December 7, 2021
Data Science

RFM Analysis for Marketing to Segment Customers

RFM analysis is a way of statistically segmenting customers to create targeted groups based on three key attributes of customer behaviour – Recency, Frequency and Monetary Value.
Recency Frequency Value
RFM (Recency, Frequency, Monetary Value) is a way of identifying key customers based on three behavioural attributes. Customers are grouped into segments from which to drive marketing activity.

It’s well known that it’s more expensive to acquire new customers than retaining existing ones (five times more, in fact), meaning maximising the value of relationships with existing customers will see a much higher increase to the bottom line.

With potentially thousands of companies all going after the same customer, it’s more important than ever to understand who your customers really are.

This poses the question – how do you identify your best customers?

If you were to only take those who have purchased recently, you’re not seeing the wider context of those transactions in the buying cycle. Sorting by those who make regular purchases may not be your best customers as they could be buying repeat, low value items. While looking at the spend of a customer may be misleading as there could be a one-off transaction of a high value throwing off the data.

The best way to quickly identify your best customers is to use all three major buying characteristics together to create a score for each customer, from which they can be segmented and grouped to a common baseline.

RFM analysis is a fundamental way of getting a well-rounded view of your customer activity.

What is RFM?

RFM is a core component for most B2C marketing teams as it’s a statistical way of segmenting customers based on three major behavioural attributes:

These behaviours are scored relative to other customers and interpreted to form groupings. This is typically done by dividing the three attributes up into quintiles meaning each customer will have a score from 1-5 (one being the lowest, five highest).

Quintile line splitting 100 customers into 5 equal segments

Customers can then be segmented into behavioural groups depending on their scoring that can be used for a range of marketing campaigns.


Recency is measured by how recently the customer has interacted or purchased from an organisation.

This can be measured in multiple ways, one being defining the start period as the most distant point before being classified as a ‘Lost Customer’. However, this doesn’t always give a true representation of customers and can leave out high value performing customers in other areas of RFM.

The most effective way to measure Recency is to take all customers and evenly split them into quintiles.

While the output could span over an extensive period, it’s the most effective way to factor in all customer behaviour, and when combined with Frequency and Monetary Value it starts to build a full view of the entire cohort behaviour.


Frequency is how often a customer interacts or purchases from an organisation.

Purchase intervals will be very different depending on the business. Looking at a supermarket and a car dealership as an example, consumable items such as food and household supplies need to be replenished regularly, meaning transactions have a much higher Frequency and are more predictable. Whereas a vehicle isn't a regular purchase for most people, often with 3+ years between transactions.

While a supermarket and a car dealership may seem opposites in terms of a buying cycle, it’s possible to use Frequency of purchases to spot patterns in buyer behaviour, identifying those customers who are most likely to purchase so marketing can be directed towards them in a much more effective way.

By assigning a description to each quintile specific to your customers buying cycle (as seen on the table below), you can start to build up customer personas which can be brought together when looked at in combination with Recency and Monetary Value – more on that later!

Monetary Value

Monetary Value looks at how much money is spent in each transaction by a customer.

Typical RFM methodology is to use the total value of all customer transactions. In practice, this is highly correlated to Frequency meaning the best method of measuring Monetary Value across multiple transactions is to take the average value per transaction.

It can be very tempting to look at the customers who have spent the most money with you and target those in the hope of another big transaction. But in reality, that could have just been a one-off purchase of high value made a long time ago and the other products you sell aren’t relevant to them.

On the flip side, focusing on just high average order value customers leaves little attention for those customers who are high Frequency but with low average order values. Even though their individual purchases may not be high, their long term cumulative spend adds up.

To have a true representation of who the best (and most profitable) customers are, all three attributes of Recency, Frequency, and Monetary Value need to be looked at in unison.

RFM Segmentation for Marketing

There are many different methods that can be used to group customers. The majority of RFM models classify quadrants of quintiles and create a rules-based system.

The most basic versions of this group customers into 'bronze', 'silver' and 'gold' segments. This is easy to interpret but is too simplistic and creates too few segments for marketing teams to work with. A more practical approach is to use a matrix to create customer segments based on value and lifecycle.

The Data Refinery’s RFM model splits customers into a total of 10 segments, using the combined score from all three attributes to segment customers.

Square RFM matrix with 1-5 Recency on X-axis and 1-5 Frequency/Monetary on Y-Axis

The benefit of using this method is that the Y axis can be a combination of Frequency and Monetary Value ((F + M) / 2), or in cases where Frequency of customers is low (e.g. car sales), Monetary Value can be used on its own.

Remember those personas we talked about in the Frequency section? Using a matrix allows marketing teams to quickly group customers together and build personas based on their overall characteristics rather than just looking at a single attribute in isolation. It’s at this point where everything is brought together that a true representation of a customer can be created.

Each persona in the matrix has its own description and associated marketing action – for example; a “Potential Champion” are recent customers with average Frequency who have the potential to become “Champions” if nurtured. To get them to make this jump, marketing actions could be to offer rewards or memberships to maintain engagement and boost the amount they shop.

It’s completely up to you on how you define your RFM personas and the marketing actions based associated with them. As long as it’s aligned with the lifecycle of your customers!

The Data Refinery and RFM

As part The Data Refinery, we automatically generate RFM segments for our customers with suggestions of what marketing actions could be taken.

Each segment can be used to directly drive campaigns through your marketing stack without you having to manually export data from multiple locations and repeatedly upload it.

Request a demo to find out more about The Data Refinery and the Advanced Analytics capabilities you’ll unlock for your organisation.

Maximise your data value

See how The Data Refinery can unlock the value in your data.
Data Scientist at The Data Refinery

Check out our latest posts

Book your free 30 minute
data value review

Are you ready to supercharge your organisation’s analytics & AI capabilities,
but not quite ready to invest £100,000s for a bespoke solution?