Propensity score predictor helped our client predict the customers likelihood to purchase a product

Authored by Ameex Technologies on 06 Nov 2019

Client is a leading e-commerce company selling baby nutrition products. They want to leverage the user’s website behavior. 

Current State

  • The client runs various marketing campaigns online and their marketing team would like to have information on their customer profile. 

  • The client can then run targeted marketing campaigns to the selected customer which matches their profile 

Gaps

  • Absence of a proper mechanism to perform customer profiling  

  • Lack of real-time data 

  • Absence of a proper solution to consume the data.

Our Approach

  • Ameex developed a solution which extracted data from Google Big Query and stored in Cloud Storage in a structured manner. The analytical layer consists of the model to generate the propensity score, which indicates the likelihood of a customer to purchase the product. This also gives the flexibility to run targeted campaigns to appropriate groups of customers. 

  • Data from Big Query is fetched and moved to Google Cloud Storage 

  • Dask Parallel processing handles data processing tasks like handling missing values, categorical variables etc… 

  • Last 30 days data was taken and feature engineering was performed on it 

  • Model is developed using LighGBM and trained using Cross-validation to prevent overfitting.                                         

  • Features: This application includes 

  • Option to enter the customer Number and view the propensity score. 

  • Option to view the list of customers within a specific score range 

Business Impact

  • A solution which gives the propensity score and determines the likelihood to purchase a product  

  • An easier consumption mechanism which uses real time data of enrolled and non-enrolled users 

Want to learn more about how we have helped our clients with their analytics journey? let’s connect!

Case Studies