A massive reduction in CO2 emissions is needed to reach the Paris climate goals on many societal levels - including individual consumer decisions. In contrast to resources as time or money, it is challenging to monitor one’s own impact on CO2 emissions. Designing digital assistants to provide users with information about their CO2 footprint could improve CO2 literacy (i.e., mental models), enabling individual behavioural change. We reviewed how characteristics of human bounded rationality (c.f. behavioural economics), such as temporal discounting, hamper effective interaction with CO2 as a resource. We examined user requirements for a carbon footprint tracker in an online study with 249 participants. Our study suggests CO2 tracking apps should encompass three types of functions: presentation of concrete CO2 emissions at a product or action level, direct comparisons between decision alternatives, and the proposal of better alternatives.