Various complexities involved in organizing and assessing computer science (CS) capstone programs are well-documented. We explore how student learning can be improved by studying the role of instructors and external industry professionals (``clients'') as project mentors in a capstone program. Three mentoring models were included: the instructor mentors teams in both semesters (I-I), a client is the mentor in both semesters (C-C), and a hybrid model where the mentor is the instructor in the first semester and a client in the second semester (I-C). The study included 287 unique participants in a two-semester capstone sequence (574 total observations) across three academic years at a public university. Collected data included aggregated grade values for assessing student performance in the course, student evaluations of the course, and self-rated student perceptions regarding their own career readiness. Analysis of the data indicated that the hybrid (I-C) mentoring model produced higher values in the course evaluation and career readiness categories. Qualitative data in the form of anonymous student comments are discussed. The study’s implications and possible limitations for replication are discussed, including capstone program length, cohort size, and client recruitment.