Analyzing RFM table

1. Analyzing RFM segments

We are now going to use the segmentation we created previously and analyze it.

2. Largest RFM segments

Let's first review the top 10 largest RFM segments. As we can see the lowest and the highest rated RFM segments are among the largest ones. It is always the best practice to investigate the size of the segments before you use them for targeting or other business applications.

3. Filtering on RFM segments

Another practical aspect of this segmentation is that it allows to do a simple selection of customers based on their RFM segment. In this case we select the bottom RFM segment with the lowest segmentation of 111.

4. Summary metrics per RFM score

The RFM Score has a smaller number of unique values and we will analyze some metrics for these groups. Lets calculate average recency, frequency and monetary value, and then the count of customers in each group. As you can see, the sizes are fairly similar across the RFM Score groups, and each of the RFM values are better with the higher RFM Score segment.

5. Grouping into named segments

Although intuitive, it is still hard to fully interpret this segmentation. Often, we will group customers based on their RFM scores into an even smaller number of segments. First we will create a segment_me() function to which we will pass a dataframe and return different segments - either gold, silver or bronze - based on their RFM_Score values. Then, we will apply this function to our datamart and create a variable called general_segment. Great, let's analyze the RFM values and group sizes across these 3 segments!

6. New segments and their values

From the first look our segmentation does make sense. In reality it can take multiple takes of trial and error to find the right cut-offs. In the next lesson we will use k-means clustering to build these segments without arbitrary chosen thresholds.

7. Practice building custom segments

Now, you will practice to build and create the segments on your own.

Create Your Free Account

By continuing, you accept our Terms of Use, our Privacy Policy and that your data is stored in the USA.