An Introduction to RFM in Analytics

Copyr ight © 2012, SAS Institute Inc. All rights reser ved.
RFM QUICK START GUIDE
PAT VALENTE/ROB WILSON
PRE-SALES, SAS CANADA

RFM MODEL QUICK START CONTENTS
• This guide will provide you with the following:
• Introduction to the RFM model
• Data Requirements
• SAS project configuration considerations
• Model Description
• Workflow Overview and Build

INTRODUCTION

RFM MODEL INTRODUCTION
• WHAT IS RFM?
• Method used for analyzing customer value.
• Commonly used in database marketing and direct marketing.
• Recency - How recently did the customer purchase?
• Frequency - How often do they purchase?
• Monetary Value - How much do they spend?
This quick start guide provides an Enterprise Guide project that categorizes
customers into a predefined number of ‘segments’ based on the score from
the RFM analysis.

DATA REQUIREMENTS

RFM MODEL DATA REQUIREMENTS
• The input data for this project is transactional data. The project aggregates the
transaction data into customer data. In transactional data, each record represents
one transaction.
• Minimum data requirements:
• There are three variables required as listed below:
• Transaction Date: the variable that specifies the date of the transaction. This information
determines the most recent transaction date. In this step, the data is sorted in descending
transaction date order and the latest transaction date is selected in a query.
• Amount of Transaction: the variable that specifies the amount of the transaction. This data
is used to compute the total amount of the transactions.
• Customer Identifier: the variable that contains a number or string that can uniquely identify
a customer.

PROJECT CONFIGURATION

RFM MODEL PROJECT CONFIGURATION
• When requesting the project please specify whether you are running Enterprise Guide
locally on a desktop or via a server with a Metadata layer.
• If you have the incorrect version, you can either request the correct version or simply
using the Migration Wizard in SAS to convert the existing project.
• Find the MigrationWizard executable file in the EG folder of SASHome and double-click
to run. Ensure that all other programs are closed or the wizard may not work properly.

• Click “Next” after you see the first
step.
• In step 2, click “Modify” to change the
active connection you have to SAS.

• In step 3, find the project you need to
convert, check it and send it to the
Selected files box.
• Click “Next” in step 4.

• Click “Done” when
parsing is completed.
• In step 5 click in the New Mapping drop down column to
select the correct local references. Click “Next”.

• Click “Done” when the migration process is
complete. The project will be ready to be
opened in Enterprise Guide.
• Click “Finish” in step 6 to
complete the migration.

MODEL DESCRIPTION

RFM MODEL MODEL DESCRIPTION
• Recency, frequency, and monetary scores are determined as follows:
• The recency score is determined by sorting the values of the most recent
transaction date in ascending order and then grouping these values into ‘bins’.
The bin with the oldest dates is assigned the lowest recency score, and so on. The
number of bins and recency score for each bin will need to be determined by the
customer.
• The frequency score is determined by sorting the values of the number of
transactions in ascending order and also grouping these values into ‘bins’. The bin
with the smallest number of transactions is assigned the lowest frequency score,
and so on. The number of bins and frequency score for each bin will need to be
determined by the customer.

RFM MODEL MODEL DESCRIPTION
• The monetary score is determined by sorting the values of the total amount of the
transactions in ascending order and grouping these values into ‘bins’. The bin with
the smallest amount is assigned the lowest monetary score, and so on. The
number of bins and monetary score for each bin will need to be determined by the
customer.
Categorizing the data into bins
• You will need to specify the number of bins for recency, frequency, and monetary
values individually.
• RFM score = Recency score + Frequency score + Monetary score.
• The least favorable customer segment has the lowest RFM score. The most
favorable customer segment has the highest RFM score.

PROJECT WORKFLOW OVERVIEW AND BUILD

RFM MODEL WORKFLOW OVERVIEW
• Below is a screen shot of the Enterprise Guide project for calculating the RFM
score and creating the segments based on the RFM score. You will see the
initial data set on the far left of the workflow.

1. The first step is to calculate
the most recent transaction
date, using the “MAX”
statistic selection in the query
builder. Right click on the
Recent transactions icon and
select “Modify” and
“Computed Columns”.
2. You can see the new column
named “Max_Trans_Date”
3. You can see the details of the
calculation which is a Max of
the transaction date.

1. Next you will calculate the
total transactions using the
‘COUNT’ statistic and the
customer IDs. Right click on
the Total transactions icon
and select “Modify” and
“Computed Columns”.
named “Total_Transactions”.
calculation which is a
frequency count for each of
the unique customer IDs.

1. The third step is to calculate
the total transaction amount
using the ‘SUM’ statistic and
the transaction amounts.
Right click on the Total
transactions icon and select
“Modify” and “Computed
Columns”.
named “SUM
_of_Transaction_Amount”.
calculation which is a sum of
the transaction amounts.

1. The next step is to add
the newly calculated
columns to the data by
joining the tables.
2. Select Tasks  Data 
Query Builder.
3. You will see the new
columns in the Select
Data tab.
4. Click on Join Tables to
see how the tables were
joined using Customer ID
as the common key.

1. You can now create scores for
recency, frequency and
monetary bins individually. Right-
click on the Create RFM Scores
icon and select “Modify”.
2. You will see the new columns in
the Select Data tab.
3. Click on “Computed Columns”,
select one of the scores and
click “Edit”.

1. You can then see how each
score has been created using
a “case” statement and
creating “bins” based on the
values of the variable.
2. This example shows a sample
calculation for frequency with
3 ‘bins’ and a score for each
bin (1, 2 or 3).
3. You can click “Next” to
complete the build.
4. You can repeat these steps
for the other scores.

1. The RFM scores are now
added together to create
the Total RFM score. Right-
click on the Total RFM
Score icon and select
“Modify”.
2. You will see the
Total_RFM_Score columns
in the Select Data tab.
3. Click on “Computed
Columns”, select the Total
RFM Score and click “Edit”.

1. You can then see how the
Total RFM Score has been
created simply by adding the
three calculated values
together.
complete the build.
3. The Total RFM Score is then
added to the data for each
Customer ID.

1. Once the number of segments
and the split values of the
segments has been determined,
the segment variable is created.
2. Right-click on the “Segment”
icon and select “Modify”.
3. You will see the “Segment”
column in the Select Data tab.
4. Click on “Computed Columns”,
select the “Segment” variable
and click “Edit”.

1. You can see how a “case”
statement is used to create
each segment based on the
values of the Total RFM Score
and how many segments are
desired.
complete the build.
3. The Segment variable is then
added to the data set.

1. In the final step, a pie chart illustrating the
frequency counts of the customers in each
segment is created.
2. Right-click on the Pie Chart icon and select
“Modify” to examine the attributes that can be
customized in creating the chart.
Note:
• This method is descriptive only, and does not
provide a mechanism to forecast behavior as a
predictive model might.
• When used to target customers, it assumes that
customers are likely to continue behaving in the
same manner. That is, it does not take into account
the impact of life stage or life cycle transitions on
likelihood of response.

An Introduction to RFM in Analytics

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Viewers also liked

Viewers also liked (18)

Similar to An Introduction to RFM in Analytics

Similar to An Introduction to RFM in Analytics (20)

More from SAS Canada

More from SAS Canada (9)

Recently uploaded

Recently uploaded (20)

An Introduction to RFM in Analytics