It is common practice in wireless communications to assume strict or wide-sense stationarity of the wireless channel in time and frequency. While this approximation has some physical justification, it is only valid inside certain time-frequency regions. This paper presents an elaborate characterization of the non-stationarity of wireless dual-polarized channels in time. The evaluation is based on urban macrocell measurements performed at 2.53 GHz. In order to define local quasi-stationarity (LQS) regions, i.e., regions in which the change of certain channel statistics is deemed insignificant, we resort to the performance degradation of selected algorithms specific to channel estimation and beamforming. Additionally, we compare our results to commonly used measures in the literature. We find that the polarization, the antenna spacing, and the opening angle of the antennas into the propagation channel can strongly influence the non-stationarity of the observed channel. The obtained LQS regions can be of significant size, i.e., several meters, and thus the reuse of channel statistics over large distances is meaningful (in an average sense) for certain algorithms. Furthermore, we conclude that, from a system perspective, a proper non-stationarity analysis should be based on the considered algorithm.