A revision of the standard approach to characterization of thin-semiconductor-layer Hall samples has been proposed. Our results show that simple checking of I(V) curve linearity at room temperature might be insufficient for correct determination of bias conditions of a sample before measurements of Hall effect. It is caused by the nonlinear behaviour of electrical contact layers, which should be treated together with the tested layer a priori as a metal-semiconductor-metal (MSM) structure. Our approach was examined with a Be-doped p-type InAs epitaxial layer, with four gold contacts. Despite using full high-quality photolithography a significant asymmetry in maximum differential resistance (Rd) values and positions relative to zero voltage (or current) value was observed for different contacts. This suggests that such characterization should be performed before each high-precision magneto-transport measurement in order to optimize the bias conditions.