RFM Analysis: Customer Segmentation for Target Marketing Strategies

An RFM analysis project using Tableau

Image by Anand Singh via Medium

One of my goals for the year 2021 is to share my work as I continue to develop more skills as a data analytics professional. So here is my first project for the year.

The purpose of writing this article is to share my insights from the RFM Analysis I conducted using Tableau. I published the charts and dashboard to my Tableau public RFM Analysis -Superstore. Before I dive into the analysis section I will introduce RFM analysis as a marketing technique and its benefits to marketing campaigns. Also, I will briefly describe how I performed the RFM analysis using Tableau and the customer segmentation framework adopted. So let’s get started.

What is RFM Analysis?

RFM analysis is a customer segmentation technique that is mostly used to segment customers based on their last purchase, how often they purchased and how much they spent. RFM stands for Recency, Frequency and Monetary value. The recency metric is measured to know how recently a customer ordered a product(s) or a customer’s last engagement on a site ( e.g. 10 days). The frequency metrics measure the total number of purchases made by the customer within a specific time frame (e.g. 5 times within 30 days). Lastly, the monetary metric calculates the total or average money spent by the customer (e.g. $2,000). The RFM metrics are significant indicators of customer’s behavior as frequency and monetary value influences customer’s lifetime value, and recency affects retention.

Why do you need RFM Analysis?

RFM analysis is an effective marketing technique that allows marketing teams to target customers with personalized marketing campaigns based on the different segments curated from their recency, frequency, and monetary scores. RFM analysis can help marketing teams provide answers to the following questions:

• Who are the best overall customers?
• Who are the potential valuable customers?
• Which group of customers could contribute to the churn rate?
• Which customers could be retained?
• Which group of customers are likely to respond to engagement campaigns?

RFM Analysis using Tableau

To accomplish this project, I utilized Tableau and its Superstore dataset. The Superstore data has 9994 transitional records and 21 attributes. The main attributes that defined the RFM metrics are in the below table:

Calculated attributes used to define RFM metrics

The R-F-M scores were derived by calculating the percentile of customer’s recency, frequency, and monetary. See the below visual for the F-score calculation in Tableau (in this post, I only showed the F-score calculation). To calculate F-score, 20% of the top frequent customers will be assigned a score of 5 and 20% of the bottom frequent customers will be assigned a score of 1. This logic also applies to recency and monetary value.

Image shows the F-score calculation (Screenshot from Tableau)

After calculating the R-F-M scores, the next step is to combine the RFM values (see below visual).

Image shows the RFM combination calculation (Screenshot from Tableau )

Further, you group the RFM scores into different segments in Tableau.

Image of RFM Segments(Screenshot from my Tableau Desktop)

RFM Customer Segmentation

Image generated by Funlola Adeeko via Word Count

I segmented the RFM groups into 11 customer segments using the same segmentation framework adopted by LINPACK for Tableau. The customer segments are :

• Champions: purchased recently, frequent buyer and higher spender.
• Loyal: these customers spend good money and respond to promotions.
• Potential Loyalists: they purchased recently, spends a good amount of money, and purchased more than once.
• Promising: This group of customers are recent shoppers, but haven’t spent much.
• New customers: They bought most recently but not often.
• Need Attention: this cohort of customers are above average recency, frequency, and monetary values. They may not have made a purchase recently.
• About to Sleep: this segment of customers has below-average recency, frequency, and monetary values. We are likely to lose them if not reactivated.
• At Risk: these customers spent a high amount of money, purchased often, however it has been long since they made a purchase. They need to be brought back.
• Cannot Lose Them: this group of customers is important because they made the biggest purchases but they have not returned for a while.
• Hibernating: this cohort of customers are low spenders, they purchased a low number of orders and it has been a while since they made a purchase.
• Lost: Lowest recency, frequency, and monetary scores.

Customers are grouped into segments based on their recency, frequency, and monetary scores

Data Exploration

I explored some business questions and generated interesting insights from them. The below charts and dashboard are published to my Tableau public RFM Analysis -Superstore for easy interaction and drilling. Below are some of the questions I looked at:

How many unique customers are there in each RFM Segment?

• The heat map displays the unique count of customers in each RFM segment.
• 555 and 111 have the highest count at 22 and 32 respectively.
• 555 is segmented as “Champions” and 111 as “Lost”.

How much in sales were made in each RFM Segment?

• The champions customers have the highest monetary value in the store.
• It is surprising and interesting to see customers that are “at risk” are second-highest in sales.

What happened between “at risk” and “loyal” customers?

• Further drilling shows that the total number of “loyal” customers were higher than “at risk “customer across the years.
• However, in 2011 and 2012 the number of “at risk” customers was higher than “loyal” customers ( explore the chart here).
• In 2013 and 2014, the count of “loyal” customers increased while the count of “at risk” customers declined.

How often did customers purchase items quarterly in each region?

All regions recorded the highest frequency of customers in the 4th quarter and lowest frequency in the 1st quarter across the years.

How did each segment perform with customer’s frequency and recency?

• Customers who are “champions” made purchases on an average of 9 times and made recent purchases within 24 days.
• Customers who are segmented as “lost” purchased items on an average of 3 times. Their recent purchases were almost 2 years ago (513 days).

How many customers purchased items within a specific period?

• The line chart shows that 229 out of 793 Superstore customers made a purchase(s) in the last 28 days.

RFM Analysis: Superstore Tableau Dashboard

The interactive dashboard includes a line chart that could help marketing teams know the number of customers that purchased within a specific time period. The map provides the number of purchases that occurred in each state. The scatter plot would inform the marketing team ton the frequency and recency of customers based on the RFM segment. The customer details provide information on each customer segment group, recency, frequency, and monetary value. And lastly, the RFM Segment Summary has details of average recency, frequency, and monetary value per segment.

Closing Remarks

On a final note, Thank you for reading my first post. I enjoyed working on this project and I hope you enjoyed reading it. Please share your comments on the insights you discovered from the charts and dashboard. You can access the charts and dashboard here.

References

Pushpa Makhija, RFM analysis for Customer Segmentation[2020] Retrieved from https://clevertap.com/blog/rfm-analysis/