Longterm eigenbeamforming techniques that combine beamforming gains with spatial multiplexing of data streams have proved to be a very strong candidate for link level processing. On system level, fast scheduling brings multi-user diversity which increases with the number of active users in the cell. We propose a system with user specific longterm eigenbeams to transmit spatially multiplexed data streams. For each data stream a feedback measurement regarding the link quality is performed and reported by each user for link adaptation and scheduling. Due to the scheduling the interference pattern usually changes at the moment of using the feedback measurement, leading to mismatches in the link adaptation and throughput loss compared with the ideal case. The aim of this paper is to investigate these new insights in the multi-user MIMO systems and propose advanced methods to decrease the mismatch in the feedback measurement. Simulation results are presented to evaluate the schemes taking as base the UMTS - HSDPA system extended for MIMO applications.