AI Low-Emission Route Selection in 2026: How It Works

Why Scope 3 Emissions Became a Priority for Corporate Travel

According to a GBTA 2024 study, business travel accounts for up to 85% of indirect greenhouse gas emissions for service-sector companies. Scope 3 category 6 (business travel) came under close scrutiny from regulators and investors after the CSRD directive took effect in the European Union. Companies with turnover exceeding €40 million must disclose the carbon footprint of business travel starting with the 2024 financial year.

The problem is that traditional TMC systems cannot rank options by emissions. An employee sees 12 flights from Moscow to Berlin sorted by price or departure time but doesn't know that a direct A350-900 flight emits 41% less CO₂ than a connection through Istanbul on a Boeing 737-800. This is where artificial intelligence comes in.

How AI Calculates Emissions: Three Data Sources

Modern machine learning algorithms gather information from three layers.

First layer: static aircraft parameters. Engine model, aircraft age, cabin configuration. An Airbus A220-300 from 2022 burns 2.2 litres of fuel per 100 passenger-kilometres, while a Boeing 757-200 from 1995 burns 3.8 litres. The ICAO Aircraft Engine Emissions Databank contains certified figures for 1,200+ engine types.

Second layer: route operational data. Distance, cruise altitude, load factor, weather conditions. Headwinds at flight level FL380 increase fuel consumption by 7-12%. Systems like Thrust Carbon use machine learning to predict actual consumption based on 340 million real flights.

Third layer: allocation coefficients. Business class occupies three times more space, so one seat carries triple the emissions share. The GHG Protocol standard prescribes a multiplier of 1.5-3.0 depending on configuration. AI models dynamically adjust the coefficient by analysing the occupancy of a specific flight over the past 90 days.

Example: flight SU2602 Moscow-Frankfurt on an Airbus A321neo, departure 14:40, direct route, economy class - 73 kg CO₂ per passenger. Alternative via Istanbul on Boeing 737-800 + Airbus A320ceo, connection 2 h 15 min, economy - 122 kg CO₂. Difference: 67%.

Integration into Corporate Booking Tools: What Employees See

Platforms like corporate travel management platform embed a carbon calculator directly into the flight selection interface. Employees see three columns: price, travel time, CO₂ emissions.

A 180-person company, an industrial software manufacturer, implemented such a system in April 2024. Over 8 months, the average carbon footprint per trip dropped from 214 kg to 142 kg CO₂-eq - a 34% decline. Key factor: employees began choosing direct flights over cheap connections when the price difference did not exceed 8%.

The technology works for ground transport too. The Moscow-St. Petersburg Sapsan train generates 12 kg CO₂ per passenger (Russian Railways data, conversion factor 0.028 kg CO₂/kWh for electric traction). A plane on the same route - 87 kg. The AI system automatically suggests the rail option if travel time fits corporate policy.

Nudge Architecture: How to Encourage Green Choices Without Bans

Banning high-emission routes triggers resistance. An Accenture 2023 study showed that strict restrictions reduce business travel satisfaction by 29 NPS points.

A more effective approach is behavioural design. The AI system highlights low-emission options with a green label and text like "Reduces emissions by 48% compared to the average option." Psychological effect: employees perceive the choice as a personal contribution, not an imposed restriction.

Another technique is default sorting by carbon footprint. Users can switch to price sorting with one click, but 63% keep the default setting (data from an internal SAP Concur A/B test, October 2024).

A third tool is a personal carbon budget. Each employee receives an annual limit, for example 1,200 kg CO₂. The system shows the remaining balance after each booking. Gamification: the top 10 employees with the smallest footprint get a mention in the corporate newsletter. A Munich electronics manufacturer reduced emissions by 22% in a year using this mechanism.

Reporting Data: Automatic Scope 3 Category 6 Calculation

Travel managers need not only selection tools but also consolidated reporting. AI platforms aggregate data from all trips and generate a report in GHG Protocol format.

Typical report structure:

• Total emissions for the period (tonnes CO₂-eq)
• Breakdown by transport mode (air / rail / car)
• Breakdown by service class
• Top 10 routes by emission volume
• Comparison with previous period
• Forecast for the current quarter based on booked trips

The system automatically applies radiative forcing coefficients (RFI = 1.9 for aviation, accounting not only for CO₂ but also contrails and nitrogen oxides at altitude). This is a requirement of the ISO 14064-1:2018 standard for corporate carbon reports.

A London financial group, 420 employees, spent 18 hours per month on manual data collection for ESG reports before implementing AI. After automation - 40 minutes for verification and export in CDP format.

Hotel Selection Optimisation: Emissions Don't End at the Airport

Hotels generate 15-25 kg CO₂ per guest night (Cornell Hotel Sustainability Benchmarking Index data, 2023). The range is huge: a chain hotel with LEED Platinum certification - 8 kg, a boutique hotel without energy management systems - 34 kg.

