Climate change is likely to lead to more frequent and more severe droughts with risk of socio-economic suffering. Utilization of satellite data, available almost instantaneously, is a great tool now providing the backbone of drought monitoring. However, the measurement of drought for socio-economic impact is poorly understood, questioning our ability to deal efficiently with future challenges. This study showcases the challenge by estimating the impact of Ethiopia’s historical 2015 drought, reported to be the worst in five decades, according to different drought indicators. The analysis shows a sever meteorological and self-reported drought, with substantially less rainfall than average and a high share of households reporting drought exposure. Also, more than half of the rural households reported lower consumption in 2016 than two years prior, despite large amounts of food aid were distributed in response to the drought. There are, however, no or limited signs of an agricultural drought, based on vegetation indices and predicted crop losses. Agricultural production, price, and wage data also support 2015 being a normal agricultural year. Further, based on spatial rank correlation, none of the different drought indicators (self-reported, vegetation anomalies, predicted crop losses, or rain anomalies) provide a coherent story of where the drought was worst. In fact, across years the spatial rank correlation between drought indicators shifts sign, highlighting that the inconsistent information on drought from different indicators is not restricted to 2015. In terms of drought impact on consumption, no impact, on average, is found for rain anomalies, predicted crop losses, or vegetation anomalies in the growing season, while a large impact is found for self-reported drought exposure and a smaller impact for vegetation anomalies in the harvest season. The self-reported drought exposure is not a trustworthy drought indicator due to endogeneity, while the impact from vegetation anomalies in the harvest season seems driven by better than average, as opposed to worse than average, vegetation. With indicators showing two such different sides to the same drought and limited spatial correlation between them, the case study calls for more research on the suitability of these indicators, and care and robustness checks when utilizing drought indicators to understand socio-economic impact.