Solvation thermodynamics-based models for predicting liquid-phase nonidealities and fluid-phase equilibria are gaining attention in modern chemical process and product development. Among this class of thermodynamic models, COSMO-RS and COSMO-SAC are two variants used extensively in industry. A key input to these models is the so-called sigma profile, i.e., a histogram of charge density distribution over the molecular surface. Typically, sigma profiles are generated from quantum mechanical calculations with molecular structure and conformation information as inputs to the calculation. We present an alternative approach for generating sigma profiles from experimental fluid-phase equilibrium data, i.e., solubility. Specifically, we incorporate the conceptual segment concept of NRTL-SAC activity coefficient model into sigma profile generation. We generate "apparent" sigma profiles from linear combination of sigma profiles of conceptual segments represented by reference molecules selected for hydrophobic, polar attractive, polar repulsive, and hydrophilic conceptual segments. Conceptual segment numbers of the molecule of interest are identified from regression of available experimental data. We show applicability of this sigma profile generation approach with solubility modeling for four drug molecules: caffeine, aspirin, paracetamol, and lovastatin.