Supplier-Specific Emission Factors in Food Scope 3: Using Supplier, Regenerative and Sourcing-Region Data Without Failing Assurance

Why now: Regulation (EU) 2023/1115 (EUDR) applies to large and medium operators from 30 December 2026, and the GHG Protocol Land Sector and Removals Standard becomes mandatory on 1 January 2027. Your farm-level data is about to be checked twice over.

Your agronomy team has funded cover cropping across a wheat region for three seasons. A dairy supplier has just emailed a product carbon footprint in Excel. Your inventory still runs on a generic database factor that knows nothing about either. And the auditor arriving in Q1 wants to know where every one of your 400 custom factors came from, including the ones a consultant added in 2021.

Sustainability leaders in food and consumer health keep asking how to swap secondary factors for better data without creating an assurance problem. Here is a practical answer, based on the standards as of November 2026.

Why do database emission factors fall short for food Scope 3?

Database averages describe a typical product in a typical place, not your supply chain. They cannot reflect a specific processing step, a specific sourcing region, or the regenerative practices you pay for, so any progress you fund stays invisible in your inventory.

The GHG Protocol's Technical Guidance for Calculating Scope 3 Emissions (version 1.0, 2013) ranks four methods for purchased goods, from most to least specific: supplier-specific, hybrid, average-data and spend-based. It defines primary data as "data from specific activities within a company's value chain" and secondary data as everything else, including industry averages and proxies.

Note the honest trade-off the same guidance flags: supplier data "may actually be less accurate than industry-average data for a particular product" when allocation is uncertain. More specific is not automatically better. It has to be better evidenced.

What does the GHG Protocol require you to disclose about supplier data?

You must disclose, for each Scope 3 category, the share of emissions calculated with supplier data and describe your data quality. That disclosure is exactly what assurance providers test.

The Corporate Value Chain (Scope 3) Standard requires "the percentage of emissions calculated using data obtained from suppliers or other value chain partners" and "a description of the data quality of reported emissions data." The Scope 3 Technical Guidance refers to five data quality indicators defined in section 7.3 of the Standard: technological, temporal and geographical representativeness, completeness and reliability. Every custom factor you introduce should be scored against them.

What changes with the GHG Protocol Land Sector and Removals Standard?

It adds traceability rules to Scope 3 for the first time, for companies with significant land-sector activities in their operations or value chain. The GHG Protocol released the Land Sector and Removals (LSR) Standard on 30 January 2026, with an effective date of 1 January 2027. Version 1.1, dated 30 June 2026, covers agriculture and removal technologies; forestry is excluded and remains open through a Forest Carbon Accounting RFI until 1 February 2027.

The LSR links your accounting boundary to what you know about origin:

• No knowledge of origin: global boundary, which the standard says generally does not meet the criteria for accurate, credible accounting.
• Known jurisdiction: jurisdiction boundary, physical traceability recommended.
• Known first point of aggregation: sourcing region boundary, physical traceability required.
• Known land management unit or field: farm or harvested area boundary.

The "sourcing region" is the LSR's term for what many teams call a supply shed: a spatially explicit area that supplies a raw material to the first point of aggregation or processing. Requirement 8 states that for land management emissions and removals, companies "shall have physical traceability to the sourcing region, LMU, or harvested area within their value chain", and it excludes book-and-claim approaches and certain mass balance setups. The standard also recommends that you "collect supplier-specific data for the GHG sources and sinks that are most significant."

Target setting follows suit. In a 19 March 2026 update, the SBTi set out version 1.2 of its Forest, Land and Agriculture (FLAG) Guidance, described as aligned with the LSR, and confirmed that companies may use either the draft guidance or the final LSR until 1 January 2027, after which the final LSR is required.

Can you use regenerative agriculture data to reduce your Scope 3 emissions?

Yes, if the practices happen on land that physically supplies you and you can prove it to at least the sourcing-region level. If you fund regenerative projects through a third-party developer in the same region but cannot trace your volumes to those farms, the result is an intervention, reported separately from your inventory.

