The World Health Organization (WHO) defines an effective round of mass drug administration (MDA) for lymphatic filariasis (LF) as one that reaches at least 65% of the target population. In its first round of MDA in 2011-2012, the National Program to Eliminate LF in Haiti achieved a 79% epidemiological coverage in urban Port-au-Prince. In 2013, coverage dropped below the WHO threshold and has declined year-over-year to a low of 41% in 2017. We conducted a retrospective qualitative case study to identify key factors behind the decline in coverage in Port-au-Prince and ways to address them. Our findings suggest that the main contributors to the decline in MDA coverage appear to be the absence of effective documentation of practices, reporting, analysis, and program quality improvement-i.e., learning mechanisms-within the program's MDA design and implementation strategy. In addition to their contribution to the program's failure to meet its coverage targets, these deficits have resulted in a high cost for the MDA campaign in both lost momentum and depleted morale. Through a proposed operating logic model, we explore how the pathway from program inputs to outcomes is influenced by a wide array of mediating factors, which shape potential participants' experience of MDA and, in turn, influence their reasoning and decisions to take, or not take, the pills. Our model suggests that the decisions and behavior of individuals are a reflection of their overall experience of the program itself, mediated through a host of contextual factors, and not simply the expression of a fixed choice or preference. This holistic approach offers a novel and potentially valuable framing for the planning and evaluation of MDA strategies for LF and other diseases, and may be applicable in a variety of global health programs.