RFM Customer Segmentation Analysis In Action!
Learn the principles of RFM analysis and explore a real-world application in Looker Studio.
You can check out the real-world application of RFM analysis here.
What is RFM?
RFM Segmentation Analysis is a data-driven customer segmentation technique used by businesses to understand and categorize their customers based on their past purchasing behavior.
RFM stands for Recency, Frequency, and Monetary value.
This method aims to segment customers based on three essential components of their past purchasing behavior:
• Recency (R): The time since their last purchase.
• Frequency (F): The total number of times they’ve purchased in the past.
• Monetary (M): The total amount they’ve spent.
These metrics serve as reliable indicators to efficiently identify and target distinct customer segments for marketing campaigns, aiming to retain valuable customers, engage existing ones, and re-engage those who have left.
An example
Imagine, you are the owner of a restaurant and you want to send emails to get some customers back.
You grab your notebook and figure out the 6-month RFM score so you can send different emails to different customers based on their RFM score.
What metrics will you need to calculate for each customer?
• Recency (R):
Days since the customer’s last visit in the past 6 months.
• Frequency (F):
Number of times a customer dined at the restaurant in the past 6 months.
• Monetary (M):
Total amount spent by a customer at the restaurant over the last 6 months.
When was RFM first used?
RFM segmentation was first used in the 1930s and 1940s when direct marketers sent printed catalogs through direct mail. For each customer, they manually maintained a 3×5 index card, ranking them based on their last purchase date, purchase frequency, and lifetime spend.
The core objective of this ranking system was to identify high RFM / value customers, who were most likely to respond to new catalogs, thereby optimizing costs by not sending catalogs to those less likely to make a purchase.
Nowadays, although email marketing is much cheaper than printing catalogs and sending them via direct mail, ineffective targeting can lead to annoyed customer, brand harm, and lack of interest.
Are there any variations of the original RFM model?
While the core principles of RFM remain unchanged, advancements in tools and data collection methods have enabled us to gather richer datasets about customers optimizing the RFM methodology.
RFD (Recency, Frequency, Duration): This variation replaces the monetary value dimension with the duration of time that a customer spends using a product or service. This is useful for businesses that are monetized through sponsored content, viewership, or other non-monetary means.
RFE (Recency, Frequency, Engagement): This variation replaces the monetary value dimension with a measure of customer engagement. Engagement can be measured in a variety of ways, such as the number of pages viewed, the number of videos watched, or the amount of time spent on a website or app.
RFMT (RFM + Tenure): This is a variation where “T” stands for Tenure. Tenure is the age of the customer or how long it’s been since a customer’s first purchase. This helps businesses identify loyal, long-term customers who may not purchase frequently but have been with the business for a long time.
RFMt (Recency, Frequency, Monetary Value, Time) or Historical RFM: This variation adds the time dimension to the RFM analysis.
This is useful for understanding how customer behavior has changed over time and identifying trends. For example, a business might use RFMt to identify the behavior of their most valuable customers over time and act accordingly.
So why is RFM so important?
Identify and keep valuable users:
• Which customers are most loyal?
• Who are the highest spending customers?
• Who are the most valuable customers?
• Which customers should be prioritized in a customer loyalty or rewards program?
• How effective are customer loyalty programs?
Retain users:
• Which customers used to buy frequently but have reduced their engagement recently?
Get back users who have left:
• Who hasn’t made a purchase in a long time and might need re-engagement strategies?
Maximize buying potential:
• Are there customers who make large but infrequent purchases, indicating potential to buy more often?
What does RFM score say about the customer?
An RFM score is a numerical value that represents our customer’s Recency (R), Frequency (F), and Monetary value (M).
By calculating RFM scores, we evaluate each customer across three components: Recency (R), Frequency (F), and Monetary value (M). For each of these components, we typically assign a numerical value on a scale from 1 to 5, where 1 is the lowest and 5 is the highest. Therefore, each customer receives three separate scores — one for Recency, one for Frequency, and one for Monetary value — reflecting their respective performance in each area. By analyzing these scores, we can segment our customers and understand their behavior.
