Feedback plays an indispensable role in pre-service teachers' microteaching practice. It provides essential information about their microteaching performance, which is of great significance in their reflection and improvement. As AI and teaching analytics advance, feedback is no longer exclusively human-generated. AI technologies are increasingly capable of delivering feedback on microteaching performance. Yet, the effects of differing feedback types on the microteaching practices of pre-service teachers are not well documented. This study examines the impact of three types of feedback--observation-based, teaching analytics-based, and combined (a combination of both)--on pre-service teachers' microteaching performance, scope of reflection, perceived usefulness, and satisfaction through an experimental research design. Sixty-five pre-service teachers voluntarily participated and were randomly assigned to three groups: observation-based feedback (N = 21), teaching analytics-based feedback (N = 23), and combined feedback (N = 21). The findings indicate that combined feedback was most effective in enhancing pre-service teachers' scope of teaching reflection, perceived usefulness of feedback, and satisfaction, but not on microteaching performance. However, when only teaching analytics-based feedback was provided, pre-service teachers perceived it as least useful and were least satisfied. The study discusses the implications of different types of feedback in teacher education.