By continuously tracking the users’ innovative posts of the “Product Ideas” section of the Lego IDEAS community from January 2022 to January 2023, this paper obtains the data of 1,032 innovations posted by 720 users and 93,783 comments posted by 23,663 users. Furthermore, the text content analysis method is used to deconstruct online peer feedback, which is divided into constructive feedback and diverse feedback, and the hierarchical regression analysis and robustness testing methods are used to empirically test the theoretical hypothesis. The results show that there is an inverted U-shaped relationship between diverse feedback and users’ continuous update contributions, and constructive feedback positively affects users’ continuous update contributions. As important contextual factors, relational embeddedness and structural embeddedness strengthen the “too much of a good thing” effect between diverse feedback and users’ continuous update contributions, but relational embeddedness weakens the positive effect between constructive feedback and users’ continuous update contributions.