A small haulier or forwarder can use AI to organise fuel, distance and load records into a per-shipment emissions sheet, but must select the method the tender asks for first. Calculate with documented factors, show whether fuel was measured or estimated, allocate shared trips consistently, and have a person verify units and totals before submitting.
A single “carbon per shipment” number can mean several different things. A customer may ask for tailpipe CO2, well-to-wheel CO2e, a particular framework, or a total for a contract period. The same 100 gallons of fuel cannot satisfy all those requests with one unexplained multiplier. Put the reporting boundary, factor source and data quality beside every result.
Read the tender field before building a formula
Copy the customer's exact request into a small requirements table. Record whether it asks for mass of CO2 or CO2e, tailpipe or well-to-wheel boundary, shipment or contract period, measured or estimated activity data, distance treatment, allocation to shared loads, and any named method. Ask the customer to clarify a vague field before claiming compliance. The Smart Freight Centre's GLEC Framework is the method on which ISO 14083, the international standard for transport emissions, was built; the current edition listed there is version 3.3, published in October 2026. If a tender names GLEC, ask which version it expects and record it on the sheet. Naming that method in a tender does not make a rough spreadsheet automatically compliant with it.
Define the reporting unit too. “kg per shipment” differs from “kg per tonne-kilometre”, an intensity measure that divides emissions by cargo mass and transport distance. If a tender asks for both, report both with the underlying tonne-kilometres. A small shipment can have high intensity because it occupies a vehicle on a route with empty running. Do not compare intensity across very different services without checking the allocation method.
A forwarder may not operate the vehicles. In that case, request carrier-provided activity and emissions data, factor sources and method boundaries. If a carrier supplies only a headline total, ask what loads and legs it covers and how it allocated shared vehicle emissions. Record missing carriers as a gap, not zero. If an estimate is permitted, label it and keep its assumptions traceable. The tender response should say what proportion uses primary operational data and what proportion is estimated.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowDesign the worksheet so a reviewer can trace one load
Use one row per shipment leg, with a stable shipment ID and leg ID. Add date, vehicle or carrier reference, service type, loaded distance, empty distance where attributed, cargo weight, whether the trip is dedicated or shared, fuel quantity, fuel unit, measured or estimated flag, factor source, factor version, emissions boundary, allocation rule and calculated kg. Keep the original fuel receipt or telematics export reference in a separate evidence column. A route can have several legs, so a single shipment may appear on several rows and then be summed by ID.
Maintain a factor table separate from shipment rows. It should contain fuel type, factor value, units, boundary, source URL or document title, publication year, date checked and who approved it. Lock or protect the factor cells in the working copy so a copied formula cannot overwrite them. Do not ask the model to fill the factor table from memory. Check the official source each reporting cycle and when a customer specifies a new method. In a real tender, a factor that covers only combustion should not be used for a field requiring the full fuel chain.
Keep units explicit. A fuel record in gallons and a factor in kilograms per litre will give a wrong answer if multiplied directly. Distance in miles and a factor per kilometre need conversion. A load weight in pounds and an intensity factor per tonne-kilometre need two conversions. Add a “unit check” column that says whether the activity unit matches the factor. The assistant can spot suspicious joins, but the spreadsheet should display converted values and formulas so a reviewer can reproduce them.
Link these rows to the dispatch system's completed loads and to fuel purchase records. The fuel used on a vehicle during a week may cover several customer loads, positioning miles and engine idling. If you allocate it, write the allocation rule and ensure all shipment shares add up to the total fuel being allocated. Assigning the full vehicle fuel to every load is a common spreadsheet join error that inflates the tender total.
A measured-fuel example with a narrow boundary
Suppose, illustratively, a haulier records 100 gallons of diesel for a dedicated completed trip and the tender explicitly asks for tailpipe CO2 from fuel combustion. One official government calculation reference, last updated in August 2026, gives 10,180 grams of CO2 per gallon of diesel, on the assumption that all of the fuel's carbon becomes CO2. It counts combustion CO2 only: no methane, nitrous oxide or upstream fuel production. Convert that to 10.18 kg per gallon. The calculation is 100 × 10.18 = 1,018 kg of tailpipe CO2 for the trip. Do not label it as well-to-wheel CO2e; that wider boundary needs its own approved method and factors.
