The Complete Guide to GHG Protocol Reporting with AI

This guide is for the person who just got told, in a Tuesday afternoon meeting, that the company needs a GHG Protocol inventory for FY2025 — Scopes 1, 2, and 3 — and that it is now their problem. You run finance, or operations, or HSE. You are not a carbon accountant. You have an ERP, a utility bill folder, and a list of 180 suppliers.

By the end of this article you will have walked through a full inventory for a fictional-but-realistic 150-employee industrial coatings manufacturer, with real emission factors and real numbers. You will also know where the manual version stops being worth your time.

Prerequisites

Before you start, pull the following. If you can't find any of these, that's the first finding of the inventory.

• Legal structure. A list of every entity the parent consolidates, with ownership percentages. Finance has this.
• Utility bills. Twelve months of electricity, natural gas, district heat. Invoices, not summaries.
• Fleet and fuel. Vehicle list, fuel card statements, any on-site diesel for forklifts or gensets.
• Refrigerant logs. HVAC and process cooling service records — the "2.4 kg R-410A topped up" lines.
• Procurement export. The AP ledger for the fiscal year, with vendor name, amount, and ideally a commodity code.
• Business travel. Expensed flights and hotels. If you have a TMC (Amex GBT, BCD, whatever), ask them for the emissions export and don't be surprised when it's in kg instead of tonnes.
• Employee commute survey. You don't have one yet. That's fine.

The GHG Protocol Corporate Standard (WRI/WBCSD) is the methodology. The Scope 3 Standard adds 15 categories on top — eight upstream, seven downstream. You will not report all 15. Nobody does, and the standard doesn't ask you to.

The worked example: Dalton Coatings Ltd.

Dalton Coatings is a UK-based industrial coatings manufacturer. 150 employees across one factory in Manchester and a small R&D office in Bristol. FY2025 revenue £38M. They import solvents and pigments, blend and package in Manchester, and ship finished product to construction and automotive B2B customers across the UK and EU. They've never produced a GHG inventory.

The controller pulled the files last week. Let's do the inventory.

Step 1. Pick the organizational boundary

The GHG Protocol gives you two choices, and they are not equivalent:

• Equity share. You report emissions from each operation in proportion to your ownership stake. A 40% joint venture? You take 40% of its emissions.
• Operational control. You report 100% of emissions from operations where you set the operating policies. The 40% JV where you run the plant? All of it.

Dalton wholly owns the Manchester factory and leases the Bristol office. There are no JVs, no minority stakes, no franchised operations. Under either boundary the answer is the same: Manchester and Bristol, 100%. They pick operational control, because that's what 85% of corporates pick and it will align with CDP and CSRD later without a re-mapping exercise.

Write this decision down. The boundary choice goes on page one of the inventory report, and an auditor will ask why.

Manual way: Half-hour conversation between sustainability lead and CFO, a paragraph in the methodology appendix. Fine, nothing to automate.

Step 2. Scope 1 — what you burn and what leaks

Scope 1 is direct emissions from sources the company owns or controls. For Dalton there are three:

Fleet. Sales reps drive 12 company-leased diesel vehicles. Fuel cards show 48,200 litres of diesel pumped in FY2025.

Emission factor from UK DEFRA 2025: 2.51 kgCO2e per litre of diesel (well-to-wheel, but use tank-to-wheel if you want to report WTW separately under Scope 3 cat 3).

48,200 L × 2.51 kgCO2e/L = 121,000 kgCO2e = 121 tCO2e

Boilers. The Manchester plant runs two natural gas boilers for process heat and building heat. Twelve months of British Gas invoices total 8,940,000 kWh of gas (gross calorific value).

DEFRA 2025 factor for natural gas: 0.18290 kgCO2e/kWh (gross CV).

8,940,000 kWh × 0.18290 = 1,635,000 kgCO2e = 1,635 tCO2e

Fugitive HFCs. The HVAC service log for 2025 shows two top-ups of R-410A totalling 3.8 kg and one top-up of R-134a at 1.1 kg for a chiller.

GWP (AR5, 100-year): R-410A = 1,924; R-134a = 1,300.

3.8 × 1,924 = 7,311 kgCO2e (R-410A) 1.1 × 1,300 = 1,430 kgCO2e (R-134a) Total fugitive: 8,741 kgCO2e = 8.7 tCO2e

Scope 1 total: 1,765 tCO2e.

Manual way: A sustainability analyst reads 12 invoices, pulls the DEFRA spreadsheet, copies factors into an inventory workbook, writes the formulas, checks the units. About a day and a half, including chasing down the R-410A kg value that someone wrote on a paper work order.

