Public opinion and individual-level preferences have become a key explanatory factor for policy development in many theories in the field of comparative political economy. However, most of them (e.g. the new politics of the welfare state, varieties of capitalism, or dualization theories) imply arguments on the relative importance of different (welfare) policy options for individuals, rather than mere position preferences. Standard surveys, however, are blind to such measures of relative importance. In my talk, I will develop the conceptual change that a shift from position to priorities implies and argue that experimental conjoint survey designs provide us with a valuable tool to causally identify the relative importance individuals attribute to specific social policies, policy designs or reform dimensions. To illustrate this argument, I will present both a finished study on pension refrom in Switzerland, which assesses the effectiveness of compensation in welfare retrenchment, as well as the design and some pre-test findings of an ongoing ERC project "welfarepriorities", which studies social policy reform priorities comparatively across Europe.