fbpx Customer Segmentation Using RFM Analysis | Leapfrog Technology

Customer Segmentation Using RFM Analysis

We live in a dynamic world, and people are seemingly looking for growth, and many depend on establishing and running businesses for their survival. However, all businesses have one important aspect – the customers. Whether you are running an online business or a business with physical locations, you should always seek ways to adapt to bring in new customers and retain the current customers by catering to them.

Business owners should well understand different types of customers and their needs. This is where customer segmentation comes in.  

What is Customer Segmentation?

Customer segmentation is the process of categorizing your customers based on their qualities. We can segment customers in the following ways:

  • Geographic customer segmentation: Segmenting customers by location, i.e., grouping by country, state, city, or zip code.
  • Demographic segmentation: It is related to grouping through the statistical variables i.e, age, gender, income etc.
  • Psychological segmentation: Customers have various personality traits, attitudes, or beliefs. If grouping is done based on these factors, it happens to be psychological segmentation.
  • Behavioral customer segmentation: It is based on customers’ past observed behaviors that can be used to predict future actions. i.e. grouping through purchasing habits, spending habits, interactions and so on.

In this article, we will be focusing on behavioral customer segmentation.

Why do we need customer segmentation?

Before heading into our major topic, let us know the importance of customer segmentation for businesses.

Every business needs to increase its customers’ lifetime value (CLV). CLV is the total time a customer is consuming goods from a business. Consider how dogs have a lifetime value of 10-13 years; similarly in a business scenario, customers also have a particular lifetime value. The only difference is that a dog’s lifetime value is a fixed attribute, but it can be increased for customers. The more a customer stays, the more revenue a business can generate. Customer segmentation helps us divide the customer base into various segments that can be useful in making crucial business decisions. 

What are the ways of preparing segments?

There are various methods of analysis to prepare segments of customers. Among which K means clustering, and RFM analysis is the popular one. 

In this article, we will be focusing on the RFM analysis. 

What is RFM Analysis?

RFM analysis is a technique for segmenting customer behavior based on data. It may sound obscure initially, but it is a straightforward approach. You collect data of the customers, analyze them and bang, you sort out their buying behavior. 

How do we carry out RFM Analysis?

There are some prerequisites for carrying out RFM analysis. I think you must have gotten the idea, we need DATA. 

For RFM Analysis, we need three primary information:

  1. Recency: When was the last time a customer visited our business?
  2. Frequency: How often do customers visit our business?
  3. Monetary: How much money has a customer spent on our business?

Let us clear this concept by looking at the example below where I demonstrate the data that must be kept by any business to carry on a RFM analysis. Then after collecting the data mentioned above, we will be grading the customers by ranking them based on each RFM attribute separately.

I normally use RFM values to rank the customers from 1-4 (But these scores can differ depending on the business needs.)

I’ve arranged customers by recency in the table above, with the most recent purchasers at the top. Customers are given scores ranging from 1-4, so the top 25% of customers who visited the business lately get a recency score of 4, the next 25% get a score of 3, and so on. 

Similarly, we can sort customers by frequency from most to least frequent, providing a frequency value of 4 to the top 25%, and so on. The top 25% of clients (large spenders) will be given a score of 4 for the monetary element, while the bottom 25% will be given a score of 1.

Programmatically, these scores are calculated using a basic statistical technique i.e. quartiles. (p.s. Now you know where maths comes in handy.)

Remember how in school days we were separated into different groups (Toppers, Average and Weaklings) based on our grades? 

The school had its way of targeting the students for various learning strategies. We do the same now, but this time we will be grouping the customers.

What do these target groups look like?

So let’s talk about what target groups can be formed using these scores. These target groups vary according to the business needs, but I will discuss the most common and popular target groups.

Using Recency-Frequency-Monetary (RFM)

When using all three components of the RFM analysis, we can form an enormous number of target groups. But according to the business needs, we must be able to sort out the needed groups to target. This kind of target grouping is done when a business has similar products. For example, a business that sells laptops, a business that sells clothes etc. Although, for businesses that sell random items or highly ranging prices, this type of targeting might not be efficient.

  • Best Customers(R=4, F=4, M=4): These customers visit the business very frequently, very recently and have spent a lot on our business.
  • Best Customers, Churned(R=1, F=3 or 4, M=3 or 4): These are loyal clients who haven’t bought anything in a long time. 
  • Best Customers, At Risk(R=2 or 3, F=3 or 4, M=3 or 4): These are loyal clients who are on the verge of being churned. These customers need special attention. 
  • New Customers, Low Spenders(R=4, F=1 or 2, M=1 or 2): These are new customers to the business and have spent an ok-ish amount on their initial transaction.
  • New Customers, High Spenders(R=4, F=1 or 2, M=3 or 4): These new consumers spend a lot of money on their initial transactions. This is an important target group.
  • Loyal Customers, Low Spenders(R=3 or 4, F=3 or 4, M=1 or 2): These are the loyal customers, but they haven’t spent much on their transactions.
  • Idle Customers(R=1 or 2 or 3, F=1 or 2, M=1 or 2): These customers are hibernating customers and random visitors. 
  • Customers at risk(R=2 or 3, F= 1 or 2 or 3, M=2 or 3 or 4): These are the customers who are on the verge of going off. These aren’t a part of the best customers but are the customers who are at risk. 
  • Churned Customers(R=1, F=1 or 2 or 3, M=2 or 3 or 4): These are your losses. These groups of customers are the unsatisfied customers, according to this analysis. They have stopped visiting your business and didn’t spend much while involved.

