The additive mean mode is generally employed to aggregate efficiency in the traditional DEA cross efficiency method. However, this method does not take into account the characteristics of the attribute utility data and its variation, which may lead to the loss of decision information and insufficient correlation between the efficiency scores and attribute values. Oriented by cross efficiency evaluation process, the changing characteristics of attribute preference and attribute utility are analysed in this study by method, the uniqueness of the entropy and entropy weight of attribute utility are verified, which helps to aggregate DMU’s self-evaluation weight (individual preference) into group weight (DEA systematic or group preference) and forms a novel peer-efficiency evaluation method based on group decision-making with only a list of common weight. Considering the consistency of attribute utility, the average value of cross efficiency is replaced by the comprehensive group utility, which makes the aggregation object transform from cross efficiency to weight. This method realizes the direct correlation between efficiency scores and attributes values and provides a path to improve relative efficiency. Finally, the feasibility and effectiveness of our proposed method are verified by a numerical example.