The literature on economic valuation of ecosystem services increasingly recognizes that the welfare generating endpoint of biophysical changes could potentially be heterogeneous across individuals in the population. This paper suggests Q-methodology as a structured and transparent approach to attribute selection in Discrete Choice Experiments that is easy to document, by combining the in-depth subjectivity of qualitative methods and the statistical rigor of factor analysis to identify groups in the population. Valid welfare estimates are dependent on the beneficiaries being presented with relevant attributes in the scenario description. This paper develops and implements a three-step procedure to: 1) identify potential groups of beneficiaries motivated by distinct ecosystem service endpoints 2) classify respondents in a Discrete Choice Experiment into groups of beneficiaries, according to their ecosystem service motivation, and 3) tailor the design of the Discrete Choice Experiment to the group-specific ecosystem service endpoints of the beneficiaries. The application explores potential heterogeneity in the ecosystem service endpoints derived from water quality improvements in Denmark.