In this paper we discuss how three criteria – carbon effectiveness, cost efficiency, and equity and co-benefits – can be incorporated in the experimental design for Reducing Emissions from Deforestation and Degradation (REDD). We discuss how additional design components can be introduced using a propensity score matching method in pre-selected study sites. Finally, we explain how this carefully designed study has helped to prepare the ground for a difference-in-difference framework for future outcome and impact evaluation. Given the lack of adequate data in developing country scenarios, and the mismatch between project implementation and impact evaluation, this paper highlights how a reliable baseline for forest carbon, ecology, and social and livelihood indicators could be created by collecting data at community level. This paper concludes with considerations for future outcomes and an impact evaluation of the REDD+ mechanism.