Case Study

Client:

SHEIN

​

Challenge:

An online retail business needed to better understand its customer base to develop more effective and personalized marketing strategies. The goal was to segment customers based on their purchasing behavior to identify high-value and low-value groups.

​

Solution:

We used a K-means clustering algorithm in RStudio to segment the customer data1. The analysis was built upon key metrics derived from the raw dataset:

• Total Sales 2

• Order Count 3

• Average Order Value 4

​

Before running the model, the data was pre-processed to ensure accuracy. This included removing records with negative quantities (cancelled orders), dropping records without a customer ID, and excluding the incomplete month of December.

​

Results:

The K-means model successfully identified four distinct customer segments. These segments were visually represented in scatter plots that showed clear divisions between customer groups based on their purchasing habits.

• Low-Value Segment: Located in the bottom left of the plots, this segment consisted of customers with infrequent purchases and low total sales.

• High-Value Segment: Found in the top right of the plots, this segment represented customers with high order counts, high total sales, and a high average order value.

​

Further analysis using a silhouette score confirmed that four clusters were the optimal number for the model, with a score of 0.4120 for Cluster 4. Based on the results, the high-value customer segment was identified as Cluster 3.

​

Conclusion:

This project successfully segmented the customer base into actionable groups. The insights gained allow the client to:

• Tailor Marketing Campaigns: Direct high-value promotions and loyalty programs specifically to the top-tier segment (Cluster 3).

• Improve Retention: Implement targeted strategies to encourage customers in the low-value segments to increase their purchase frequency and order value.

• Optimize Budget Allocation: Allocate marketing resources more efficiently by focusing on the most profitable customer groups.

This data-driven approach provides a clear roadmap for creating a more personalized and effective marketing strategy.