The use of long-term (slow variations) for modelling radio propagation behavior in indoor scenarios limits the validity of such models, as most of the environments induced fast variations overlapped with the slow ones. Real world data gathered at different locations in two bands within Fifth Generation FR1 spectrum, 3 GHz and 5 GHz, provide insight on the effect of such fast variations and on the distribution models that would be useful to complement the long term analysis with short term behavior in order to improve the design of such 5G microwave networks. Comparisons among short term performance taking into account both the frequency, the visibility conditions and the environment size, shape and furnishing are presented with the focus on modelling as accurate as possible the narrowband channel. The use of ΔBIC (describing the difference between Bayesian Information Criteria indexes of each fitting) complements and confirms the insight provided by direct inspection on traditional fitting plots. This parameter could be interesting for future network deployments.