Crack growth modeling has always been one of the major challenges in fracture mechanics. Among all numerical methods, the extended finite element method (XFEM) has recently attracted much attention due to its ability to estimate the discontinuous deformation field. However, XFEM modeling does not directly lead to reliable results, and choosing a strategy of implementation is inevitable, especially in porous media. In this study, two prevalent XFEM strategies are evaluated: a) applying reduced Young’s modulus to pores and b) using different partitions to the model and enriching each part individually. We mention the advantages and limitations of each strategy via both analytical and experimental validations. Finally, the crack growth is modeled in a natural porous media (Fontainebleau sandstone). Our investigations proved that although both strategies can identically predict the stress distribution in the sample, the first strategy simulates only the initial crack propagation, while the second strategy could model multiple cracks growths. Both strategies are reliable and highly accurate in calculating the stress intensity factor, but the second strategy can compute a more reliable reaction force. Experimental tests showed that the second strategy is a more accurate strategy in predicting the preferred crack growth path and determining the maximum strength of the sample.