Custom scoring models operate based on predefined criteria and weightage. For instance, opening an email might score 1 point, clicking a link could score 3 points, and making a purchase might score 10 points. These scores accumulate over time, providing a comprehensive view of each subscriber's engagement level. Advanced models can also incorporate machine learning algorithms to continuously refine the scoring criteria based on historical data.