Cross-correlation of consecutive Doppler images is one of the most common techniques used to detect surface differential rotation (hereafter DR) on spotted stars. The disadvantage of a single cross-correlation is, however, that the expected DR pattern can be overwhelmed by sudden changes in the apparent spot configuration. Another way to reconstruct the image shear using Doppler imaging is to include a predefined latitude-dependent rotation law in the inversion code ('sheared image method'). However, special but not unusual spot distributions, such like a large polar cap or an equatorial belt (e.g., small random spots evenly distributed along the equator), can distort the rotation profile similarly as the DR does, consequently, yielding incorrect measure of the DR from the sheared image method. To avoid these problems, the technique of measuring DR from averaged cross-correlations using time-series Doppler images ('ACCORD') is introduced and the reliability of this tool is demonstrated on artificial data.