As the way people consume has changed, the focus of customer satisfaction has moved away from engineering and towards sensory preferences (SPs), necessitating a new approach to product development. This paper introduces a novel product development model that incorporates SP factors. First, utilizing literature and expert input, a domain-specific food sensory attribute knowledge graph is built. Then, SPs/non-sensory preferences (NSPs) are extracted from extensive online product evaluations using Latent Dirichlet allocation (LDA) and sentiment analysis. Customer preferences (CPs) are formed by combining quantifiable SPs/NSPs. Following that, CPs are used to design and distribute product components. To improve customer satisfaction, a new product development approach has been developed. Finally, a genetic algorithm is applied to the optimization model in order to determine the best product development approach. This model has been empirically and comparatively demonstrated to be practical and effective.