AI way: In Formist, an AI-powered compliance platform built by WeCarbon that works like a knowledgeable colleague sitting next to you — you talk to it, upload your documents, and it fills out the sustainability forms for you. Upload the 12 gas invoices, the fuel card statement CSV, and the HVAC service PDF. Formist extracts litres, kWh, and refrigerant kg, matches each to the correct DEFRA or AR5 factor, and produces the three Scope 1 line items with source citations. About twenty minutes, most of which is you confirming that yes, the HVAC service on 14 March was R-410A and not R-32.

Step 3. Scope 2 — purchased electricity, both ways

The GHG Protocol Scope 2 Guidance (2015) requires dual reporting: location-based and market-based. You report both. Always.

Dalton's Manchester plant consumed 2,340 MWh of purchased electricity in FY2025, per the EDF Energy invoices. The Bristol office consumed 48 MWh. Total: 2,388 MWh.

Location-based. Use the grid average for the country. UK national grid factor for 2025: 0.233 kgCO2e/kWh (DEFRA/BEIS published figure).

2,388,000 kWh × 0.233 = 556,404 kgCO2e ≈ 556 tCO2e

Market-based. Use the emission factor of the specific contract. Dalton's Manchester supply is on EDF's "Business Zero" tariff, backed by REGO certificates, so the supplier-specific factor is 0 kgCO2e/kWh for Manchester. Bristol is on a standard commercial tariff with a residual-mix factor of 0.412 kgCO2e/kWh (Green-e / AIB residual mix for the UK).

2,340,000 × 0 + 48,000 × 0.412 = 19,776 kgCO2e ≈ 20 tCO2e

Note the gap: 556 tCO2e vs 20 tCO2e. That's not an accounting trick — it's the entire point of market-based reporting. It credits Dalton for buying a renewable tariff. Under CDP and SBTi, the market-based number is what gets compared to your target.

Scope 2 total: 556 tCO2e (location) / 20 tCO2e (market).

Manual way: Find DEFRA's location-based factor. Find EDF's residual-mix disclosure (good luck — it's usually a PDF buried on the "fuel mix" page). Find the AIB European residual mix. Apply both. Write the dual-reporting note. Half a day.

AI way: Formist pulls the kWh from the invoices, identifies the tariff name on the invoice ("Business Zero — REGO backed"), looks up the supplier-specific factor, and produces both numbers in a single card labelled "Scope 2: Purchased Electricity" with both methods populated and the REGO certificates attached as evidence. About ten minutes.

Step 4. Scope 3 — pick what's material, then do it properly

Dalton has 15 Scope 3 categories available. They do not need 15. The GHG Protocol's own guidance says: report the categories that are material to your business. For a coatings manufacturer, the obvious candidates are Category 1 (purchased goods and services), Category 4 (upstream transportation), Category 6 (business travel), and Category 11 (use of sold products — which requires a product-level LCA that is not going to happen this year).

They scope in 1, 4, 6. Note the rest as "screened, not material" with a one-line justification.

Category 1 — Purchased goods and services

This is almost always the biggest Scope 3 number, and the hardest. There are two methods:

Spend-based. Multiply supplier spend by an emission factor per dollar or pound of that spend category. Fast, rough, defensible as a starting point.

Activity-based. Get actual product-level emissions data from the supplier (a PCF — product carbon footprint). Accurate, slow, requires supplier cooperation.

Best practice, and what SBTi and CDP now expect: start spend-based, then replace your top 10 suppliers with activity-based data in years 2 and 3.

Dalton's FY2025 procurement ledger: £18.2M of goods and services across 183 vendors. The top 15 vendors account for 78% of spend. Categorised (using UK ONS SIC codes mapped to EXIOBASE/US EEIO factors):

Spend-based Category 1: 9,680 tCO2e. Already roughly 4x the combined Scope 1+2 market-based total. Welcome to Scope 3.

Next year, Dalton will ask their top 3 solvent suppliers for PCFs and replace the spend-based rows with activity-based numbers. The total will change — typically down, because spend-based factors are deliberately conservative — but the reporting will get more defensible.

Manual way: Map 183 vendors to categories. Find the right emission factor database (EXIOBASE, US EEIO, DEFRA indirect factors, CEDA — pick one and stick with it). Build the multiplication table. Write the methodology note explaining why spend-based is the right first step. Two full days, minimum, and the vendor categorisation is where people make errors they later find in the audit.

