The lack of knowledge about how to effectively spread and sustain the use of research-based instructional strategies is currently a significant barrier to the improvement of undergraduate physics education. In this paper we address this lack of knowledge by reporting on an interview study of 35 physics faculty, of varying institution types, who were self-reported users of, former users of, or knowledgeable nonusers of the research-based instructional strategy Peer Instruction. Interview questions included in this analysis focused on the faculty's experiences, knowledge, and use of Peer Instruction, along with general questions about current and past teaching methods used by the interviewee. The primary findings include the following: (i) Faculty self-reported user status is an unreliable measure of their actual practice. (ii) Faculty generally modify specific instructional strategies and may modify out essential components. (iii) Faculty are often unaware of the essential features of an instructional strategy they claim to know about or use. (iv) Informal social interactions provide a significant communication channel in the dissemination process, in contrast to the formal avenues of workshops, papers, websites, etc., often promoted by change agents, and (v) experience with research-based strategies as a graduate student or through curriculum development work may be highly impactful. These findings indicate that educational transformation can be better facilitated by improving communication with faculty, supporting effective modification by faculty during implementation, and acknowledging and understanding the large impact of informal social interactions as a mode of dissemination. [This paper is part of the Focused Collection on Preparing and Supporting University Physics Educators.]