Look a Like Segmentation
The Client’s Need
An important E-commerce company had a loyalty program under a set of criteria which did not categorize their customer universe in line with the company´s goals and interests.
That meant the program was not able to discriminate between premium and standard customers. Because of this, the company spent a lot of effort and money trying to keep low value customers.
The AIM
To propose a new methodology which would allow the company to make a customer segmentation in a predictive way, on the basis of their past shopping behavior. With this new methodology, it would be possible to apply aggressive attraction campaigns focused on keeping the most valuable customers, increasing revenue income when compared with the former segmentation system.
Gutbit Data Analytics was selected to perform this task because of our wide and successful trajectory in data analytics.
Solution
- Provide a new customer segmentation (classification system) based on a new predictive model which considers past customer behavior.
- Identify the most valuable customers to whom retention campaigns should be applied. (directed)
- Develop a continuous customer surveillance solution in a quality model.
- Integrate the model to the different channels from which we could gain customers
- Replicate the model to other countries of the region
Achievements
Techniques and Technologies
Neuronal Net Model
Phyton ETL process
Amazon web services
Jupyter Notebook informs
Trello Kanban
