Nowadays, numerous tools allow you to precisely measure the success of your CRM and Online Marketing campaigns. With a data-driven approach you can even automate performance analysis per individual marketing channel or across all of them. But what about the lifecycle of a customer? How can you analyze it and ensure that the results get automatically updated over time?
One of the most efficient ways for measuring performance in CRM channels is by applying the RFM Model. This is a ranking system that considers three core dimensions of customers’ purchase behaviour: Recency, Frequency and Monetary Value. The data is primarily used to calculate the customer lifetime value (CLV) – the estimated profit per an individual buyer over a perceived time frame in the future. Additionally, it helps marketers to highlight the most valuable target audiences and alert about the dangers of churning users.
Which problems does this approach solve?
With the RFM Model, you can get better insights into the purchase behavior of your existing customer base and improve the performance of each target group in particular. Here are a few ways how you can apply the scoring system:
- Identify customers with significant spendings and extensive shopping needs – namely, your most active buyers
- Re-engage customers that have recently had fewer touchpoints with your brand
- Prevent users from churning
- Allocate marketing budgets across target groups by using a data-driven approach.
How does this work?
Originally, the idea behind the scoring model is splitting users into ranges as per each of the three dimensions (Recency, Frequency and Monetary Value) based on their past activity and customer value. The higher the range the customer is assigned to, the more value the customer brings to a company.
Pic 1. Traditional RFM Model
However, in order to visualize customer segments in the most efficient way, one can create a separate diagram for each range of Recency. Those ranges are based purely on the industry your company is in. For example, the purchase cycle for a number of eCommerce companies out there would often be around 90-100 days. Car manufacturers usually stick to the purchase frequency of multiple years, while local supermarkets would expect customers to check in fairly often.
Pic 2. Splitting the RFM Model into layer based on Recency axis
By taking Frequency and Monetary Value as axes of the diagram, it makes it convenient to place customers in the sectors they belong to based on their purchase history. For example, the segments with lower frequency would need reactivation while the active ones may react to upselling and cross-selling campaigns.
Normally, your most valuable customers are in the upper range of Monetary Value dimension but they are not always the active buyers with high spendings. Remember that frequent orders may generate extensive transactional and shipment costs on your side. The combination of purchase frequency and the cost of order processing helps you realize which segments, in fact, drive the most revenue.
Pic 3. Basic Frequency-to-Monetary-Value Matrix of the RFM Model
The number of ranges per RFM-dimension usually varies between 3 and 5. The more differences between product categories and audiences you have, the more ranges you can apply. With 3 ranges (Range 1, Range 2, Range 3) defined for each of the RFM dimensions, you get 27 possible combinations of scores.
Pic 4. Mapping ranges for Frequency and Monetary Value dimensions
Traditional approach to RFM scores
The lowest score in a default RFM Model is “111” (1 of 3 points for Recency, 1 of 3 points for Frequency and 1 of 3 points for Monetary Value). Users that belong to the segment with this score have not purchased in a while and the total of their last orders was under the average cart amount. Consequently, the highest score is “333” (or “555” if you use 5 ranges instead of 3). This score characterizes loyal customers with frequent purchases that are spending significant amounts of money in your shop.
Further combinations fall in between the above mentioned scores and characterize users in the segments according to their purchase behaviours.
When defining ranges, you should remember that outliers may have an impact on calculations so they have to be eliminated. Another scenario that needs precision in building the ranges is selling seasonal products like ice cream or snowboard equipment.
Applying weights to RFM dimensions
Another popular approach suggests applying individual weights to each of the three RFM dimensions. This generally helps to address the business priorities more precisely. However, this approach does not allow to accurately recognize and analyze the movements in RFM segments when aggregating the scores.
Can users move across RFM segments?
The data insights are not static – they change as your business evolves. Obviously, you want to keep track of that on a daily basis without having to ask your BI department to deliver it to you every single time. That’s where the real-time data flow from your database into the marketing platform helps. Having this in place, you can create dynamic segments and efficiently use them in your marketing automation.
