While considering processes of decision making we often encounter the problem of incomplete information. In group decision making (GDM) problems each decision maker is supposed to provide a matrix that describes his/her preferences over the set of given options. However, we have to take into account that an expert can not be able to define his/her preferences about all the options. Usually the problem is solved by an additional phase of estimating missing values. In the article we want to suggest and discuss a totally different approach that consists in adopting GDM algorithm so as it can deal with incomplete preference matrices.