Least Squares Monte Carlo (LSMC) is a widely used proxy modelling technique in the European insurance industry. The technique allows users to model components of their economic balance sheets. This paper introduces its readers to LSMC Surgery, an approach developed by the authors that is useful in detecting underlying model issues. LSMC Surgery allows its users to understand the behaviour of the dependent variable – such as best estimate liabilities – in one, two or three dimensions of the risk space, thereby allowing users to zoom into the underlying sources of fitting issues. We also show how LSMC Surgery can help detect heteroscedasticity of residuals and suggest ways to improve regression models.