How to choose an emission factor you can defend
Updated 2026-10-11
An emission factor turns an activity amount (kWh, litres, tonne-km) into an emissions estimate. Picking the wrong one is one of the easiest ways to get a footprint wrong, and one of the hardest to spot later. Use this checklist.
1. Does the activity really match?
Read the activity label and the method note, not just the headline. “Electricity” can mean electricity generated in a country or electricity consumed there, and it can include or exclude transmission losses. The Ember factors on this site, for example, describe generation — see the method note on any Ember factor page.
2. Region and year
Use the region where the activity happened and the factor year closest to your reporting year. Grid factors move year by year: Germany’s Ember value went from 0.50359 kg CO2e per kWh in 2015 to 0.32965 kg CO2e per kWh in 2025. Using an old factor for a recent year can overstate or understate emissions by a large margin.
3. Units
Check the denominator. A factor “per litre” cannot be applied to kilograms of fuel without a density, and a factor “per tonne-km” needs both weight and distance. Our calculator converts only between units of the same kind (MWh to kWh, miles to km) and refuses everything else.
4. Gases: CO2 or CO2e?
A CO2-only factor leaves out methane and nitrous oxide. Some datasets publish each gas separately; others publish a combined CO2e value. Do not add them up yourself unless you know which global warming potentials the dataset intends. See CO2 vs CO2e.
5. Boundary: direct, well-to-tank or lifecycle
Some factors cover only direct emissions at the point of use. Others add the upstream supply chain of the fuel (well-to-tank) or the full lifecycle of the plant. Mixing boundaries within one inventory makes numbers incomparable.
6. Scope label
The scope label tells you how the dataset intends the factor to be used. Where a dataset does not assign a scope, we show “No scope assigned” rather than guessing.
7. Source quality and licence
Prefer factors published by the dataset owner (labelled primary here) over republications (secondary). Check that the licence allows your use — our licence register lists what we import and why some datasets are excluded.
8. Write it down
For each factor record the dataset, edition, year, region, value, unit and the date you looked it up. Every factor page gives you an APA and a plain-text citation plus a CSV row with exactly these fields.
When datasets disagree
Different datasets give different values for what looks like the same activity, because they draw different boundaries or use different input statistics. That is expected. Pick the dataset that matches your reporting framework, use it consistently, and explain the choice. Never average factors from different datasets: the result would describe no method at all.
Sources
- Licence register on this site (Data on this site)