Scott Gillespie, founder, tClara

With corporate travel and sustainability managers increasingly relying on flight emissions data for reporting purposes and to influence decisions at the point of sale, more questions are being asked of the methodologies being used to calculate those figures, and why different methodologies can produce very different results.

A joint analysis by climate tech firm Squake and travel analytics consultancy tClara compared six leading aviation carbon emission models and found differences of up to 6,700kg per passenger ¨C a margin of over 600 per cent ¨C for the same flight and cabin. These discrepancies call into question the reliability of emissions data at a time when business travel needs trusted metrics. Squake¡¯s research and development sustainability manager Yury Erofeev, and tClara founder Scott Gillespie, tell BTN Europe about the findings of the study.

Yury Erofeev, research and development sustainability manager, Squake

BTN Europe: What was the rationale for the study?

Yury Erofeev: The business travel industry¡¯s need for credible emissions reporting continues to grow as regulatory requirements such as the EU¡¯s Corporate Sustainability Reporting Directive (CSRD) and emerging mandatory climate disclosure laws in jurisdictions like California and Japan take hold. An earlier study with BTN Europe revealed significant differences between emission models, but also highlighted gaps in understanding their assumptions, limitations and appropriate use cases ¨C gaps this study sought to address.

BTN Europe: Explain how you conducted the study ¨C specifically which methodologies you examined and the datasets used ¨C and tell us about the key findings.

Scott Gillespie: We evaluated six flight emission models: ICAO¡¯s Carbon Emissions Calculator Methodology, the UK¡¯s DEFRA, France¡¯s ADEME, the Dutch CO2 Emissiefactoren, Google¡¯s Travel Impact Model (TIM), and Advito¡¯s GATE4. For each model, we calculated its per-passenger carbon emissions value for each of four possible cabin classes on 22,000 flights across 3,000 city pairs worldwide.

The answers across the ~88,000 flight-cabin tests varied greatly: by at least 200kg in 57 per cent of the tests; by at least 500kg in 31 per cent of the tests; by at least 1,000kg in 15 per cent of the tests; and by at least 2,000kg in 6 per cent of the tests. In one instance, for a first class flight between London and Perth, the results varied by more than 6,700kg. For context, the average amount of CO2 across all flights in the economy cabin was 254kg, and the average percentage gap between models was 300 per cent of the lowest model¡¯s amount.

BTN Europe: What¡¯s causing that variation?
?
Yury Erofeev:?Each model uses its own set of inputs and logic. Differences appear in several critical areas, one of them being user inputs. In this respect, some models use city pairs, others ask for distance flown, one model wants to know the aircraft type, while another expects users to know or guess at the aircraft¡¯s seating capacity. Others don¡¯t ask anything about the aircraft.
?
With flight distance, all models start with great-circle distance but five add extra distance for weather and traffic. For fuel burn, three models use aircraft and distance-specific estimates, while two use seat count and distance and one uses only distance and country-specific data.
?
The variables continue with load factors ¨C three models use industry sources, one omits it, another applies averages, and one uses cabin class and distance ¨C and with cabin multipliers, with one omitting them and others using different factors to allocate emissions among cabin classes.
?
Lastly, one methodology uses Tank-to-Wake CO2, while others may include Well-to-Tank or Well-to-Wake, with or without a Radiative Forcing Index (RFI) factor. Best practice for corporate travel and sustainability reporting is to use Well-to-Wake emissions ¨C which provides the most complete picture of a flight¡¯s carbon impact ¨C without Radiative Forcing Index (RFI) adjustments, for consistency and comparability, unless the reporting standard specifically requires RFI inclusion. Our analysis normalised all model outputs to this standard.
?
BTN Europe: What are the main factors driving the inconsistent emissions figures we¡¯re seeing?

Scott Gillespie:?Cabin class is a key driver of model variance and premium cabins and short flights showed the biggest gaps. Since premium cabins take up more space per passenger, most models allocate them a higher share of emissions. But cabin multipliers vary widely, especially for first class. Business class estimates also showed large differences, though less extreme. Economy class values on long-haul flights were more consistent across models.

Flight distance matters too. Emissions estimates diverged more on short-haul flights, which are proportionally less fuel-efficient due to takeoff and climb. Some models account for this more aggressively than others.

