Retraining vs Fine-tuning or Transfer Learning? [D]

Hi! I am currently working on a project that is basically an e-commerce clickstream data. We take in data, find the intent of the user(XGboost) and price sensitivity(Xgboost), segregate the user in different segments based on their purchasing intent or their research or price behaviour(Xgboost), recommend the benefit like discount or free shipping(Linucp or Thompson sampling), etc.