Valuing ecosystem services with stated preference methods requires operationalizing these services for the purposes of the survey. This study presents a process for selecting agricultural ecosystem services for attributes of a choice experiment and analyzes how the relevance of these attributes affects respondents’ preferences, allowing for individual preference and scale heterogeneity. The results show that a non-significant cost attribute is associated with low relevance of the attributes, and that respondents who consider the attributes relevant have less uncertainty in their answers. The findings emphasize the importance of attribute selection when the object of the valuation is complex, such as certain ecosystem services.