In recent years, crowdfunding has increasingly become an important financing alternative of raising startup funds. To date, drawing upon Kickstarter’ Statistics, merely 35.79% of the crowdfunding campaigns can achieve their initial funding goal to successfully launch their projects. Previous studies have explored several determinants of crowdfunding success including the early stage of funding, project duration, categories and social networks. However, dynamic features such as project comments and updates are yet to be systematically studied. In this work, we propose a method to extract important features from backers’ comments and creators’ updates in different periods of project duration to predict the crowdfunding success. Experimental results show that incorporating dynamic features from backers’ comments and creators’ updates increases the accuracy of success prediction. Sentiment and textual quality features derived from project comments further boost the prediction accuracy. In addition, our research reveals the significant factors from comments and updates to improve the success rate of crowdfunding campaigns, which could be beneficial to the project creators.