Most methods to assess ecosystem services have been developed on large scales and depend on secondary data. Such data is scarce in rural areas with widespread poverty. Nevertheless, the population in these areas strongly depends on local ecosystem services for their livelihoods. These regions are in focus for substantial landscape investments that aim to alleviate poverty, but current methods fail to capture the vast range of ecosystem services supporting livelihoods, and can therefore not properly assess potential trade-offs and synergies among services that might arise from the interventions. We present a new method for classifying village landscapes into social-ecological patches (landscape units corresponding to local landscape perceptions), and for assessing provisioning ecosystem services and benefits to livelihoods from these patches. We apply the method, which include a range of participatory activities and satellite image analysis, in six villages across two regions in Burkina Faso. The results show significant and diverse contributions to livelihoods from six out of seven social-ecological patches. The results also show how provisioning ecosystem services, primarily used for subsistence, become more important sources of income during years when crops fail. The method is useful in many data poor regions, and the patch-approach allows for extrapolation across larger spatial scales with similar social-ecological systems.