Modular product architecture design allows product variants to be derived with little effort and cost. An essential part of designing modular product architectures is fulfilling customer-relevant product features, which represent the differentiating aspects for the customer and, thus, stand for the economic success of a product variant. However, product features are not only subject to high fluctuations due to changing environmental influences, but are also of varying relevance to customers over time. Not considering these aspects in the development phase can lead to costly changes in the product architecture later on. To counteract this, these two perspectives are merged and a methodical approach is introduced that identifies new product features and analyses their future development in the context of customer benefit. For this purpose, the current and future relevance to customers as well as uncertainties are calculated using Adaptive Conjoint Analysis and a Monte Carlo simulation. The results are consolidated in a visualization and the product features are classified according to their future implementation in robust product architecture. The procedure is explained using the example of a product family of vacuum cleaning robots.