When modeling consumer satisfaction in product design, consumer satisfaction is always assumed to be static. In fact, they are dynamic and evolve over time. Moreover, their relationships exhibit nonlinearity and fuzziness. While previous studies have introduced several methods to develop consumer satisfaction models, the resulting models typically suffer from “black box” issues, making them difficult for humans to interpret, as they do not provide explicit representations. This challenge has led to the emergence of explainable artificial intelligence. To address all the above concerns in the modeling, this paper proposes the nonlinear dynamic fuzzy regression model for consumer satisfaction modeling based on online reviews. The methodology employs sentiment analysis on reviews of the different time periods and obtains the values of time series consumer satisfaction as the datasets. Then, a multi-objective optimization algorithm is integrated into the method of fuzzy regression to determine the nonlinear model structure, and fuzzy coefficients are determined for each item through fuzzy regression. The developed models can transparently express the nonlinear, fuzzy, and dynamic characteristics of the relationship between product design attributes and consumer satisfaction. The proposed methodology is validated using the case study of smartwatch design and compared with the modeling methods of fuzzy regression and dynamic fuzzy regression. The results demonstrate that the proposed methodology outperforms the other two methods by reducing the modeling errors and enhancing the system’s credibility.