Advances in thermoelectric materials in recent years have led to significant improvements in thermoelectric device performance and thus, give rise to many new potential applications. In order to optimize a thermoelectric device for specific applications and to accurately predict its performance ideally the material's figure of merit ZT as well as the individual intrinsic properties (Seebeck coefficient, electrical resistivity, and thermal conductivity) should be known with high accuracy. For that matter, we developed two experimental methods in which the first directly obtains the ZT and the second directly measures the individual intrinsic leg properties of the same p/n-type thermoelectric couple device. This has the advantage that all material properties are measured in the same sample direction after the thermoelectric legs have been mounted in the final device. Therefore, possible effects from crystal anisotropy and from the device fabrication process are accounted for. The Seebeck coefficients, electrical resistivities, and thermal conductivities are measured with differential methods to minimize measurement uncertainties to below 3%. The thermoelectric couple ZT is directly measured with a differential Harman method which is in excellent agreement with the calculated ZT from the individual leg properties. The errors in both the directly measured and calculated thermoelectric couple ZT are below 5% which is significantly lower than typical uncertainties using commercial methods. Thus, the developed technique is ideal for characterizing assembled couple devices and individual thermoelectric materials and enables accurate device optimization and performance predictions. We demonstrate the methods by measuring a p/n-type thermoelectric couple device assembled from commercial bulk thermoelectric Bi2Te3 elements in the temperature range of 30 °C-150 °C and discuss the performance of the couple thermoelectric generator in terms of its efficiency and materials' self-compatibility.