The combination of individual Recency, Frequency, and Monetary value scores forms the RFM score, indicating the overall value or engagement level of the customer.
Note that we will focus on the rule-based approach, assigning scores to each RFM component to calculate RFM. This is one of the two primary methods, the other being the Machine Learning clustering approach.
Step 1 : How to calculate RFM components?
Recency
As the difference between the current date and the date of the customer’s last transaction.
e.g. if Today is June 13, 2023 and the the customer last purchased on May 10, 2023, the recency would be 34 days.
Frequency
As the number of unique transactions made by a customer over a particular time period.
e.g. If a customer made 20 purchases over the last year,
the frequency would be 20 transactions.
Monetary Value
As the total amount of money a customer has spent on your brand over a particular time period.
e.g. If a customer has spent 300$ over the last year,
the monetary value would be 300$.
Step 2 : How to assign scores to RFM components ?
Let’s assume we’ve calculated the Monetary Value for each customer and now wish to assign a score from 1 to 5. A common scoring method to achieve this is by using percentiles that segment customers into equal groups, known as quintiles.
• 1 (0%-20%) -> score 1
• 2 (20%-40%) -> score 2
• 3 (40%-60%) -> score 3
• 4 (60%-80%) -> score 4
• 5 (80%-100%) -> score 5
Each group corresponds to a score, effectively segmenting customers into “buckets” based on their Monetary value.
How to calculate
1. We sort our customers based on their Monetary value (in Ascending Order).
2. We calculate the positions of the percentiles:
Position = x / 100 * N, where x is the percentile and N is the Sample size (Number of Customers).
• For the 20th percentile (20%), we calculate the position as Position = 2, indicating that 20% of our customers have spent $30 or less. These customers fall into the 1st bucket, receiving a score of 1.
• Similarly, for the 40th percentile (40%), we calculate the position as Position = 4, indicating that 40% of our customers have spent $41 or less. Accordingly, these customers are categorized into the 2nd bucket, with a score of 2.
3. By applying this methodology, we can systematically organize all our customers into buckets, by assigning a score from 1–5 based on their Monetary value.
Note that we used a 1–5 scale to divide our customers into 5 groups (20%, 40%, 60%, 80%).
Other methods include:
• A 1–3 scale using 3 groups (25% and 75%).
• A 1–4 scale using 4 groups (25%, 50%, 75%).Scales recommended by customer base size:
• 1–3 for up to 30,000 customers.
• 1–4 for 30,000 to 200,000 customers.
• 1–5 for over 200,000 customers.But we can always choose a 1–5 scale and add hyper categories if needed.
Step 3 : How to combine RFM componets to calculate the final RFM score ?
Let’s assume we have assigned scores to every customer for all the RFM components: Recency, Frequency, and Monetary Value. To calculate the RFM, the most common method we can use is:
• Concatenation: Combine the Recency, Frequency, and Monetary scores into one code.
Then, for every customer, we’ll have a score in the form of ‘531’, ‘425’, ‘311’, etc. What we need to do now is map these scores into human-readable categories.
For example, Champions. These are your best
customers, who buy frequently and spend the
most. A typical RFM score for a Champion might be ‘555’, ‘554’, ‘544’, ‘545’, ‘454’, ‘455’ or ‘445’ indicating high Recency, Frequency, and Monetary values.
Other categories might include:
Loyal, Potential Loyalists, Recent, At Risk, Lost etc
Which are the RFM Segments and their Corresponding Strategies ?
And how to proceeed?
Essentially, you need to filter a segment and export these email addresses into the email campaign software of your choice to start you campaigns !
Remember, the above strategies are examples. You need to form and follow the strategies that align best with your business and its specific requirements.
You can check out the real-world application of RFM analysis here.