Assume the dedicated trip serves one shipment, ID S-701. The worksheet records fuel 100 gallons, factor 10.18 kg CO2 per gallon, factor boundary tailpipe, calculated emissions 1,018 kg CO2 and allocation share 100%. The dispatch record shows 500 loaded miles and 40 positioning miles. Those distances help explain the operation and check plausibility, but they are not multiplied into the fuel-based result. Multiplying by distance again would double count activity.
For a simple intensity calculation, suppose S-701 carries 10 tonnes over 500 loaded miles. Do not insert “500” into a tonne-kilometre formula without converting the distance and checking whether the chosen method includes empty distance in the allocation. If the tender wants a per-shipment mass only, submit the 1,018 kg figure with its boundary and data quality. Adding an intensity that was not requested can confuse the response and expose an unreviewed conversion.
Now suppose the same 100-gallon trip carried two customers' consignments rather than one. Allocating 1,018 kg to both would claim 2,036 kg from 100 gallons. If the approved rule allocates 60% to shipment S-701 and 40% to S-702, their illustrative tailpipe allocations are 610.8 kg and 407.2 kg; together they remain 1,018 kg. The choice of 60/40 requires a defensible basis, such as occupied capacity or a method required by the tender. A model must not choose shares merely because one customer invoice is larger.
Make the assistant explain the method and the gaps
ChatGPT or Claude can help create a sheet schema, check for inconsistent units, identify duplicate shipment IDs and draft plain-English method notes. Give it the tender requirements and the approved factor table. Ask it to flag any row that lacks fuel, distance, weight or allocation evidence. It should return a list of questions, not a confident estimate for every blank. Use a business-approved account for customer and carrier information, and remove names or addresses that the arithmetic does not require.
Review this shipment emissions worksheet for a tender.
Tender request: [copy exact boundary, unit and named method]
Approved factor table: [paste source, year, value and units]
Check each row for fuel-unit mismatch, missing activity data,
duplicate shipment/leg IDs, allocation shares that do not total 100%,
and factors that use the wrong reporting boundary.
Recalculate only from supplied values. Return a row-level exception list,
the formula for each result, a total by data-quality class, and a short
method note. Do not invent a factor or silently fill missing fuel.An illustrative AI response might say “all rows use the diesel factor; total 12,430 kg CO2e”. The reviewer should correct two things: the source factor is tailpipe CO2, so the label “CO2e” may be wrong for the tender, and “all rows” is unproved until fuel type and units have been checked individually. Ask for a list of every row ID and factor selected. A useful assistant exposes its working; a fluent summary cannot stand in for the evidence table.
Use a separate method note in the tender. For example: “The following illustrative estimate uses recorded diesel fuel for dedicated trips, a published tailpipe CO2 factor, and a documented 60/40 capacity allocation for one shared trip. It excludes fuel-chain emissions and is not presented as a full ISO 14083 result.” Adapt the language to the actual method and do not submit a caveat that contradicts the tender's requirements. If the tender requires a full framework, obtain the correct methodology and factors before responding.
Fuel cards, reefers and empty returns that skew the figure
A fuel card covers two vehicles: the invoice total is 400 gallons, but 70 gallons belong to a vehicle outside the tender. Do not divide the full 400 gallons across tender loads. Match transaction IDs to vehicle IDs first. If 330 gallons remain in scope, reconcile that subtotal to the fuel records and disclose any transactions that cannot be assigned.
A refrigerated load uses auxiliary fuel: the road vehicle's propulsion fuel may not include fuel used by refrigeration equipment. If the tender boundary includes it, capture the separate consumption or an approved estimate and factor. A model cannot infer it from the commodity name. The worksheet should have a separate activity field and note if the data is unavailable.
