Stores develop customized assortment based on localization characteristics. In terms of performance analysis, theses changes make more difficult to analyse store performance. This adjustment procedure entails an internal benchmarking which needs small samples to control regional and assortment effects. This paper explores the influence of competitive environment and store neighbourhood characteristics on store efficiency by using three data envelopment analysis models (DEA). This research is illustrated by using real data from a French supermarket retail chain at the product category level. It is showed that the two-steps DEA model is a relevant analysis tool to take into account location aspects. It presents a discriminant power strong enough to derive managerial implications with small samples.