We report a comprehensive experimental study to analyze the limiting factors and physical mechanisms that determine the achievable performance of transverse magneto-optical Kerr effect (T-MOKE) ellipsometry. Specifically, we explore different approaches to achieve high sensitivity and reduced acquisition times. The best sensitivity is observed for an incident light polarization with balanced s-p components. We also verify experimentally that the method’s theoretical description is accurately describing data for any s-p combination of the incoming light. Furthermore, two alternative measurement strategies are explored by using different measurement sequences for the polarization sensitive optics, which both achieve a very comparable, high quality of results. Signal-to-noise ratios and systematic deviations are measured and analyzed based on a large number of nominally identical measurement repeats, both for entire signal sequences as well as for individual Fourier components of the magneto-optical signal generated by a sinusoidal magnetic field sequence. Hereby, we observe that while higher order Fourier components have a significantly reduced signal amplitude and correspondingly exhibit reduced signal-to-noise and repeatability performance, signal-to-noise ratios always exceed values of 100 even for the lowest signal Fourier component and the lowest signal sample that we investigated, illustrating the extremely precise nature of T-MOKE ellipsometry.