• users
• criteria
• scoring model
• behavior
• traditional customer segmentation
• e-commerce
• customer acquisition cost

This work of thesis has been carried out in the Dubai branch of the multinational company Caendra Inc., a company that produces and sells IT security training. Like any e-commerce company, it faces the common limitations of traditional Customer Acquisition Cost measurement and traditional customer segmentation.
These problems and needs led to the design and development a behavior based dynamic scoring model.
The model criteria were defined based on the literature and on all the possible actions that the user is able to make on the website. Such criteria have been implemented on IBM Marketing Cloud. In the end, we performed two training phases in order to calibrate the model, for obtaining the thresholds for the users’ different categories, and a test phase in order to verify its effectiveness, that resulted to be higher than 70%.