Few studies examine the effectiveness of community-based early interventions in unselected samples of autistic children, especially those from minoritized backgrounds. These types of studies require attention to strategies for recruitment, retention, data collection, and support for community providers beyond those used in university-based trials. We conducted a pilot trial of Project ImPACT, a promising caregiver-mediated naturalistic developmental behavioral intervention, in partnership with the early intervention system in a poor, large city. We assessed recruitment and retention strategies, training protocol, and measurement battery. We recruited representative provider and family samples, and had good measure completion among retained participants. Retention varied by study arm, providers had relatively poor fidelity to the intervention despite substantial support, and our measures did not appear sensitive to change. Lessons learned include the need for (a) a ramp-up training period prior to starting the study, (b) intensive implementation supports, and (c) additional strategies for family retention.