AI systems integrate with Hotel Carbon Measurement Initiative (HCMI) and Travalyst Coalition databases. The algorithm considers:

• Electricity source (renewable / fossil fuel)
• Heat recovery system presence
• Linen and towel replacement policy
• Energy efficiency of heating and air conditioning systems

Employees see the hotel's carbon footprint next to the price. Marriott Bonvoy began displaying this data in its mobile app from March 2024 for 1,800 properties.

Practical case: an Amsterdam IT company, 95 people, changed its hotel selection policy. Instead of the criterion "maximum €150 per night," they introduced "maximum 20 kg CO₂ per night, budget up to €180." Over six months, the average carbon footprint per night fell from 22 to 14 kg, while average cost rose by only €8.

Multimodal Optimisation: When Train + Taxi Beats Plane

AI platforms analyse not only individual segments but the entire door-to-door chain. The algorithm considers:

• Travel time to airport / station
• Check-in and security
• Delay probability (historical punctuality data)
• Transfer from arrival point to office

Munich-Zurich route: plane 1 h, but with transfers and check-in - 4 h 20 min, 98 kg CO₂. Train 4 h 10 min centre to centre, 18 kg CO₂. The system automatically suggests rail as the optimal option.

A Paris consulting firm implemented multimodal optimisation for routes up to 800 km. The share of train trips rose from 12% to 41% over a year. Savings: 340 tonnes CO₂ across 1,200 trips.

Predictive Analytics: Planning Carbon Budget for the Quarter

Machine learning enables emission forecasting based on historical patterns. The model analyses:

• Trip seasonality (exhibitions, conferences, sales peaks)
• Correlation with business metrics (revenue, number of deals)
• Changes in corporate policy

The system outputs a forecast: "At the current booking pace, the quarterly limit of 45 tonnes CO₂ will be exceeded by 18%. We recommend limiting business class travel or replacing 6 intercontinental flights with video conferences."

A Basel pharmaceutical company uses such forecasts to align carbon budgets with financial ones. If the forecast shows an overage, the travel manager initiates a discussion with the CFO 6 weeks before quarter-end, not after the fact.

Emission Offsetting: Automatic Carbon Credit Purchase

Some AI platforms integrate with carbon offset registries (Gold Standard, Verra VCS). After each booking, the system offers to offset emissions.

Important nuance: not all projects are equally effective. The algorithm filters projects by criteria:

• Additionality (the project would not have been implemented without credit sales)
• Permanence (carbon sequestered for at least 100 years)
• No leakage (the project doesn't shift emissions to another location)
• Third-party verification

The system prioritises Direct Air Capture projects and reforestation with long-term monitoring. Cost of offsetting 1 tonne - €20-45 depending on the project.

A Stockholm retailer, 310 employees, implemented automatic offsetting for all air travel. Annual offset budget - €14,200 with total emissions of 520 tonnes. Employees see an offset certificate in the booking confirmation.

Practical Steps for Implementing AI Route Optimisation

Step 1: audit current emissions. Collect data on all trips over the past 12 months. If you use a TMC, request a report with aircraft type details. If bookings go through corporate cards, export transactions and classify manually. Goal - obtain a baseline in tonnes CO₂-eq.

Step 2: choose a platform with a carbon calculator. Check whether the system uses ICAO data or proprietary methodology. Request a white paper describing the algorithm. Ensure the platform accounts for RFI for aviation and supports integration with your expense management system.

Step 3: pilot launch with one department. Select a department with high trip frequency (sales, consulting). Give access to the new system but don't impose mandatory restrictions. Measure the change in average carbon footprint over 3 months. Collect feedback: how convenient is the interface, does displaying emissions influence choice.

Step 4: update corporate policy. Specify priority for low-emission options if the price difference doesn't exceed a set threshold (e.g. 10%). Introduce mandatory justification for bookings with emissions above 200 kg CO₂ per segment. Set a quarterly emission limit for each division.

Step 5: train employees. Conduct a 30-minute webinar: how to read carbon footprint, why direct flights are preferable, how offsetting works. Record a short video (3-4 minutes) and post it in the corporate knowledge base. Answer the question "Why does this matter to the company" through the lens of ESG goals and investor expectations.

Step 6: monthly monitoring. Set up a dashboard with key metrics: total emissions, emissions per trip, share of low-emission bookings, savings compared to baseline. Discuss results at monthly meetings with department heads.

Implementation Barriers: What to Expect in Practice

The main problem is data quality. Not all airlines disclose aircraft type at booking. Aircraft substitution 24 hours before departure makes forecasts inaccurate. GDS systems (Amadeus, Sabre, Travelport) began transmitting emission data only from 2023, and coverage is currently 70-80% of flights.

