Understanding people’s
perception of their
carbon footprint

For the climate, flying to New York is worse than taking a long shower. But is it 10 times worse or is it 1000 times worse? The carbon footprint of our actions have been widely analyzed and quantified, but it does not mean that people are well aware of their impact.

With this project, we would like to understand people’s perception of their carbon footprint, and how does it compare to the actual carbon footprint of actions they take. This could help climate scientists, sociologists, news outlets, politicians, and the general public to improve climate communication and enhance climate action.

In an ideal world, we would ask people to answer questions such as "How much CO2 does taking the plane emit?", for all actions. But this is a tedious, difficult task for non-experts. Hence, we propose to compare pairs of actions. It is probably easier to say "Flying emits 1,000 times more CO2 than taking a shower".

To tackle this problem, we use recent advances in machine learning to run Climpact. To estimate the actual impact of actions from pairwise comparisons, we developed a statistical model of the observed impact ratio (i.e., "1,000 times more" from the flying and showering example above). This model enables us to:

  1. compute implicitly the carbon footprint of all actions;
  2. measure the uncertainty inherent in the answers we collect; and
  3. select the best pair of actions to show in order to obtain meaningful data.

Individual versus government and corporate responsibility

Although our goal is to evaluate the perception of individuals, all actions presented on Climpact also bears the responsibility of governments and companies, which can move things in the right direction through their policies and practices. One example is heating home for one year, which emits 3,646 kgCO2eq. (the highest in our dataset). Tenants cannot decide to change the heating system of the place they rent on their own, it's the landlord responsibility who in turn is largely influenced by laws and regulations set up by policymakers.


We computed the carbon footprint of 52 actions from scratch using a life-cycle analysis methodology (ISO 14040). Whenever possible, we computed the carbon footprint bottom-up using the Ecoinvent database. We curated two datasets. The carbon footprint dataset contains the carbon footprint of an average Swiss citizen for most aspects of his or her life (food, transportation, housing, leisure activites, etc.). The actions dataset is built upon the carbon footprint and derives actions of a citizen's daily life (flying from Geneva to New York, eating one year's worth of meat, heating home for one year, go skiing, etc.).

These two datasets are publicly available for download below:


Climpact is an original idea from Victor Kristof and Lucas Maystre in Patrick Thiran’s and Matthias Grossglauser’s INDY Lab at EPFL. This website has been designed by Jason Reynolds and developed by Andrija Kolić. The life-cycle analysis and the creation of the dataset is the result of an academic project by Alexis Barrou, Edouard Cattin, and Blanche Dalimier, under the supervision of Jérôme Payet. This work has won EPFL’s Durabilis Award 2021.


We are grateful to Robin Zbinden and Valentin Quelquejay-Leclère for the development of the proof-of-concept app and the collection of the initial dataset that kickstarted this project. We are also grateful to Lars Klein for helping with the German translations. Last but not least, we are really grateful to Jayson Reynolds for his amazing job giving a soul to Climpact with his impressive design skills.

Want to know more? Shoot us an email.