The widespread use of the product line approach allows companies to realize significant improvements in time-to-market, cost, productivity, and quality. However, a fundamental problem in software product line engineering is that a product line of industrial size can easily incorporate several thousand variable features. The complexity caused by this amount of variability makes variability management and product derivation tasks extremely difficult. To address this problem, we present a new method to optimize the variability provided in a software product line. Our method constructs a visualization that provides a classification of the usage of variable features in real products derived from the product line. We show how this classification can be used to derive restructuring strategies for simplifying the variability. The effectiveness of our work is demonstrated by presenting a case study of optimizing the variability in a large industrial software product line.