Unobserved quality challenges the empirical content of signaling theory, and often precludes the valuation of quality signals such as wine names. This paper uses the location of vineyard plots to control for unobserved wine quality when estimating the causal value of wine names on vineyard prices. The identification tackles unobserved spatial heterogeneity by newly combining a multi-cutoff spatial regression discontinuity design with plausibly exogenous name variations. We deal with standard requirements of causal inference – unconfoundedness and overlap – with instrumental variables and high-dimensional propensity models in a double robust framework. For the Burgundy region of France, we then recover the full causal signaling scheme of nested wine names with both a horizontal and a vertical dimension. This typical structure of names is monotone and complementary, as the names are consistently ordered within each dimension (rank preservation) and they present spillovers between them (umbrella effect). We find a high importance for unobserved wine quality, which produces heterogeneous signaling values.