While it is well known which curricular practices can improve student performance on measures of conceptual understanding, the sustaining of these practices and the role of faculty members in implementing these practices are less well understood. We present a study of the hand-off of Tutorials in Introductory Physics [McDermott and Schaffer (Prentice-Hall, Upper Saddle River, NJ, 2002)] from initial adopters to other instructors at the University of Colorado, including traditional faculty not involved in physics educational research. The study examines the impact of implementation of tutorials on student conceptual learning across ten first-semester, and seven second-semester courses, for 15 faculty members over 13 semesters, and includes roughly 5000 students. It is possible to demonstrate consistently high, and statistically indistinguishable, student learning gains for different faculty members; however, such results are not the norm and appear to rely on a variety of factors. Student performance varies by faculty background--faculty involved in, or informed by physics education research, consistently post higher student learning gains than less-informed faculty. Student performance in these courses also varies by curricula used--all semesters in which the research-based Tutorials and learning assistants are used have higher student learning gains than those semesters that rely on nonresearch-based materials and do not employ learning assistants.