Around the world, 1.1 billion people are severely affected by their lack of access to electricity. Other vulnerable communities receive low quality access, or face expensive prices that force them to restrict their consumption because of suboptimal technology choices made by their suppliers, which are sometimes forced by local regulation. Microgrids, properly sized and managed, may represent the best option to overcome these dilemmas, offering a tailor made supply. Today's standard methodologies to design isolated microgrids optimize the cost of supply as well as the cost of the energy not served with an exogenous per unit value for the lost load. They do not include community's restrictions, such as willingness to-pay, consumption level, budget constraints or its particular (endogenous) value of lost load. We developed a novel methodology that offers a range of microgrid designs to an isolated community, where each of them is optimal for a particular consumption pattern and value of lost load, from which the community may choose the one that best suits their needs. For this purpose, a Pareto optimal cost-coverage trade-off was constructed for an isolated community in northern Chile. A three-stage optimization was done: capacity (Genetic Algorithm), operation (robust optimization and mixed integer linear programming) and configuration (DC or AC). Diesel, gas, PV, wind and storages were modeled and 176 designs were found in total. More expensive microgrids (and with a larger electricity coverage) have hybrid mixes (conventional and renewable) and have an almost linear total cost from 298 to 249 USD/MW h for ENS from 0% to 28%. Lower quality microgrids are fully renewable, providing a very cheap but unreliable supply. The direct impact of lower-cost/limited supply microgrids offered here is the improvement of the quality of life of millions of vulnerable people, but it requires adjustments in the country's public policies of electrification programs. (C) 2017 Elsevier, ltd. All rights reserved.