Achievement of the ambitious environmental sustainability targets requires improvement of energy efficiency practices in private households. We demonstrate how utility companies, having access to smart electricity meter data, can automatically extract household characteristics related to energy efficiency and adoption of renewable energy technologies (e.g., water/space heating type, age of house, number and age of electric appliances, interest in installation of photovoltaic systems etc.) by using supervised-machine-learningbased green IT artifacts. The gained information enables design of customtailored interventions (such as promotion of personalized energy audits, ecologic services and products, or load shifting mechanisms) that trigger residents’ behavioral change toward environmental sustainability as well as improvement of utilities’ key performance indicators. Moreover, realizing privacy preservation concerns, we investigate the influence of smart meter data granularity and the amount of survey responses required for the artifact development on the household classification quality.