Carbon labelling of food products serves as a demand-side tool with the potential to drive the essential shift in consumption patterns toward reducing climate impact. For carbon labels to influence food choices, they must enable consumers to recognize and adopt purchasing behaviour that lower their climate footprint. While inference plays a critical role in facilitating behavioural change, evidence remains sparse regarding how specific characteristics of carbon labels affect consumers' ability to accurately identify low-carbon products. This study investigates how different carbon labels affect consumers' efficiency in identifying low-carbon-emitting food products. Three labels are evaluated: (i) ‘Digit’ specifies the amount of CO2e-emissions from the production of the product, (ii) ‘Colour-Coded’ label indicates the overall climate impact from A to E, (iii) ‘Logo’ identifies the lowest-emitting products within each product category. Respondents in a survey in the United Kingdom were asked to identify the lowest-emitting food product in a set of tasks. All labels improved accuracy in the tasks when products from the same food category were included. Importantly, in the tasks that included products from different categories, the Digit outperformed both the Colour-Coded and the Logo labels. Notably, the Logo did not improve accuracy compared to no-label tasks. It is important that a carbon label informs about the overall climate impact rather than the within-category performance, should the label help consumers identify changes that contribute to significant reductions in climate impact. © 2025 The Authors