One of the main features of 5G networks is to provide high data rates with low variability to mobile users. However, this has been shown to be ineffective from the operator's perspective as it leads to abundantly unused network resources. Reallocating the unused resources to the same users increases back the data rate variability. A more efficient way is to provide different low-variability data rates to the users depending on their channel conditions, while not wasting network resources considerably. In this paper, the goal is to minimize the coefficient of variation across all users in the cell. To that end, we consider two approaches, one with reserved resources for every user and the other where the amount of allocated resources is decided on the fly. Then, for each of them, we look at different allocation policies and derive the maximum achievable constant rate for every user. We evaluate the performance on a real trace. Results show that utilizing policies with no-resource-reservation increases considerably the utilization of network resources and the achievable data rates at the expense of a slight increase in variability. Also, using our policies, the waste of resources drops to 15%, which is almost 4x lower than with other state-of-the-art techniques.