Market size and aircraft type also influence results. On trunk routes like New York to London, models often assume similar aircraft with high load factors, leaving less room for disagreement. But on lower-volume routes, model assumptions about aircraft type, seat configuration and fuel burn can diverge sharply.

These compounding factors lead to large disparities in premium cabins on short-haul, low-traffic routes. Values for business class flights on short routes often saw model differences of 30 to 40 per cent. In contrast, long-haul economy results typically differed by less than 15 per cent.

BTN Europe: Did you find that any of the models consistently produced the highest or lowest figures?

Yury Erofeev:?No model is consistently the highest or lowest. It all depends on the flight, cabin class, and each model¡¯s formulas. The six models fall into two categories: global and regional. ICAO¡¯s methodology, GATE4 and TIM are designed for worldwide use, rely on broad and deep datasets, and use more granular inputs. The three regional models ¨C DEFRA, ADEME and CO2 Emissiefactoren ¨C are grounded in local data or policy frameworks and ask for less detailed inputs. Although designed for national use, they are often applied globally.

In general, the median outputs of the three global models were lower than those of the three regional models. Across 25 test segments, such as long-haul economy flights in major markets, the TIM (global) and DEFRA (regional) models most often represented the median value within their respective comparison groups.

BTN Europe: What is the upshot for travel managers and indeed the wider industry?

Scott Gillespie:?Inconsistencies between emissions models undermine the credibility of carbon data across the corporate travel ecosystem. TMCs, OBTs, airlines and OTAs all display flight emissions, but which model and whose standard should they rely on?

While the ISO14083 standard focuses on calculating transport-related emissions, it is just one piece of a broader ecosystem that includes ISO14064, ISO14067, ISO14040/44, the GHG Protocol, the SBTi Transport Guidance, etc. True convergence will require aligning across these frameworks, not just meeting a single standard¡¯s criteria.

With frameworks like the EU¡¯s CSRD demanding clear and consistent Scope 3 emissions reporting, our industry needs its emissions data to converge. Business travel procurement can¡¯t manage what it cannot credibly measure. Having a trusted value for carbon is becoming as vital as one for cost.

BTN Europe: What are the chances of the industry coalescing around a ¡®single source of truth¡¯ or one methodology? TIM is often put forward as the most likely to succeed in this respect.
?
Yury Erofeev:?It seems unlikely that all the current emission models would be reformed to use a single standard and robust methodology. As has been pointed out, some models are designed for use on very basic travel data.
?
Scott and I agree that the TIM model has the greatest potential. It utilises a robust set of data sources and allocation logic, features the best and most transparent documentation, and has a long track record of making substantive and rigorously researched improvements to its model year after year.

BTN Europe: You set out some calls for action in your report. Tell us more about those.
?
Scott Gillespie:?Our analysis sends a clear message: flight emission modelling must improve. Ideally, there will be a single trusted source of flight emission values. While we wait for that lofty goal to be achieved, we call for several changes.
?
Firstly, models should provide transparent documentation about their sources and uses of fuel burn estimates, passenger and cargo load factors, cabin multipliers, and RFI factors. TIM sets an exemplary standard for this.??

Secondly, flight emission modellers should allocate emissions based on cabin mass rather than floor space, taking into account cabin configurations ¨C such as seats per row ¨C and standard weights for seats, passengers and luggage.
?
Lastly, just as financial metrics are audited, carbon emission models should report their audited margins of error annually against operational flight data, including airline-reported fuel burn, cargo loads, passenger load factors, and per-flight emissions.
?
Moreover, we want to see less friction for corporates. TMCs, OBTs and other emissions-displaying platforms should make it easy for customers to choose their preferred emissions model, configure how the values are displayed, and provide meaningful reports to their stakeholders.
?
Beyond that, we hope that influential organisations in the travel and sustainability spaces ¨C such as ITM, GBTA, BT4E, VDR and others ¨C will encourage buyers to use the models they deem to have sufficient credibility for reporting Scope 3.6 emissions.

By adopting these changes, flight emission modellers will make strides in improving the accuracy and credibility of all models. Emissions estimates will converge, and business travellers will make more carbon-responsible travel decisions. Business travel has no shortage of obstacles on the road to net zero; an unreliable carbon odometer should not be one of them.