A shared route has an empty return: a one-way customer shipment may be assigned some empty running under the chosen allocation method. Ignoring 80 return miles can understate its footprint; adding all 80 to every shipment can overstate the total. Record the route and allocation basis. The backhaul tutorial addresses the operational decision to fill empty capacity, which is separate from reporting an emissions number.
A contractor supplies only a distance estimate: 420 planned miles is not the same as measured fuel or actual route distance. Use the tender's permitted distance-based method and a verified intensity factor appropriate to the service, or seek better carrier data. Mark the row estimated and keep it separate from measured-fuel rows. Do not multiply a tailpipe fuel factor by miles; its denominator is gallons.
A revised delivery adds a second leg: the original shipment estimate covers one leg, then a failed delivery creates a redelivery. Version the shipment and add the extra leg if the reporting boundary requires it. Do not overwrite the original distance with the new total without preserving what changed. This helps explain a higher figure to the customer and reconcile it to dispatch records.
A spreadsheet uses the wrong factor year: a tender asks for a named reporting year, while the factor tab has a different version. The arithmetic may be flawless yet the submission may fail the stated method. Record factor year and date checked in the output. Ask the tender owner whether a newer factor is required; do not silently swap versions because a chatbot finds a more recent web page.
Resolve the allocation rule before producing a total
A shared vehicle creates a denominator problem. Suppose, illustratively, a trip carries six pallets for Customer A and four for Customer B. An allocation by pallet spaces gives shares of 60% and 40%, provided the pallets occupy comparable space and the method is permitted. If Customer B's four pallets are unusually heavy and drive a different vehicle choice, a weight-based or capacity-based allocation may be more defensible. The worksheet should state why the chosen rule reflects this operation. An AI-generated percentage without a documented denominator is not evidence.
Check the allocation at three levels. For each vehicle trip, shares across consignments must total 100%. For each shipment with multiple legs, the leg emissions must sum to the shipment result. For the tender period, the included shipment results plus any explicitly excluded operational fuel should reconcile to the fuel-based trip total. If a route contains an empty repositioning leg, decide under the tender method whether to attribute it to the preceding deliveries, the next pickup or a broader operating pool. Record that decision before calculating, because changing it can move emissions between customers while leaving fleet fuel unchanged.
For an illustrative audit, a vehicle trip has 100 gallons and three consignments. The approved capacity shares are 50%, 30% and 20%. Using the narrow tailpipe example factor, the trip produces 1,018 kg CO2; shipment allocations are 509 kg, 305.4 kg and 203.6 kg. Their sum is 1,018 kg. If the sheet instead reports 509, 305.4 and 305.4, the allocation percentages total 110% and emissions are overstated by 101.8 kg. This is an arithmetic check that a spreadsheet can enforce without AI.
When distance is the available activity measure, distinguish planned from actual distance. A route planner's 400 miles may omit a diversion to a second pickup, waiting movements or a different road taken by the driver. Do not “correct” the distance by a made-up percentage. Ask the carrier or dispatch team for actual distance if the tender demands it; otherwise disclose the planning basis and apply the named method's allowed adjustment, if any. If a customer asks for a per-tonne-kilometre figure, confirm whether the cargo mass means billed weight, actual weight or an allocation weight and document the conversion to kilometres.
Double-counting can also appear when a subcontractor provides a shipment emissions figure and the forwarder adds a separate estimate for the same transport leg. Give every leg one source type and one included result. If carrier data covers only the line-haul leg, the forwarder may need to estimate a separate pickup leg, but label the two legs and their methods. A reviewer should be able to answer “which operation generated this kilogram figure?” for every row. A total formed by stacking all available numbers is not a trustworthy total.
