We present a new methodology that enables a significant sensitivity improvement for transverse magneto-optical Kerr effect (T-MOKE) detection. For this purpose, we developed a novel measurement scheme, in which the polarization detection conditions are changed during the measurement sequence in a pre-defined way. An analytical expression of the associated T-MOKE signal pattern was derived, which allowed us to analyze and classify our experimental data in a straightforward way. Furthermore, this new measurement approach enables the identification of noise and false background signals that might be generated by the sample under investigation, the environment or the detection system itself and it provides a pathway to unambiguously separate all these effects from true T-MOKE signals. These capabilities significantly increase the sensitivity and robustness of T-MOKE detection. The method enabled us to measure magneto-optical signals for samples that are paramagnetic at room temperature or exhibit really small magneto-optical responses, even in the presence of false signals that were far larger in size. Our new methodology was integrated into a scanning wafer tool, which allows for nondestructive, laterally resolved surface characterization measurements and even has the capability of measuring optical and magneto-optical properties simultaneously.