# Finding Lookalike Customers for E-Commerce Marketing

**Authors:** Yang Peng, Changzheng Liu, Wei Shen  
**arXiv:** [2301.03147](https://arxiv.org/abs/2301.03147)  
**Date:** 2023-01-09  
**Summary page:** https://www.summarizepaper.com/en/paper/2301.03147v1/

## Key points

- Customer-centric marketing campaigns are crucial for generating e-commerce website traffic
- Expanding the marketing audience is essential for driving business growth and delivering value to customers
- The authors developed a scalable system capable of handling hundreds of millions of customers
- The system uses a deep learning-based embedding model to represent customers and employs an approximate nearest neighbor search method to identify lookalike customers quickly
- Constructing interpretable customer similarity metrics allows the model to cater effectively to various business interests
- Extensive experiments demonstrate exceptional performance of both the system and the customer embedding model
- Leveraging advanced technologies like deep learning and nearest neighbor search methods can enhance e-commerce marketing strategies, optimize targeting efforts, drive traffic, and improve customer engagement and satisfaction

## Summary

In their paper titled "Finding Lookalike Customers for E-Commerce Marketing," authors Yang Peng, Changzheng Liu, and Wei Shen discuss the importance of customer-centric marketing campaigns in generating e-commerce website traffic for Walmart. With the increasing scale of customer data, expanding the marketing audience to reach more customers has become crucial for driving business growth and delivering value to customers. The authors present a scalable and efficient system designed to expand the targeted audience of marketing campaigns, capable of handling hundreds of millions of customers. The system utilizes a deep learning-based embedding model to represent customers and employs an approximate nearest neighbor search method to quickly identify lookalike customers of interest. By constructing interpretable and meaningful customer similarity metrics, the model can cater to various business interests effectively. Extensive experiments conducted by the authors demonstrate the exceptional performance of both the system and the customer embedding model. Overall, this research highlights the significance of leveraging advanced technologies like deep learning and nearest neighbor search methods in enhancing e-commerce marketing strategies. By efficiently identifying lookalike customers, businesses can optimize their targeting efforts, drive traffic, and ultimately improve customer engagement and satisfaction.