Optimising the Fashion E-Commerce Journey: A Data-Driven Approach to Customer Retention
A fashion e-commerce company offers a wide range of products from domestic and international brands that are popular with young people. However, there has been an increase in non-organically acquired customers, many of whom do not return to make repeat purchases. This has led to a higher customer ch...
Saved in:
| Published in: | Knowledge engineering and data science (Online) Vol. 7; no. 1; p. 58 |
|---|---|
| Main Authors: | , , |
| Format: | Journal Article |
| Language: | English |
| Published: |
24.08.2024
|
| ISSN: | 2597-4602, 2597-4637 |
| Online Access: | Get full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | A fashion e-commerce company offers a wide range of products from domestic and international brands that are popular with young people. However, there has been an increase in non-organically acquired customers, many of whom do not return to make repeat purchases. This has led to a higher customer churn rate, with a significant proportion of non-organically sourced customers failing to become repeat purchasers. Consequently, a churn analysis and prediction model were developed to address this issue. This paper employs the Recency, Frequency, and Monetary (RFM) framework for churn analysis and prediction. The framework is underpinned by three key dimensions: last purchase recency, purchase frequency, and total transaction value. Seven machine learning algorithms were evaluated to identify the optimal approach. Following a comparative analysis of these models, Random Forest emerged as the superior algorithm, demonstrating an accuracy of 0.99, precision of 0.97, recall of 0.99, ROC AUC of 0.98, and F1-score of 0.97. Consequently, this model will be utilized for churn prediction. Based on the analysis and modelling, several recommendations are offered to enhance customer retention for the fashion e-commerce platform. In addition to predicting churn, this paper provides insights into potential refinements to the churn prediction model, such as real-time monitoring, personalized customer experiences, analysis of customer feedback, and lifetime value analysis. |
|---|---|
| ISSN: | 2597-4602 2597-4637 |
| DOI: | 10.17977/um018v7i12024p58-70 |