Teachers' instructional design skills influence their teaching practices and student learning performances. However, researchers have found that the traditional one-to-many model of preservice teacher education prevents preservice teachers from receiving timely and individualized feedback, making it difficult to fill in theoretical knowledge gaps to achieve better instructional design. In order to address this issue, a chatbot-based 5E learning approach developed on the constructivist 5E (engage, explore, explain, elaborate, and evaluate) learning model was proposed in this study. To investigate the effects of this learning approach, we conducted a quasi-experimental study at a university (31 in the experimental group and 37 in the control group). The chatbot-based 5E learning approach was applied by one class as the experimental group, while the control group applied the conventional 5E learning approach. The results showed that students in the experimental group reported more significant increases in learning performance, learning attitudes, and extrinsic motivation. In addition, learners were generally satisfied with the information quality supplied by the chatbot. The interview results showed that the approach could help students create a clarified learning framework and thus improve their learning performance.