Experimental evidence of the effect of providing households with cheap energy saving technology is sparse. We present results from a field experiment in which autopoweroff plugs were provided free of charge to randomly selected households. We use propensity score matching to find treatment effects on metered electricity consumption for different types of households. We find effects for single men and couples without children, while we find no effect for single women and households with children. We suggest that this could be because of differences in saving potential (e.g. some households do not have appliances where using a plug is relevant), differences in the skills relevant for installing the technology and differences in the willingness to spend time and effort on installation. We conclude that targeting interventions at more responsive households, and tailoring interventions to target groups, can increase the efficiency of programs