Some essential marketing campaigns you can automate are the following:
- Reactivation of users with the lowest activity rate (increase Frequency and Monetary Value)
- Driving more purchases from occasional buyers with heavy spendings (Improve Frequency)
- Offering complementary products or goods with higher profit margin to frequent customers that are spending minimum amounts on purchases (Enhance Monetary Value).
Pic 5. Marketing best practices for targeting customers with most recent purchases
What kind of data is required for building the RFM Model?
In order to successfully implement the scoring, you need to ensure that you collect past orders with cart totals per each individual buyer. The age of data varies purely on the industry you are in, although mostly lies in the range between 90 and 365 days if you are in online business.
In CrossEngage, Recency and Frequency can be depicted solely with Completed Order event. Monetary Value should be represented by the sum of cumulative spendings per user over the selected time period. This data should be available in user profiles – you can create a custom field with the name you prefer and populate it with the numbers accordingly.
What is the process of setting up the RFM scoring in CrossEngage?
Assigning scores to users is only part of the process. In order to successfully implement the RFM Model, you have to decide which type of campaign – if any – each target group should receive and how campaign success is going to be measured.
The following steps will help you to keep everything on track:
- Build segments that include all three dimensions (see more details below).
- Define what goals you would like to achieve for each segment.
- Create campaigns and assign them to segments accordingly.
- Launch the campaigns and analyze conversions.
How should the RFM based segments be built in CrossEngage?
It goes without saying that you can replicate the RFM Model in your segments on our platform. Here is a quick starter guide to that.
Each score that you wish to depict has to be created using three conditions within the segment: one for Recency, one for Frequency, and one more for Monetary Value. In case you would like to apply the same communication strategy to several groups, you can add the score configurations to segment blocks and connect them with OR condition.
Use Completed Order event to identify the last purchase date.
At least one purchase in the last 3 months:
Completed Order > 0 from 90 days before dispatch until dispatch
At least one purchase in the last year but none of them in the last 3 months:
Completed Order > 0 from 365 days before dispatch until 91 days before dispatch
Completed Order = 0 from 90 days before dispatch until dispatch
At least one purchase within the customer lifetime but none of them in the last year:
Completed Order > 0 from the beginning of time until 366 days before dispatch
Completed Order = 0 from 365 days before dispatch until dispatch
Use Completed Order event to identify the number of purchases within the last 12 months.
No purchases within the last year:
Completed Order = 0 from 365 days before dispatch until dispatch
Between 1 and 3 purchases in the last year:
Completed Order => 1 from 365 days before dispatch until dispatch
Completed Order <= 3 from 365 days before dispatch until dispatch
More than 3 purchases in the last year:
Completed Order > 3 from 365 days before dispatch until dispatch
Create a custom user attribute such as traits.totalSpendings – you can pick any name you like. Afterwards, update all user profiles with the cumulative cart totals for all purchases in the defined period of time.
The total amount of money spent for the defined period is less than 25 Euros:
traits.totalSpendings < 25
The total amount of money spent for the defined period is somewhere between 25 and 100 Euros:
traits.totalSpendings => 25
traits.totalSpendings <= 100
The total amount of money spent for the defined period exceeds 100 Euros:
traits.totalSpendings > 100
How can the RFM Model be automated in CrossEngage?
Once your segments rely on real time data like in the example from above, they get dynamically updated, so you do not have to worry that they can expire at some point of time. Therefore, you can automate the campaigns that use these segments – just define the campaign cycle and schedule the deliveries accordingly.
Here are some more hints on how to ensure that the messages are relevant for users within each segment:
- Set a frequency capping for each campaign so that users do not get the same kind of message too often.
- Should you want to only send a specific message once and never again, please follow our instructions here.
- Sometimes images and texts in the messages get obsolete due to adjustments in your company’s strategy or even seasonal changes. Keep track of your media assets and don’t forget to refresh them from time to time.
We are always happy to discuss your marketing challenges and find the most appropriate solutions to them. Should you have questions regarding the RFM Model or its setup in CrossEngage, please do not hesitate to contact us.