The Middle Magdalena Valley (VMM) is well known as one of the more important basins in Colombia. It has numerous conventional reservoir targets from the tertiary period. The hydrocarbon in these formations was generated in La Luna formation. La Luna is part of the South American upper Cretaceous sequence, has been recognized as one of the most important hydrocarbon source rocks in the world. This formation is described as low permeability calcareous shale and limestone, black in color, with high foraminifera content and limestone concretions. The continuity of this formation has been established by several authors from Ecuador towards the Northwest of Venezuela. The primary focus of this paper is the optimized lithofacies definition in La Luna formation through the Hetorogenous Rock Analysis (HRA) and the correlation of these classes with core description, core analysis and productivity potential. HRA is a log-based unsupervised rock classification method applied in unconventional reservoirs. HRA defines rock classes based on their fundamental attributes of texture and composition as discriminated by log inputs while promoting the uniqueness of the classes. The electrical logs input selection for HRA model and quality control for optimal number of classes definition are critical processes to obtain the best correlation with the lithology description. In this case, several basic and advanced electrical logs combinations were used as HRA input to determine the best result taking into account the core facies. The productivity potential for this formation was defined by both, reservoir and completion quality criteria. The reservoir quality is mainly based on petrophysical and geochemistry parameters while the completion quality is based on geomechanical parameters. Finally a relationship between litho facies and productivity potential is found through the different domain results integration. This association is applied in nearby wells where no core data is available. The HRA model is used to estimate the litho facies lateral continuity in these wells to then predict the productivity potential. Once this methodology is properly calibrated with production data, the costs associated with core could be drastically reduced.