The need for targeted policies, particularly those that induce behavioral changes, to achieve universal access to clean cooking by 2030 has been well recognized. Yet no systematic approach to the development of such policies has thus far been proposed. We attempt to fill that gap by integrating the techniques from data analytics (group segmentation) with the insights from behavioral economics (loss aversion and social norm) in developing targeted “nudges” to clean cooking. Individually customized nudges would be theoretically ideal but likely to be prohibitively costly; in contrast, nudges based on a “one-size-fits-all” approach are unlikely to be effective. Therefore, segmenting households into more homogenous subgroups based on certain socioeconomic factors is an important first step. Analyzing the post Gorkha-earthquake data from Nepal on 747,137 households’ fuel-choice pattern, we find that ethnicity, as a factor of segmentation, explains the highest intergroup diversity (39.12%) followed by income (26.30%), education (12.62%), and location (4.05%). Once the factor with the highest discriminatory power is identified and households are segregated into subgroups, we develop several subgroup-specific nudges, which may be less costly and yet outperform the traditional economic incentives in promoting clean cooking.