Through a combination of ethnographic research and grounded theory, this dissertation explores the mechanisms underlying a heavily bureaucratic multilateral organization’s ability to design and implement a social innovation program in Zambia to tackle the pressing issue of HIV transmission among young people. It also utilizes a unique dataset to assess how the said innovation performs against organizational and user expectations. The first empirical study argues that the perpetual criticism of bureaucracy and its portrayal as the least suitable form of organizing for innovation are misleading and need to be re-examined in light of the evidence presented here. The complexities involved in most societal problems require a disciplined approach to innovation in which bureaucratic attributes do not hinder progress; they may in fact be beneficial. The study concludes that the common practice to regard innovation as a monolith impedes knowledge development in the fields of innovation and organizational design. The second empirical study uses naturally occurring data on an mHealth (mobile health) platform to understand how the target group interacts with the service and subsequently makes design and policy suggestions to improve the outcomes. Conducted as an observational exploratory study, it relies on a set of newly developed constructs that can be used as a foundation in future research. Overall, the dissertation contributes to the fields of innovation and organizational studies by providing evidence on the mechanisms and outcomes of organizing for social innovation in the context of development.