Multi-user MIMO (MU-MIMO) is a precoding technique that allows for an Access Point (AP) to transmit data to multiple receivers in parallel resulting in an overall capacity increase. However, achieving these gains requires significant overhead in first choosing the subset of users to serve and measuring the Channel State Information (CSI) between that user group and the AP. Management of this overhead and other Media Access Control (MAC) decisions are key to the performance of MU-MIMO transmissions. Previous solutions attempt all-purpose approaches which attempt to work well regardless of deployment environment or frequency band. However, through thorough analysis of MU-MIMO transmissions with respect to band and environment, we show that an all-purpose solution will not allow for MU-MIMO transmissions to reach their full potential. In fact, leveraging the differences in MU-MIMO transmission characteristics for different bands and environments to develop separate protocols will allow for increased overall system performance. To address transmission scenarios with high channel variability, we present Pre-sounding User and Mode selection Algorithm (PUMA), a transmission mode and user selection protocol that leverages the characteristics of the highly variable 2.4/5.8 GHz transmission scenarios for efficient MU-MIMO overhead amortization. While PUMA results in significant MU-MIMO capacity gains for 2.4/5.8 GHz indoor environments, it will not allow UHF-band indoor and outdoor MU-MIMO transmissions to reach their full potential. To that end, we design and implement Feedback Removal with Opportunistic Zero-overhead channel EstimatioN (FROZEN), a protocol that harnesses the channel stability of the UHF band to eliminate sounding overhead by relying on CSI measurements from previously received uplink packets.