We consider the problem of evaluating the performance of a set of heterogeneous units. The heterogeneity may originate from environmental conditions or the unit itself. We propose a comparison of the units with benchmark units that perform well and represent various types of heterogeneity. The benchmark units are specified by a decision maker and used as alternative most preferred units in the value efficiency analysis. Each unit is associated with the best benchmark unit, in the spirit of the data envelopment analysis. The approach simultaneously takes into account preference information and the heterogeneity of units. We developed the approach to provide a framework for quantitative performance comparison for the Helsinki parishes. The parishes have heterogeneous structures and operate in various environments leading to different emphases of outputs. The special feature of the approach is that it is does not require the characterization of the heterogeneity by quantitative measures.