Mapping food and drink products to environmental sustainability metrics using retail transaction data

With increasing concerns around food sustainability, estimating the environmental footprint of diets is critical. Supermarket transaction data are becoming prominent as a valuable source of objective dietary purchase data. We developed a method to map environmental sustainability metrics to foods and beverages sold by a major UK supermarket, using sales data from the Yorkshire and the Humber region (2022). Products were mapped to global Greenhouse Gas Emissions (GHGE) estimates for 45 commodities in four stages. Initially, products were grouped into categories and linked to commodities. Subsequent stages disaggregated high-complexity or high-sales categories into subcategories with similar ingredients, using retailer categorisations (Stage 2) and word searches within product descriptions (Stage 3), and finally refined categorisations to aid interpretability (Stage 4). The product with the highest sales in each subcategory was selected as an indicator product and mapped to commodities using data on ingredient proportions. Land Use and Water Use estimates were generated using the final mapping scheme. A look-up tool was produced linking categories to environmental sustainability metrics for use with other food product data. By Stage 4, 98·6% of >27,000 products were mapped to GHGE, using 200 category/subcategory-based GHGE estimates. Disaggregation revealed significant variation in GHGE estimates: up to a three-fold difference between Stage 1 and 4 estimates for the same category, and up to a 30-fold difference between subcategories within the same category. Disaggregation of complex categories is important for accurate estimation of sustainability indices. Our sales-guided approach balances accuracy and efficiency when dealing with large supermarket data and could support a wide range of research into sustainable diets.