This paper presents a comprehensive framework for causal uplift modeling that moves beyond binary conversion metrics to estimate heterogeneous treatment effects at the individual level. We develop three meta-learner architectures (S-learner, T-learner, X-learner) with explicit bias correction via propensity scoring, augmented by psychographic transition priors for behavioral context. The framework introduces a four-quadrant decision taxonomy (Persuadables, Sure Things, Sleeping Dogs, Lost Causes) grounded in decision theory. This work develops individual-level causal inference methodology complementary to Robinson (2026a, 2026b).