Sustainability reports used to be voluntary disclosures, designed for stakeholders who read them with goodwill. Under the Corporate Sustainability Reporting Directive (CSRD) [1], they have become audited documents, reviewed by the same firms that audit your financial statements. Limited assurance is mandatory from the first year of application [2].
This is a fundamental change in how sustainability information is treated. An auditor under CSRD does not read your report. They test it. They ask where each number came from, who verified it, and what evidence supports the methodology behind it. They will challenge claims that lack substantiation, and they will issue qualifications when the documentation does not hold up.
The Omnibus I package, agreed at the end of 2025 and formally adopted in early 2026, has reduced the number of mandatory ESRS datapoints by approximately 61% [3]. It has also removed the previously planned transition to reasonable assurance, keeping CSRD reporting under limited assurance for the foreseeable future. The volume of disclosures has decreased; the standard of rigour expected on what remains has not. Climate disclosures, and Scope 3 in particular, are firmly in the core set.
If your value chain touches agriculture, your Scope 3 emissions are likely to represent the largest part of your total corporate footprint. The World Wildlife Fund estimates that around 70% of food-related emissions originate directly from farms [4], with most food companies seeing 80 to 90% of their total footprint sitting in Scope 3. Most of these emissions come from on-farm activities: fertiliser, livestock, fuel, soil management, and land-use change. They are generated far from your factories, by suppliers you do not directly control, on tens or hundreds of thousands of hectares scattered across geographies.
That distance is the problem. Farm-level data is harder to collect than industrial data. It is harder to verify. It depends on a fragmented ecosystem of platforms, methodologies, and assumptions that rarely align. And because it represents the bulk of the footprint, it is also the part of the report that auditors spend the most time examining.
This applies whether your climate strategy is built around insetting, contribution, or both. The audit logic does not change between models. What changes is the type of evidence each model needs to produce.
After working with corporate clients preparing for CSRD assurance, we have observed that auditors apply three principles when they review agricultural Scope 3. None of them are surprising in isolation. Together, they form a standard that most current sustainability reports were not designed to meet.
An auditor is not trying to recompute your emissions. They are trying to verify that the process you used is sound, that the inputs are credible, and that the decisions you made along the way are documented. A modest data set with strong documentation will often perform better in audit than a sophisticated data set with weak documentation.
This catches most teams off guard. The work goes into producing the number, but the audit value sits in the rationale: why this scope was chosen, why this methodology was applied, why this category was excluded. If those decisions were made implicitly, they will need to be reconstructed under pressure. If they were documented at the time, they hold up.
Every claim in your report needs to be backed by data of sufficient quality, drawn from real agricultural activity rather than generic averages. For Scope 3, this means activity data collected at farm level: what fertiliser was used, what crop rotation, what tillage practice, what yield. You do not necessarily need to trace each tonne back to a specific field, but you do need to demonstrate that your numbers come from actual fields, not from estimates layered on top of estimates. The GHG Protocol Corporate Standard [5] requires that companies disclose information of sufficient quality to allow verification, and auditors will sample-test the chain that connects on-farm activity to the disclosed figure.
This is where many corporate teams discover gaps they did not know they had. Suppliers report numbers without showing how they were calculated. Carbon programmes deliver claims without supporting field data. Multiple systems report different numbers for the same supplier with no documented reconciliation. Each gap is a place where the audit can stall.
The CSRD framework is unambiguous on this point: gross emissions are reported separately from any credits, removals, or contribution claims. They never cancel each other out. A single "net" figure that mixes the two is a frequent cause of audit qualification.
This is also where the distinction between insetting and contribution becomes decisive. Insetting reduces emissions inside your value chain and reports against your Scope 3. Contribution finances climate action outside your value chain and is reported as a separate claim. Auditors will look closely at how each is presented, and at whether the boundary between the two is documented and consistent with the underlying accounting.
Sustainability reports used to be voluntary disclosures, designed for stakeholders who read them with goodwill. Under the Corporate Sustainability Reporting Directive (CSRD) [1], they have become audited documents, reviewed by the same firms that audit your financial statements. Limited assurance is mandatory from the first year of application [2].
This is a fundamental change in how sustainability information is treated. An auditor under CSRD does not read your report. They test it. They ask where each number came from, who verified it, and what evidence supports the methodology behind it. They will challenge claims that lack substantiation, and they will issue qualifications when the documentation does not hold up.
The Omnibus I package, agreed at the end of 2025 and formally adopted in early 2026, has reduced the number of mandatory ESRS datapoints by approximately 61% [3]. It has also removed the previously planned transition to reasonable assurance, keeping CSRD reporting under limited assurance for the foreseeable future. The volume of disclosures has decreased; the standard of rigour expected on what remains has not. Climate disclosures, and Scope 3 in particular, are firmly in the core set.
If your value chain touches agriculture, your Scope 3 emissions are likely to represent the largest part of your total corporate footprint. The World Wildlife Fund estimates that around 70% of food-related emissions originate directly from farms [4], with most food companies seeing 80 to 90% of their total footprint sitting in Scope 3. Most of these emissions come from on-farm activities: fertiliser, livestock, fuel, soil management, and land-use change. They are generated far from your factories, by suppliers you do not directly control, on tens or hundreds of thousands of hectares scattered across geographies.
