Since their introduction feature models have been widely accepted and used in Software Product Lines. However, feature models may become too large easily, which complicates the management of the model. One of the reasons of complexity arises from the constraints imposed by the context on the product family. In this article we propose a strategy that uses context information for performing staged configurations on feature models. We introduce an organization structure for the context variability model and use this organization to determine the configurations for specification stages. This approach enables the elimination of context-related variability from the feature model, thus results in a reduction in the complexity of the model and enables to focus on the functional features of the product family.