Fig: Insights derived using RFM

2. Using Recency-Frequency (RF)

In this method, only the recency and frequency values are considered while forming a target group while the monetary value is ignored. This type of customer grouping is effective for a business where the business demands more flow of customers rather than focusing on how much a customer spends. 

Consider a scenario where person A buys chocolates regularly for a whole year spending Rs. 10 every day (Rs. 3650/year), and person B buys a shoe costing Rs.4000 once. When considering monetary value, person B has spent more than person A in a year. This brings unfair results when considering the loyalty of a customer. So businesses that sell items ranging from low to high prices, such as supermarkets, can consider using just the RF values while targeting groups. 

Below is the list of groups that can be formed using RF values.

  • Champions(R=4, F=4): These customers visit the business frequently and very recently. These are loyal, and we must always be able to keep them.
  • Loyal Customers(R=3, F=4): These customers have been visiting the business for a very long time. 
  • Hibernating (R= 1 or 2 or 3, F=1 or 2)  These customers can be considered random shoppers or mood swingers. It’s kind of hard to deal with them.
  • New Customers (R=4, F=1) : These customers have visited the business very recently. 
  • Potential Loyalists (R=3 or 4, F=2 or 3): These customers have recently had an increasing number of visits. 
  • At-Risk(R=1, F=3 or 4): These customers were once the most frequent customers, but they have stopped visiting the business now.
  • Can’t Lose them(R=2, F=3 or 4): These customers were once persistent to your business but slowly have been fading away. However, there is a possibility of retaining them.

Fig: Insights Derived using RF

3. Using RFM Added Scores

The Final Target Strategy I will talk about is the RFM added scores. You can use this strategy when your business is small, and if you have limited budgeting considering your marketing department.

What we do here is add up the RFM scores.

  • Platinium(R+F+M = 10 or 11 or 12): These are Top 25% of the customers per the RFM score. 
  • Gold(R+F+M = 9 or 8 or 7): These are the next 25% of the customers in accordance with the RFM score after platinum. 
  • Silver(R+F+M = 6 or 5 or 4): These are the next 25% of the customers following the RFM score after gold. 
  • Bronze(R+F+M = 3 or 2 or 1): These are the Bottom 25% of the customers per the RFM score. 

Some strategies that can be used on these target groups:

After successfully targeting the groups, it’s all up to the business development team to develop actions to keep their customers. However, I will be pointing out a few plans and actions that a business can implement for their business growth:

Groups Strategies Impact
For the Top Customer group– Communication is likely to make them feel respected and appreciated. 
– More in-depth analysis of their unique tastes and affinities will open up even more possibilities for more targeted content.

These customers are likely to account for a disproportionately large share of overall revenue, therefore keeping them satisfied should be a major priority.
For the customers who have been showing signs of non-engagement– Win them back via renewals or newer productsDon’t lose them to competition, talk to them.
For the group of random visiting customers– Running a social media ads targeting campaign.
– Give off random discount offers
More engaging just might hold them on your side.
For the group of newly visiting customers– Having clear first-time buyer strategies in place, such as triggered welcome emails, might pay off.
– Provide them with fantastic incentives to interact with the company in the future
It is always a good idea to carefully “incubate” any new customers. The majority of customers never progress to being loyal. We should aim to make them feel respected and appreciated.
For Customers who seem to be cautious with their money– The marketers can target them for a reward with special deals if they spread the word about the company to their friends, such as through social media.If these customers are loyal and yet spend less, marketers should build ads for this demographic that make them feel appreciated and encourage them to spend more.

These were just a few groups and their respective strategies. You can always go on and build a full-fledged plan for your own business. A little bit of study and getting to know your customers will always help you grow your business through RFM analysis.

Amun Timilsina

Amun is an Associate Software Engineer, AI at Leapfrog Technology.

More in Blogs

Using FAISS Advanced Concepts to Speed Up Similarity Search in Recommender Systems  (Part II) Artificial IntelligenceInsights

Using FAISS Advanced Concepts to Speed Up Similarity Search in Recommender Systems (Part II)

Continued from the last blog. Improving on Flat Indexes Beyond the flat indexes that perform exhaustive searches, FAISS also has

Read more
Using FAISS basics to Speed up similarity search in Recommender Systems (Part I) Artificial Intelligence

Using FAISS basics to Speed up similarity search in Recommender Systems (Part I)

Measuring the similarity between vectors or arrays is a calculation that we encounter often while developing recommendation systems or other

Read more
Why is a No-Code Recommendation System worth a try? Artificial IntelligenceInsights

Why is a No-Code Recommendation System worth a try?

Lately, in the tech world, no-code systems have been getting immense attention, and for good reasons. These types of systems

Read more