In a previously published report, our researcher-practitioner partnership demonstrated the effectiveness of a coursework plus higher-dosage coaching professional development intervention (Expanding Quality, EQ) at improving the quality of teacher interactions with children birth-to-3 in center- and home-based child care settings. The current study examined the 4,796 written feedback statements from 515 coaching sessions, provided to the two groups that received coaching (n = 68) to attempt to shed light on how variation in feedback may have contributed to the coaching's effectiveness. In the first study of its kind, detailed content of coach-to-coachee feedback was analyzed in relation to changes in the quality of teacher-child interactions. Twelve types of feedback were qualitatively coded, and consolidated into three, nonexclusive categories: "Positively influential" (e.g., suggesting specific changes, explaining why teacher actions are good for children); "noninfluential" (e.g., related to professional development outside of teacher-child interactions, vague comments); and "CLASS-related" (i.e., referring to support of children's language and learning, or emotional-behavioral development). Results indicated that coaching feedback was strongly aligned with the strengths-based, behaviorally-anchored model put forth by EQ as well as the measures of interaction quality. Nevertheless, there were variations in coaching feedback, primarily due to coach, and which impacted the two coaching dosages (5 vs. 15 hours) differently. For three of the four interactional outcomes, the impact of the shorter coaching dosage on interaction quality was more vulnerable to different amounts and types of feedback, whereas the longer dosage's effectiveness was not impacted by feedback differences. For the fourth outcome--support of language and learning at 4 months postcoaching--positive impacts were seen only when the dosage was longer, and when more influential feedback was intensified.