The emissions sheet and the commercial sheet answer different questions. A low-emission route may cost more, or a low-cost load may use more empty running. If the owner wants to evaluate both, join the final emissions result to the same shipment ID used in the lane margin analysis, and keep the two calculations separate. For example, a tender option might add a $75 carrier charge while reducing an approved emissions estimate; show both changes with their sources. Do not let a model describe the trade-off as a guaranteed saving or environmental benefit before the carrier confirms the route and fuel data.
Put a short data-quality statement beside the tender figure. “Of 40 shipment legs, 28 use measured fuel allocated from vehicle trips, eight use carrier-provided values with stated boundaries, and four use distance-based estimates” is an illustrative disclosure. It helps a customer see where the estimate is strong and where better data is needed. It also gives the haulier a concrete improvement list for the next tender: obtain fuel by vehicle, request carrier method notes, and record actual route legs. Such a statement is more useful than rounding the total to an implausibly precise decimal.
Reconcile emissions before sending the tender
Count every shipment and leg in the tender period, then compare with completed dispatch records. Sum vehicle fuel by source document, sum fuel allocated to included shipments, and show the difference. Some fuel may legitimately sit outside the contract or outside the reporting boundary; explain it. Sum emissions by measured, carrier-provided and estimated data classes. A total without those classes can hide that half the work rests on assumptions.
Recalculate three rows by hand: one dedicated trip, one shared trip and one estimated trip. Check units, factor boundary, allocation and rounding. Then ask another staff member to trace one reported shipment from the tender figure back to a fuel record, route and approved factor. If they cannot reproduce the figure, the worksheet is not ready. The separate shipment dashboard tutorial can help present checked measures, but the tender method note should remain attached to the numbers.
Review the final wording as carefully as the calculation. State the period, method, emissions boundary, activity source, factor source and estimate share. Avoid “certified”, “compliant” or “carbon neutral” unless an appropriate external process has actually established those claims. A small haulier can submit a useful, transparent estimate without overstating what its records prove. That is more credible than a precise-looking number whose boundary cannot be explained.
Tender emissions questions worth clarifying
Can a chatbot supply the emissions factor?
Use a current, official factor source that matches the fuel, unit, year, geography and reporting boundary the tender requires. Record the source and version in your sheet. Ask AI to check units and formulas, but do not accept a factor it invents or recalls without a verifiable reference.
What if the haulier knows distance but not fuel for one load?
Label the load as an estimate and use the method the tender permits, such as a documented intensity factor or fuel-use estimate. State the vehicle class, route distance, load share and factor source. Keep estimated rows separate from measured fuel rows and report how much of the tender total relies on estimates.
Is tailpipe CO2 the same as well-to-wheel CO2e?
No. Tailpipe CO2 describes direct carbon dioxide from burning fuel. Well-to-wheel CO2e includes a wider fuel-chain boundary and other greenhouse gases according to the chosen method. Do not label a tailpipe-only number as a full well-to-wheel result. Ask the customer which boundary and standard its tender requires.
Further reads
- Can AI Read Bills of Lading and Fill Your TMS? An Accuracy Test — Check the underlying shipment record.
- AI for Freight Forwarding: What It Does Well and What It Doesn't — Set boundaries for AI analysis.
- Freight Rate Sheets: AI Extracts Carrier Rates From PDFs — Keep source data structured.
- Factory Energy Audit With AI: Find Peak Loads and Idle Waste — Structure a separate energy audit.
- AI ESG Report Drafting for Sustainability Consultants — From a client's data table to an ESG narrative: a claim register, number tracing, wording tests for green claims and what assurance providers will still ask.
- Detention Pay: AI Tracks Wait Times and Invoices Brokers — A carrier-side workflow: capture timestamps at the dock, test them against your contract terms, then invoice the broker with AI-drafted emails.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Smart Freight Centre GLEC Framework page (version 3.3, October 2026; describes GLEC as the foundation of ISO 14083); Environmental Protection Agency, Greenhouse Gas Equivalencies Calculator: Calculations and References (updated 4 August 2026; 10,180 g CO2 per gallon of diesel, combustion CO2 only); both read 5 November 2026. All tender figures and fuel allocations are illustrative.