Enter the same route into several emissions calculators and the answers can differ substantially. The disagreement is not carelessness. It reflects genuine choices about what to count.
Fuel burn is the easy part
Fuel consumed on a flight is known reasonably well from aircraft type, route distance and typical operating conditions, and converting fuel to carbon dioxide is straightforward chemistry.
The difficulty begins with dividing that total among the people and goods aboard, which is an allocation question rather than a measurement one.
Every allocation decision that follows moves the per-passenger figure, sometimes by a large margin, without changing the fuel burned at all.
Occupancy and cargo change the divisor
A full aircraft spreads its fuel across more passengers than a half-empty one, so the same flight yields a lower figure per person when it is busy.
Most aircraft also carry freight, and a share of the fuel is fairly attributed to it. How that share is calculated, by weight or by revenue, changes the passenger figure.
Calculators using average load factors will therefore differ from those using route-specific data, and both are defensible.
Cabin class multiplies floor space
A premium seat occupies several times the floor area of an economy seat, so allocation by space assigns it a correspondingly larger share of the fuel.
Applying class multipliers can raise a business-class figure to several times the economy figure on the same aircraft.
Calculators that ignore class produce a single average, which understates premium travel and overstates economy on the same route.
Non-carbon effects are the largest uncertainty
Aircraft emit at altitude, where water vapour forms contrails and nitrogen oxides alter atmospheric chemistry. Both affect warming beyond the carbon dioxide alone.
Contrail effects are short-lived and depend heavily on atmospheric conditions, time of day and route, which makes them difficult to attribute to an individual flight.
Some calculators apply an uplift factor to account for this and some do not, and that single decision can nearly double a reported figure.
How to read the numbers usefully
Comparisons within one calculator are reliable, because the assumptions are held constant. Comparisons across calculators are not.
The robust conclusions survive every methodology: flying less often matters most, direct routes beat connections, economy beats premium, and short flights are inefficient per kilometre.
Treating the output as an order of magnitude for comparing choices, rather than as a precise quantity, matches what the underlying method can support.