Multicriteria Disaggregation - Aggregation (D-A) approaches results to the estimation of Decision Makers' preference models (usually of additive value) through interactive procedures, where the global preferences of DMs are analysed. The low robustness of preference models, presented in many cases, can be the result of the ill-structured problem formulation or can reflect the real thoughts of DMs. The case of collaborative decision making presents more complicated situations. This research work describes the use of visual techniques based on 3d graphs and a set of indices, which can be used for picturing and comprehension of the low robustness in collaborative decision making problems. Also, the frame of feedbacks which can be utilised for the reducing and the exploitation of the low robustness is described and illustrated through a case study.