Stock deviation of casted parts needs to be handled by adapting the CNC machining code to every new batch. Usually, human experts deal with this workpiece referencing task, but the demand for automation is expressed by the industry. This paper introduces a Digital Twin (DT) supported workpiece referencing method, implemented in the following steps: building the DT of the CNC machining cell, loading measurements of the casted part into the DT, solving the workpiece referencing problem as a convex optimization problem, and generating the compensated CNC code. The proposed approach is illustrated in a case study from the automotive industry.