The adoption of a global index (GI) for performance evaluation has been increasingly recognized as a useful strategy for decision-making as it simplifies the interpretation and monitoring of the results. Because the GI is often built by adopting a combination of different procedures for normalization, weighting, and aggregation of indicators, it is challenging to select the optimal combination of procedures, since the countless combinations lead to different results. This paper proposes a method for the development of a robust and original GI for the evaluation of table grape production units (TGPUs). Various combinations of procedures were used to develop eighteen GIs for each TGPU. These are located in the lower-middle San Francisco valley in the northeast of Brazil, where their robustness was assessed by identifying outlier GIs and via a graphical analysis. Plausible GIs were reliably identified and a cluster analysis was conducted to categorize the TGPUs into groups considering each performance objective. The identification of the outlier GIs and the use of the plausible GIs in cluster formation constitute a new scientific approach to the topic, which can be extended to other applications and contribute to the sustainable development of several industries.