In promoting cleaner cooking, one of the sustainable development goals of the United Nations, the need for a systematic approach to group segmentation and target-group identification has been well recognized. Yet, no such approach has thus far been proposed. Here, we propose such a framework based on a Euclidean distance of households’ fuel-choice behavior. Applying this framework to the post Gorkha-earthquake data on 747,137 households from Nepal, we find that ethnicity explains the highest intergroup diversity (39.12%) in fuel-choice pattern, followed by income (26.30%), education (12.62%), and location (4.05%). Among the ethnic groups, Chepang-Thamis have the lowest intragroup diversity (Shannon index = 0.101) and Newars the highest (0.667). Once the factor that segments households into subgroups with the highest intergroup, but lowest intragroup, diversity is identified, better targeted policy instruments that align with the behaviours of the subgroups can be designed. The proposed framework can also be useful in designing many socio-economic policies, including those aimed at achieving several other sustainable development goals.