Nowadays, different techniques exist to study macromolecular structures, however, X-ray crystallography still being one of the more powerful technique to obtain atomic structural information. Crystallography has two main limitations - obtain crystals and solve the phase problem. Even the data collection and analysis methods have highly improved in the last decades, the phase problem can be critical if there is not a model with high similarity for molecular replacement. SAD/MAD are the second approach after MR to succeed solving the phase problem. These techniques require a tunable X-ray beam and the presence of an atom in the crystal that absorb X-ray radiation and produce anomalous signal, such as Se, S, Zn⁺²... In this study, we analyze two different SAD data collection approaches to increase the success of direct methods for phase determination - continuous collection and inverse-beam geometry (where Friedel’s pairs are collected close to each other avoiding radiation damage effect). This analysis has been done using crystals from the well study egg-hen lysozyme and the human protein survivin, which test the methods when the resolution and the diffraction are not optimal. In addition, a pairing reflections analysis using Bayesian inference has also been done. Our preliminary results show that inverse-beam geometry collection not only does not improve phasing but can also make it difficult to solve the structure when the collection is not optimal. Moreover, the use of bayesian inference for reflection pairing looks to improve the further data analysis.