The LSR says companies reporting on impact traceability "must report GHG emissions or removal impacts of projects and interventions separately" from the physical inventory. The broader rules for market instruments and interventions are still being written. The GHG Protocol's Actions and Market Instruments workstream has produced a Phase 1 White Paper, which the public consultations page describes as "not a draft standard"; that page also says a draft consolidated Corporate Standard is expected for public consultation in Q2 2027. Until then, do not net intervention claims into your inventory.

The practical consequence: ask your project developer which supplying farms or aggregation points the practices sit on. If the answer overlaps your physical supply, you may have an inventory claim. If not, you have a credible separate disclosure, which is still worth having.

Which emission factor types are acceptable, and what evidence should you keep?

Each factor type is defensible in the right context. The test is whether the evidence pack matches the claim.

• Database average (process LCA or EEIO). Acceptable for the long tail, low-materiality items, or when origin is unknown. Keep: database name and version, dataset ID, geography and year, mapping rationale, data quality scores.
• Sourcing-region (supply shed) factor. Acceptable when origin is known to the first point of aggregation, with physical traceability. Keep: region boundary definition, traceability evidence (volumes, aggregation points), model and inputs, calibration data, reviewer sign-off.
• Supplier PCF. Acceptable for supplier-specific product data, cradle-to-gate, with a compatible boundary. Keep: PCF file, methodology (ISO 14067, PACT), reference period, primary data share, data quality rating, verification status, allocation method.
• Farm-level regenerative data. Acceptable when practices happen on farms or fields that physically supply you. Keep: farm or field IDs and geolocation, practice records, model and version, baseline, measurement or calibration data, reversal monitoring for any removals claimed.
• Funded intervention outside your physical supply. Never in the inventory; separate disclosure only. Keep: project contract, developer methodology, location, verification, separate reporting line.

How do you reuse supplier PCFs and EUDR data as emission factors?

Treat supplier PCFs as candidate factors that must pass a quality gate, and treat EUDR geolocation as traceability evidence, not emissions data.

For PCFs, check the methodology first. ISO 14067:2018 sets requirements for quantifying a product carbon footprint. The PACT Methodology v3, launched in April 2025, requires companies to calculate and report a primary data share with each PCF, and makes data quality ratings required by 31 December 2027. It also makes biogenic and land-sector emissions mandatory, unless biogenic carbon content is below 5% or those emissions are below 3% of the cradle-to-gate footprint. A spreadsheet PCF with no stated method, period or boundary should not replace a documented database factor.

For land data, the European Commission confirms the EUDR covers cattle, cocoa, coffee, oil palm, rubber, soya and wood, applies to large and medium operators from 30 December 2026 following Regulation (EU) 2025/2650, and requires information such as "the geolocation of the plot of harvest." That plot data is exactly what the LSR asks for to move from a jurisdiction boundary to a farm or harvested-area boundary. Collect once, use twice.

If you want to see how supplier PCF collection and matching run at scale, explore the Supplier Engagement Hub.

How do you document custom emission factors so they pass assurance?

Build one evidence pack per factor and version everything. Follow these steps:

1. Inventory every custom factor. Export all non-database factors with owner, source and creation date.
2. Retire undocumented legacy factors. Any factor without a traceable source goes back to a database default before sign-off. An unexplained number is a finding.
3. Classify each factor using the list above and confirm its traceability level under the LSR.
4. Score data quality on the five GHG Protocol indicators and record PACT primary data share where relevant.
5. Attach the source pack: original file, methodology, reference period, geography, boundary and verification status.
6. Write explicit matching rules. Document which activities, suppliers, products and regions each factor applies to, and what it replaces.
7. Freeze a versioned configuration for each reporting year so restatements are explainable.
8. Disclose separately any intervention impacts that fail the physical traceability test.

Public examples show the payoff. Reckitt obtained precise emissions data for 25,000 products in just under four months, and one brand went from 18 representative products to 2,190 unique items, quadrupling the number of emission factors it used. Accor modulated 150 industry emission factors into 1,800 based on survey responses, with full data lineage.

Where does a carbon management platform help?

At CO2 AI, we keep custom emission factors, their documentation and their versions in one place. Supplier PCFs are collected and reviewed, then matching rules swap secondary factors for supplier-specific ones only where they apply. AI emission factor matching covers the long tail, so your experts spend their time on the factors that move the number.