PredictN For Remarketing
Description
A Prediction SaaS platform which predicts qualified audience list for remarketing who carry the higher probability of conversion on your site within the next ‘N’ days
Responsibilities
- Extracting data from Google Analytics
- Training a predictive model to generate the optimal list of audience
- Deploying the model on Google Compute Engine for serving
Achievements
- Improved CPL of insurance company AegonLife by >55% using this solution
- Improved CVR of Eureka Forbes by 30%.
- Similar improvements were measured for one of the largest vehicle & health
- Insurance company in India which used the solution to target only a specific segment of the audience to remarket instead of traditional remarketing approaches
Technologies Used
- BigQuery and Google Analytics Core Reporting API were used to extract user session data on client’s websites.
- Jupyter Notebook, DataFlow and Cloud ML/Compute Engine were used to explore and process data and train the machine learning models
- Cloud ML/Compute Engine
- Google Analytics Management API to automate creation of audience
Libraries and Frameworks Used
- Numpy, Pandas, Scikit-Learn, TensorFlow, Keras, Catboost