DoorDash Customer Segmentation Analysis Using RFM

Overview

This report outlines the exploratory data analysis (EDA) conducted on DoorDash customer data to build an RFM (Recency, Frequency, Monetary) segmentation model. The goal of the analysis was to identify key customer segments based on purchasing behavior and to provide actionable insights for targeted engagement strategies.

---

Data Preparation

The dataset was initially cleansed to ensure accuracy and ensure analysis was done with distinct records:

• Duplicate Removal: A TEXTJOIN function was applied to each row to concatenate all columns into a single identifier. Conditional formatting was used to detect duplicates, leading to the removal of 184 duplicate records. While there was no unique identifier to guarantee that these were duplicate records, the likelihood that two people make the same income, are the same age, spent the exact same amount on various categories, and have the same education level is very unlikely.

---

RFM Metric Calculation

We constructed the RFM model using the following logic:

Recency

• Measured as the number of days since a customer’s last purchase.

• Ranked using the RANK.EQ function in ascending order (i.e., more recent = higher rank).

• Divided into quintiles using nested IF statements to assign scores from 1 (least recent) to 5 (most recent).

Frequency

• Calculated as the total number of purchases across four channels: deals, web, catalog, and in-store.

• Ranked using RANK.EQ in descending order (more frequent = higher rank) and binned into 5 equal groups (quintiles) using the same IF logic as the one used to calculate Recency.

Monetary

• Derived from the MntTotal field, representing the total spending per customer.

• Ranked with RANK.EQ (descending) and converted to a 1–5 scale using the same quintile logic.

  
  

---

3. RFM Score Construction

I then assigned an RFM score for each customer by concatenating the individual Recency, Frequency, and Monetary scores using the CONCAT function, producing identifiers like 555, 432, etc.

This condensed RFM string captures a customer’s behavioral profile and is a key feature for segmentation.

---

Segment Assignment

Now with some critical thinking and a multi-layered IF formula, each RFM score was mapped to one of eight customer segments:

• Champion

• Loyal Customer

• New Customer

• Promising

• Potential Loyalist

• Need Attention

• Hibernating

• Churn Risk

These segments are behavior-driven and allow for precise marketing actions.

Nested IF function used to categorize segments

RFM Scores Segmented (Sample)

With the segmentation created, we have successfully identified each type of customer. This is crucial as it will allow us to tailor the marketing strategy for each customer and provide the appropriate promotion, which we will get to in just a moment.

---

Interactive Dashboard Design

To enhance interpretability, I developed a dynamic report using dropdowns and the VLOOKUP function. Based on the selected metric (Recency, Frequency, or Monetary), associated bar charts and summaries update automatically.

Visualizations were generated to showcase the behavior of each customer group by:

• Recency

• Frequency

• Monetary value

These heatmaps reveal patterns such as:

• Champions scored highest across all three dimensions.

• Churn Risk customers had high past frequency/monetary values but poor recency.

• Hibernating and New Customers show opportunity for re-engagement.

---

Engagement Strategy & Tailored Promotions

As mentioned prior, successfully segmenting each customer based on their calculated RFM score allows us to tailor the marketing strategy for each customer and send them the appropriate promotion.

We've given every customer an RFM score and a specific segmentation based on the score. All that's left to do now is to send out the tailored promotion to each customer type.

---

Conclusion

This RFM analysis is a very insightful analysis in regards to understanding your customer base and creating a powerful customer segmentation model. It combines statistical rigor with real-world usability, supporting personalized marketing, customer retention, and lifetime value optimization.

Through clear segmentation, dynamic reporting, and strategic messaging, this analysis creates a direct bridge from data insights to customer engagement actions.