The second barrier is employee resistance. A Deloitte 2024 survey showed that 38% of business travellers consider carbon footprint "not their responsibility." Behaviour change requires not only technology but cultural shift.

The third is implementation cost. Integrating a carbon calculator into an existing TMC system costs €15,000-40,000 depending on complexity. Subscription to a platform with a ready AI module - €3-8 per active traveller per month. ROI occurs when emission reduction converts to reputational capital or avoids carbon tax.

Regulatory Context: Why Delay Is No Longer an Option

The CSRD directive requires companies to disclose not only direct emissions (Scope 1 and 2) but also indirect ones (Scope 3). Business travel falls into category 6. First reports under new standards must be published in 2025 for the 2024 financial year.

The UK introduced similar requirements through Streamlined Energy and Carbon Reporting (SECR) back in 2019. Companies with turnover exceeding £36 million must report transport emissions.

China included aviation emissions in its national emissions trading system (ETS) from January 2024. The cost per tonne of CO₂ on the Chinese exchange - ¥80-100 (€10-13). Companies exceeding the limit buy quotas from those who reduced emissions.

Australia plans to introduce mandatory Scope 3 reporting for public companies from July 2025. Penalty for non-disclosure - up to AUD 1.1 million.

Ignoring these requirements leads not only to fines but reputational losses. Institutional investors use ESG ratings in decision-making. A company with a high carbon footprint and no reduction plan receives a lower rating from MSCI, CDP, Sustainalytics.

Future of the Technology: What Will Change by 2027

Machine learning is moving toward predictive optimisation. Instead of showing emissions for an already-selected flight, the system will suggest alternative dates and routes with minimal footprint.

Example: an employee plans a meeting in Milan on 15 March. AI analyses flight schedules for ±3 days and discovers that on 14 March an A350 flies with 68 kg emissions, while on 15 March only a 737 MAX with 94 kg. The system suggests moving the meeting a day earlier and automatically sends a calendar invite.

Another direction is integration with video conferencing systems. The algorithm assesses whether a trip can be replaced by a virtual meeting based on meeting type (negotiations, training, audit) and historical effectiveness of online format for that category.

The third trend is blockchain verification of carbon credits. Smart contracts automatically retire offsets after each flight, and employees see a record in the distributed ledger. This increases transparency and eliminates double counting.

The industry is moving toward standardisation. The Travalyst Coalition (founded by Prince Harry, partners: Booking.com, Skyscanner, Trip.com, Visa) is developing a unified emission calculation methodology. Goal - to make figures in different systems match rather than differ by 30-40%, as they do now.

FAQ

How does AI calculate CO₂ emissions for air travel?

Algorithms analyse three data layers: aircraft and engine type (from the ICAO database), route operational parameters (distance, altitude, weather conditions, load factor), and allocation coefficients by service class. Machine learning systems use historical data from millions of flights to predict actual fuel consumption with 7-12% accuracy.

How accurate are carbon calculators in booking systems?

Accuracy depends on data source. Platforms using ICAO methodology and accounting for radiative forcing (RFI = 1.9) give ±15% margin of error. The main problem is aircraft type substitution before departure, which happens in 8-12% of cases. GDS systems (Amadeus, Sabre) began transmitting emission data from 2023, coverage is currently 70-80% of flights.

Are companies required to report business travel emissions?

In the EU, the CSRD directive requires disclosure of Scope 3 category 6 (business travel) for companies with turnover exceeding €40 million from the 2024 financial year. The UK introduced the requirement through SECR in 2019 (threshold £36 million). Australia plans mandatory Scope 3 reporting from July 2025. Penalties for non-disclosure reach AUD 1.1 million.

How much does implementing an AI system for low-emission route selection cost?

Integrating a carbon calculator into an existing TMC system costs €15,000-40,000. Subscription to a platform with a ready AI module - €3-8 per active traveller per month. A 200-employee company with 40 trips per month will pay €120-320 monthly. ROI occurs in 8-14 months through reputational capital and regulatory readiness.

Can air travel be replaced by trains to reduce emissions?

For routes up to 800 km, trains generate 4-7 times less CO₂. Moscow-St. Petersburg: train 12 kg, plane 87 kg. Munich-Zurich: train 18 kg, plane 98 kg. AI platforms with multimodal optimisation automatically suggest rail when door-to-door time is comparable. A Paris consulting firm increased train trip share from 12% to 41%, cutting 340 tonnes CO₂ per year.

How does carbon offset compensation work through AI platforms?

After booking, the system offers to purchase carbon credits from projects certified by Gold Standard or Verra VCS. The algorithm filters projects by additionality, permanence (CO₂ sequestration for at least 100 years), and third-party verification criteria. Priority goes to Direct Air Capture and reforestation with monitoring. Cost of offsetting 1 tonne - €20-45.