Segmenting the target audience using RFM analysis
Posted: Wed Jan 22, 2025 6:16 am
RFM analysis of the target audience is based on three parameters:
Recency is an activity indicator. It represents the time since the buyer's last action (purchase, authorization in a personal account, opening an email newsletter, etc.).
Frequency – the number of purchases (other actions) of the client.
Monetary – Lifetime value , that is, the life value of the buyer. It represents the sum of his purchases or the profit received from them by the company.
Often, in the process vk database of performing RFM analysis for each of the indicators, the groups are divided into equal intervals (from smaller to larger). For example, the recency of the last purchase is up to 1 week, up to 2 weeks, up to 3 weeks. Then you should calculate the boundaries of the clusters by determining the sum and difference of the mean with the standard deviation. Ultimately, you will get the largest number of customers in the r2f2m2 segment.
Indices 1 and 3 in RFM analysis refer to exceptional users who have certain behavioral characteristics. For example, buyers of cluster r1m3 (for any value of f) are customers who brought profit to the company, but after some time stopped buying goods. The reason for such changes will need to be determined using surveys.
The r3f3m1 segment has a great potential for increasing LTV (monetary). These are active customers who make purchases but do not spend much. In this case, it makes sense to offer them a discount on purchases of a certain amount or advise them to purchase related products based on their previous purchases.
RFM segmentation can help you create a much more effective way of communicating with consumers than sending messages to the entire target audience. Some customer data and Excel will help you with this. The analysis process itself will not take much time.
Read also!
"Cross-marketing: 5 examples and 8 mistakes"
Read more
Segmentation of target audience using cluster analysis
The main task of cluster analysis is to divide buyers into groups based on certain parameters. This method is most often visualized using a hierarchical tree. Each of its successive levels reflects narrowing factors of difference.
As a rule, one of the variations of cluster analysis is used – k-means. Let's consider the entire sequence of actions.
First, the number of clusters k is determined. The components of the division into groups will be divided by this number. It can be assigned manually (for example, using tree clustering), or calculated as an optimal value (using machine learning).
Then k arbitrary points should be selected as cluster centers. After that, the distance between these centers and the remaining points is calculated. To establish the belonging of a point to a cluster, it is necessary to calculate the minimum distance to one of the k-centers.
After this, new centers should be selected whose coordinates will be equal to the average value of the coordinates of the points inside the cluster. The points are again distributed among k-clusters. This sequence of actions is reproduced until the distance values inside the clusters begin to repeat. This indicates that the optimal division has been achieved.
Once you have formed clusters, you need to determine what characteristics the points in them are most similar to each other. In other words, you need to understand which customer behaviors are repeated most often. Boxplots can help with this. The values are each customer's performance on a specific parameter.
Recency is an activity indicator. It represents the time since the buyer's last action (purchase, authorization in a personal account, opening an email newsletter, etc.).
Frequency – the number of purchases (other actions) of the client.
Monetary – Lifetime value , that is, the life value of the buyer. It represents the sum of his purchases or the profit received from them by the company.
Often, in the process vk database of performing RFM analysis for each of the indicators, the groups are divided into equal intervals (from smaller to larger). For example, the recency of the last purchase is up to 1 week, up to 2 weeks, up to 3 weeks. Then you should calculate the boundaries of the clusters by determining the sum and difference of the mean with the standard deviation. Ultimately, you will get the largest number of customers in the r2f2m2 segment.
Indices 1 and 3 in RFM analysis refer to exceptional users who have certain behavioral characteristics. For example, buyers of cluster r1m3 (for any value of f) are customers who brought profit to the company, but after some time stopped buying goods. The reason for such changes will need to be determined using surveys.
The r3f3m1 segment has a great potential for increasing LTV (monetary). These are active customers who make purchases but do not spend much. In this case, it makes sense to offer them a discount on purchases of a certain amount or advise them to purchase related products based on their previous purchases.
RFM segmentation can help you create a much more effective way of communicating with consumers than sending messages to the entire target audience. Some customer data and Excel will help you with this. The analysis process itself will not take much time.
Read also!
"Cross-marketing: 5 examples and 8 mistakes"
Read more
Segmentation of target audience using cluster analysis
The main task of cluster analysis is to divide buyers into groups based on certain parameters. This method is most often visualized using a hierarchical tree. Each of its successive levels reflects narrowing factors of difference.
As a rule, one of the variations of cluster analysis is used – k-means. Let's consider the entire sequence of actions.
First, the number of clusters k is determined. The components of the division into groups will be divided by this number. It can be assigned manually (for example, using tree clustering), or calculated as an optimal value (using machine learning).
Then k arbitrary points should be selected as cluster centers. After that, the distance between these centers and the remaining points is calculated. To establish the belonging of a point to a cluster, it is necessary to calculate the minimum distance to one of the k-centers.
After this, new centers should be selected whose coordinates will be equal to the average value of the coordinates of the points inside the cluster. The points are again distributed among k-clusters. This sequence of actions is reproduced until the distance values inside the clusters begin to repeat. This indicates that the optimal division has been achieved.
Once you have formed clusters, you need to determine what characteristics the points in them are most similar to each other. In other words, you need to understand which customer behaviors are repeated most often. Boxplots can help with this. The values are each customer's performance on a specific parameter.