This paper is the first product of a project which aims to build a Collaborative Smart Mapping of Mini-grid Action (CoSMMA), whose principal objective is to identify best practice in decentralized electrification projects. Using evaluations of 421 projects, from published research papers, we built a pilot CoSMMA which proves its feasibility. Its relevance is demonstrated by a meta-analysis, which reveals the principal characteristics of decentralized electrification projects which have positive impacts on sustainable development. Four main characteristics were considered: technology (source or energy), system size (power), decision level (from local to country level) geographic location. When searching for best practices, technology and system size must be considered together, because the chosen technology may constrain the power, which is provided by the system. We find that the most popular projects, which are based on Solar Home Systems (SHS) are not the most effective. The problem with SHS is not the use of solar energy, but the small system size often chosen for SHS. Mini-grids, of larger size, especially those which use hybrid renewable sources of energy, have more positive impacts, because these systems combine the benefits of sustainability and flexibility. In terms of decision level, we find that both top-down and bottom-up approaches have advantages, with the observation of a U-shaped curve for the influence of the decision level on the probability of obtaining positive impacts. Geographical location matters, as it is very often the key to system feasibility. We find that DEPs are more effective in Latin America than in Asia, and more effective in Asia than in Africa. We also attempted to study the type of effects resulting from DEPs. Descriptive data suggest that for some types of effects, positive impacts are more likely than for others. Decentralized electrification projects have a more positive impact on Lifestyle & NICT or Household agenda than on Economic transformation or Community life. However, this pilot CoSMMA does not contain enough information to study precisely the types of effects, because some types of effects have not been studied frequently in the existing literature. This is the case, for instance, for environmental effects, which have been rarely measured scientifically. Finally, we attempted to broaden our information set by including expert data, which was entered into the CoSMMA meta-analysis. We define expert data as data that are not supported by statistical tests with measures of significance, whereas the evaluations based on scientific data were supported by statistical tests of significance. The expert data may be valid, but our attempt to include it in the analysis failed at this stage. The determinants of unproven effects appear to be quite different from the determinants of proven effects in our meta-analysis, and using expert data would imply merging proven and unproven effects, which would totally blur the conclusions.