (5) means the most recent, the most frequent, or the highest monetary value. The system concatenates each measure into a three-digit score ranging three measures together range from 111 to 555, a.k.a., the RFM score. The highest score is 555, meaning the customer purchased most recently, most frequently, and at the highest monetary value. The lowest score is 111, meaning the customer has purchased the least recently, least frequently, and at the lowest monetary value. 

The system derives RFM scores based on a selected customer population and how an individual's RFM score plots on a relative scale of 100%. 

  • A score of 5 represents a rank in the top 20% of the population, that is, scores in the 80-100th percentile.
  • A score of 4 represents a rank in the next 20% of the population, that is, scores in the 60-79th percentile. 
  • A score of 3 represents a rank in the next 20% of the population, that is, scores in the 40-59th percentile. 
  • A score of 2 represents a rank in the next 20% of the population, that is, scores in the 20-39th percentile. 
  • A score of 1 represents a rank in the bottom 20% of the population, that is, scores in the 0-19th percentile.

The analytical technique entails establishing a benchmark RFM score and saving it in for future reference.  In the CRM record grid, the system identifies that value as the "Last RFM."  You would use the scores to develop segments and outreach strategies. After some timea while, you would regenerate RFM scores and compare the results to their Last RFM score to identify customers whose scores improved or worsened, increased or decreased.  When regenerating RFM scores, and develop strategies accordingly.you must use the same start-stop date range and selection criteria to compare apples to apples.  

Table of Contents

Table of Contents
excludeTable of Contents|Introduction



...

Press Enter then Continue to review and confirm before and after results.



PSCUM.CRM - RFM Program Logic - Building Buckets

Method:  Inspect the items being measured. Find the top and bottom values. Take the difference and divide by 5 giving bucket ranges. Assign items to buckets. Done (Tim Smith, 1981- )

Recency Example (the R-Score of RFM)

  • Scenario:  Order Selection in 24-month date range, from Jan 1, 2020, through Dec 31, 2021 
  • 24 months / 5 = 4.8 months (rounded to 5)
  • Yields five 5-month ranges
Last Purchase in Month Range...R-Score

Last Purchase in Date Range (assuming SOLD or SHIP date?)

21-251January 2020 - April 2020
16-20

2

May 2020 - September 2020
11-153October 2020 - February 2021
6-104March 2021 - July 2021
1-55August 2021 - December 2021

Questions for Programming: 

  1. What date does the system use: Batch, Sold, Shipped?  Assumes program uses Batch Date since individual line items can be SOLD and SHIPPED on different dates.  
  2. When rounding, where does the program put the short range?  In the example above, priority was given to R-Score 5, Last Purchase within last 1-5 months. R-Score 1 represents last purchase 21-25 months ago.  Since our date range only encompasses 24 months, the program will not encounter anyone whose last purchase was 25 months ago. 
  3. Is the system selecting negative amount transactions? That would negatively affect a RECENCY score.

Frequency Example (the F-Score of RFM)

  • Scenario:  N customers, each having completed a different number of purchases within whatever date range. The value of n does not matter. 
  • Tally purchasers per customer
  • Highest individual count is 15
  • Lowest individual count is 1 
  • 15+1 = 16 /5 = 3.2 (rounded to 3)
  • Yield five 5 ranges in increments of 3 
Frequency ranges...F-Score

Individual tally/count of purchases ...

1-3 1Same as frequency range
4-6

2

"
7-93"
10-124"
13+5"

Monetary Example

  • Scenario:  N customers, each having completed a different number of purchases within whatever date range at whatever monetary value. The value of n does not matter. 
  • Tally net sales amount (question) per order per customer 
  • Highest individual net sale amount is $2,500
  • Lowest individual net sale amount is $20 (assuming CREDIT and negative amount transactions are not being selected 
  • $2,500+$20 = $2,520/5 = $504
  • Yield five 5 ranges in increments of $504
Monetary ranges...RFM Score

Individual sum of monetary value ...

$1 - 5041Same as monetary range
$505 - 1,009

2

"
$1,010 - 2,5233"
$2,524 - 3,0284"
$3,029+5"


Questions for Programming:  

  1. Is the program selecting $0 dollar transactions?
  2. Is the program selecting transactions with negative amounts, including CREDIT transactions?
  3. For monetary scores to be accurate, the system should include all transactions.