AWCF can facilitate academic writing development, especially for novice writers of English as a foreign language (EFL). Existing AWCF studies mainly focus on teacher and learner perceptions; fewer have investigated the error-correction effect of AWCF and factors related to the effect. Especially lacking is research on how successfully students can use AWCF to correct errors in source-based academic writing. Meanwhile, "Grammarly," an automated writing evaluation (AWE) tool, is under-represented in AWCF studies despite its extensive use in EFL classrooms and its significant advantages. Correspondingly, we evaluated how effectively EFL students could use "Grammarly"-provided feedback for error correction in research writing and what factors were related to user responses (i.e., revision operations) and response accuracy. The results show that the error scores decreased significantly after revision and the participants correctly addressed 85% of the "Grammarly"-flagged usages. The error-correction success was impressive, considering how the technical nature of research writing might affect feedback accuracy and user responses, both related to response accuracy. In fact, feedback accuracy contributed to response accuracy both directly and indirectly (via user responses). The findings have implications for source-based academic writing instruction and AWE tool development.