That distance is the problem. Farm-level data is harder to collect than industrial data. It is harder to verify. It depends on a fragmented ecosystem of platforms, methodologies, and assumptions that rarely align. And because it represents the bulk of the footprint, it is also the part of the report that auditors spend the most time examining.
This applies whether your climate strategy is built around insetting, contribution, or both. The audit logic does not change between models. What changes is the type of evidence each model needs to produce.
After working with corporate clients preparing for CSRD assurance, we have observed that auditors apply three principles when they review agricultural Scope 3. None of them are surprising in isolation. Together, they form a standard that most current sustainability reports were not designed to meet.
An auditor is not trying to recompute your emissions. They are trying to verify that the process you used is sound, that the inputs are credible, and that the decisions you made along the way are documented. A modest data set with strong documentation will often perform better in audit than a sophisticated data set with weak documentation.
This catches most teams off guard. The work goes into producing the number, but the audit value sits in the rationale: why this scope was chosen, why this methodology was applied, why this category was excluded. If those decisions were made implicitly, they will need to be reconstructed under pressure. If they were documented at the time, they hold up.
Every claim in your report needs to be backed by data of sufficient quality, drawn from real agricultural activity rather than generic averages. For Scope 3, this means activity data collected at farm level: what fertiliser was used, what crop rotation, what tillage practice, what yield. You do not necessarily need to trace each tonne back to a specific field, but you do need to demonstrate that your numbers come from actual fields, not from estimates layered on top of estimates. The GHG Protocol Corporate Standard [5] requires that companies disclose information of sufficient quality to allow verification, and auditors will sample-test the chain that connects on-farm activity to the disclosed figure.
This is where many corporate teams discover gaps they did not know they had. Suppliers report numbers without showing how they were calculated. Carbon programmes deliver claims without supporting field data. Multiple systems report different numbers for the same supplier with no documented reconciliation. Each gap is a place where the audit can stall.
The CSRD framework is unambiguous on this point: gross emissions are reported separately from any credits, removals, or contribution claims. They never cancel each other out. A single "net" figure that mixes the two is a frequent cause of audit qualification.
This is also where the distinction between insetting and contribution becomes decisive. Insetting reduces emissions inside your value chain and reports against your Scope 3. Contribution finances climate action outside your value chain and is reported as a separate claim. Auditors will look closely at how each is presented, and at whether the boundary between the two is documented and consistent with the underlying accounting.
Both models can pass external assurance. Both face new questions under it. The audit logic is consistent, but where it concentrates depends on the model.
For insetting programmes, the audit pressure concentrates on supplier data. Auditors will ask how supplier-reported emissions were verified, what the data quality is on the categories that drive most of the footprint, and how primary farm-level data was obtained on material sources. The entire value of an insetting claim rests on its ability to demonstrate real reductions inside the supply chain, with evidence that holds up to external review.
For contribution programmes, the audit pressure shifts to claim integrity. Auditors will look at whether the underlying outcomes are independently verified, whether retirement records exist on a credible registry, and whether permanence and additionality are documented. The contribution claim is reported separately from gross emissions, but the supporting evidence is no less rigorous than for an insetting claim.
Companies running both models in parallel, which is increasingly the norm, face an additional governance question: how do you ensure that a tonne reduced inside the value chain is never confused with a tonne financed outside it? Documentation of this boundary becomes part of what the audit tests.
There is no universal protocol, because the right approach depends on the maturity of your data, the structure of your supply chain, and the model you have chosen. But there is a structured way to think about the work.
It starts with a hard look at where your numbers actually come from, followed by a methodical upgrade of data quality on the categories that matter most. It continues with documenting every methodological decision in a form an auditor can review. It requires drawing a clear, defensible line between insetting and contribution claims. It depends on independent verification of every programme you rely on, under recognised international standards. And it ends with connecting your climate plan to real budget and real supplier contracts, so that the transition is more than an announcement.
Soil Capital has built its corporate programmes around exactly these principles: every tonne of CO₂ reported is traceable to a specific farm, a specific year, and a specific set of practices, verified annually by TÜV Rheinland under ISO 14064-2, with a clean separation maintained between insetting reductions and contribution-grade outcomes.
We have distilled what we observe in audit conversations into the CSRD Audit Readiness Guide, which translates these principles into the specific questions auditors ask and the evidence each one requires. It is the practical companion to this article, designed for sustainability leaders who want to walk into external assurance with confidence.
[1] European Commission, Corporate sustainability reporting — finance.ec.europa.eu
[2] ICAEW, CSRD Sustainability Assurance — icaew.com
[3] EFRAG, Amendments to the European Sustainability Reporting Standards — efrag.org
[4] World Wildlife Fund, Scope 1, 2, and 3 Emissions in Food Supply Chains — worldwildlife.org
[5] GHG Protocol, Corporate Accounting and Reporting Standard — ghgprotocol.org
🌱 Insetting: Reduce your Scope 3 emissions by investing in regenerative agriculture within your supply shed.
Discover our Insetting offer
🌍 Contribution: Support regenerative farmers locally and contribute to global net zero, beyond value chain mitigation.
Discover our Contribution offer
