Before we get into just how you can perfect your RFM marketing strategy to see ecommerce sales skyrocket, let’s explain:
What is percentiles-based scoring in RFM Marketing?
As is true for any percentile math equation, splitting customers into percentiles helps you tell where a group of customers, or segments, stand relative to other segments.
Statistics by Jim gives this example: “a person with an IQ of 120 is at the 91st percentile, which indicates that their IQ is higher than 91 percent of other scores.”
By first measuring the behavior of your customers in terms of recency, frequency, and monetary (RFM) of their purchases from your store, you can improve audience targeting. This is because percentiles-based RFM scoring allows you to group customers together. Grouping customers means you only need to create a handful of marketing strategies, rather than personalizing follow-up to every single customer.
Why might your ecommerce use percentiles-based scoring?
You’re running a small, rapidly growing business. You don’t have time to write an email to every single client, one day after their purchase to thank them, two days later to propose their second purchase, five days after to offer them a discount, ten days later to share exclusive product releases if they sign up to your newsletter…
You certainly don’t have time to personalize each of these touches to every customer’s individual profile.
With percentiles-based scoring, you can reduce the number of customer journeys to plan; this means fewer nurture campaigns to design. Essentially, you can plan communications with hundreds of clients with ease.
Of course, scoring customers and grouping them into audience segments works well when you can automate nurture campaigns by triggering them at the right moment. Find out how our platform calculates RFM for you or read on to understand different ways to use percentiles in an RFM score calculation.
What are the different kinds of percentiles-based scoring?
Method 1: Quartile or quintile RFM score calculation
There are three variables by which an analyst will classify shoppers for an RFM segmentation strategy: the recency of their last purchase, frequency of purchase, and its monetary value. Each variable is divided into 5 groups of customers, equal to 20%. Then, each group is given a score, from 1 to 5, representing the level of importance of the behavior.
The top 20 percentile are assigned the number 5 representing the customers with the highest level of a variable (for example, purchase frequency in your store). The next 20% is number 4 and so on. The same is done for each of the three variables until each audience segment has a three-digit number, or RFM score, such as 523. Each customer is eventually assigned a value and, using a quintile score system, you will have 125 audience segments (5 x 5 x 5).
The percentile score method:
- Requires few resources to analyze
- Is the easiest method to interpret
- Is not ideal if the segments are very diverse
- Only considers three variables and no external factors
Method 2: Clustering RFM score calculation by k-means
If you’re not keen on the quartile or quintile RFM score calculation, an analyst can cluster RFM marketing scores to build statistical cohesion. This will mean that similarities within each group of customers are greater. Furthermore, clustering ensures fewer similarities between different groups.
According to research published in Segmenting Bank Customers via RFM Model and Unsupervised Machine Learning, segments should be calculated using mean averages of all the shoppers across the ecommerce audience, using a K+means calculation that selects the optimal (k) recency, frequency, and monetary value.
The clustering method:
- Takes longer and more resources to calculate
- Can include more variables than RFM
- Makes it easier to target narrower audiences with messaging
- Still does not consider all external variables
Method 3: LRMF score calculation with clustering by k-means
This final method for RFM score calculation requires a little studying. It also adds an important variable: length of relationship with the shop (hence the L added to RMF) by measuring the time between the first and last purchases. This extra variable ensures greater accuracy, by measuring the length of the customer-store relationship, leading to a better segmentation performance.
The LRMF with clustering method:
- Requires significant resources to calculate and analyze
- Groups customers by cohesive qualities with clear dissimilarities between segments
- Can extract individual behaviors by comparing the mean values of the LRMF variables within the segments, and between segments
- Facilitates significant personalization
The LRMF with clustering method is how Aument segments audiences. We’ve used this method for the Gogo online store, in fact. We calculated the standardized distribution of LRFM variables in the segments and defined them by the LRMF with clustering method, plus some refinement.
Gogo store is a highly specialist shop, and their team was keen to try something equally innovative in the market. We were thrilled to hear their feedback:
“Aument is about to be a game-changer for busy ecommerce stores. Keep an eye on their upcoming new releases, as will we!” said Diego Arévalo, Co-founder of Gogo store.
Adopting one of these methods is key to ecommerce. But if you’d rather not do any RFM score calculation and instead get straight to it with data-driven automations, let us do the leg-work: Activate Aument on Shopify.