AI way: Upload the AP ledger. Formist classifies each vendor line into EXIOBASE categories, applies the factor, flags the 12 vendors it's uncertain about for you to confirm, and produces the table above with full audit trail — vendor, amount, assigned category, factor source, factor value, resulting tCO2e. An hour of work, mostly reviewing the 12 edge cases.

Category 4 — Upstream transportation and distribution

Only the legs Dalton pays for. Finished product goes out on customer-paid freight — that's Category 9, downstream, and Dalton scopes it out as not material. Upstream: solvent and pigment deliveries into the Manchester plant.

Freight invoices show 1,240,000 tonne-km of road freight (articulated truck, average load). DEFRA 2025 factor for HGV average laden, diesel: 0.10585 kgCO2e/tonne-km.

1,240,000 × 0.10585 = 131,254 kgCO2e ≈ 131 tCO2e

Cat 4: 131 tCO2e.

Category 6 — Business travel

Amex GBT export for FY2025: 140 flights totalling 412,000 passenger-km, and 290 hotel nights.

Flights, weighted by cabin and haul (DEFRA 2025): blended factor 0.158 kgCO2e/passenger-km including RF uplift.

412,000 × 0.158 = 65,100 kgCO2e ≈ 65 tCO2e

Hotels, DEFRA UK-average: 10.4 kgCO2e per room-night.

290 × 10.4 = 3,016 kgCO2e ≈ 3 tCO2e

Cat 6: 68 tCO2e.

The inventory

Putting the worked example together:

• Scope 1: 1,765 tCO2e (fleet 121, boilers 1,635, fugitive 9)
• Scope 2 (market-based): 20 tCO2e
• Scope 2 (location-based): 556 tCO2e (disclosed in parallel)
• Scope 3 Cat 1: 9,680 tCO2e
• Scope 3 Cat 4: 131 tCO2e
• Scope 3 Cat 6: 68 tCO2e

Reported total (Scope 1 + 2 market + Scope 3 in scope): 11,664 tCO2e.

About 83% of that is Category 1. This is typical for a manufacturer and it tells Dalton exactly where to push next year: supplier engagement on the solvents line.

Common mistakes

Mixing gross and net calorific value on gas. UK invoices report gross CV, DEFRA publishes both, Europe often quotes net. Using a net-CV factor on a gross-CV kWh reading under-reports by about 10%. Match them explicitly.

Forgetting fugitive refrigerants. Small numbers, easy to overlook, and auditors check precisely because they're easy to overlook. Pull the HVAC service log before you close Scope 1.

Single-method Scope 2. Reporting only market-based because it's a better number, or only location-based because the finance team asked. The Protocol requires both. Omitting one is a qualification in assurance.

Double-counting Scope 3. The logistics services line in Cat 1 (spend-based) and the freight tonne-km in Cat 4 (activity-based) cover the same inbound moves. Pick one method per emission source. Dalton's fix: drop logistics out of Cat 1 next year once the Cat 4 activity data is solid.

Using last year's grid factor. The UK factor fell from 0.257 in 2022 to 0.233 in 2025. If you're benchmarking reductions, use the year-specific factor. Otherwise you'll report a 10% "reduction" that is just the grid decarbonising without you.

When to automate

The Scope 1 section of this inventory — fleet, boilers, fugitive — is two pages of invoices and one spreadsheet. Automate it if you want to, but a competent analyst can do it in a day and it will be right.

The Scope 2 dual reporting is where mistakes start creeping in, specifically around residual-mix factors and REGO certificate handling. Worth automating.

The Scope 3 Category 1 work is where the manual method breaks down. Mapping 183 vendors to emission factors, applying them consistently across years, preserving the reasoning for audit — that's exactly the kind of structured-but-tedious work that a Formist AI agent handles better than a stressed analyst at 6pm on a Thursday. You upload the AP ledger, it produces the categorised table, you review and sign off. The same Formist workspace also feeds the numbers into your CDP questionnaire, your CSRD E1 datapoints, and your SBTi progress tracker without re-entry, because it's one data model underneath.

Dalton's first inventory — done manually — will take a sustainability analyst roughly three weeks and produce a workbook only they can read. The same inventory in Formist takes most of a working day and produces a reviewable card-based record with source documents linked to every number. Both get to the same answer. Only one of them is repeatable without redoing the whole thing next year.

The GHG Protocol is a methodology, not a piece of software. You can do it in Excel. You probably shouldn't, past year one.

Formist is built by WeCarbon, a climate-tech company with offices in Paris, Shanghai, and Dubai. It supports the GHG Protocol Corporate and Scope 3 Standards, CBAM, CSRD/ESRS, EU Taxonomy, CDP, ISSB, SBTi, and 15+ other sustainability frameworks.