Verifying Carbon Claims: Assurance and Audit

A carbon figure held internally is an estimate. Published, it becomes a claim, and claims are examined by assurance providers, market surveillance authorities, competitors and NGOs, each with different powers and similar questions.

Understanding what those questions are is more useful than trying to make the number lower, because most failures are about whether a figure can be traced rather than whether it is favourable.

The levels of verification

The distinction between limited and reasonable assurance is not marginal. Limited assurance concludes that nothing came to the provider’s attention suggesting the figure is materially misstated; reasonable assurance concludes positively that it is fairly stated, and costs substantially more to obtain.

What a verifier actually examines

1. Boundary: which life cycle stages are included, and whether that matches the claim.
2. Data sources: which figures are primary, which secondary, and whether sources are traceable.
3. Allocation: how co-product burdens were split, and whether the method is applied consistently.
4. Calculation: the arithmetic and the emission factors used, including their vintage.
5. Completeness: what was excluded under cut-off criteria, and whether exclusions are justified.

Calculation sits fourth for a reason. Arithmetic errors are rare and easy to correct; boundary and traceability problems are common and frequently cannot be corrected without redoing the collection.

Why traceability causes most findings

A verifier asked where a figure came from needs an answer that ends at a source rather than at a spreadsheet. A supplier value transcribed into a model by an analyst who has since left is not traceable, however accurate it may be.

This is the recurring first-engagement finding: the data exists and the audit trail does not. Correcting it means going back to suppliers for evidence they already provided informally, which costs relationship capital that has usually already been spent.

Which claims fail most often?

Carbon neutral is the most exposed. Directive (EU) 2024/825 restricts claims that a product has neutral, reduced or positive environmental impact based on offsetting, which removes the mechanism most such claims relied on.

Comparative claims need a panel

ISO 14044 requires a critical review by a panel of interested parties for comparative assertions disclosed to the public. Publishing that your product has a lower footprint than a competitor is exactly such an assertion.

Marketing teams frequently make this claim on the basis of a study that conforms to the standard in every other respect. Citing ISO 14044 in support of a comparative claim that skipped the panel requirement is a specific and checkable problem.

What good evidence looks like

For each material input: a figure, its functional unit, its boundary by stage, its allocation method, whether it is primary or secondary at each stage, the reference period, and the party asserting it, retrievable without contacting anyone.

That is precisely what a signed supplier credential held against a product record provides, which is the practical link between passport infrastructure and verification cost. The passport does not make the claim true; it makes it demonstrable.

How to prepare for a first engagement

1. Write the method down before the verifier asks, including every choice and its justification.
2. Trace ten figures at random from the published result back to their source, and time it.
3. Identify which figures are secondary and quantify what share of the total they represent.
4. Check that any comparative or neutrality claim has the process behind it that its standard requires.
5. Restate baselines deliberately, with the reason disclosed, rather than quietly.

The second item is the most informative hour you can spend. If tracing a figure to its source takes longer than a few minutes, the first assurance engagement will find that, and finding it yourself is considerably cheaper.

Who verifies, and does it matter?

For CSRD, an accredited assurance provider, which in most member states means a statutory auditor or an independent assurance services provider subject to equivalent requirements. For product claims outside that regime, the market is looser and the choice matters more.

A verifier without demonstrable competence in life cycle assessment will check arithmetic and miss boundary problems, which is the failure mode that actually causes trouble later. Accreditation against a recognised scheme is worth more than a certificate that simply says verified.

Independence matters for the same reason. A review by the consultancy that produced the study is a quality step rather than verification, and describing it as third-party verification is a claim that will not survive scrutiny.

What does it cost?

Less than the work of preparing for it, in most cases. The verifier’s fee is a fraction of the internal effort spent assembling evidence, reconstructing derivations and answering questions about figures nobody documented at the time.

That ratio is the argument for structuring evidence at collection. Organisations that record source, method and asserting party as they go find verification a review exercise; those that do not find it an excavation, and the excavation is where the cost sits.

How often does a claim need re-verifying?

Whenever the underlying figure changes materially, and on a defined cycle regardless. A verified footprint attached to a product whose supplier, formulation or production route has since changed is a claim about something that no longer exists.

Recording which version of a figure applies to which production period is what makes this manageable, and it is the same versioning problem the passport handles for every other field. A single current value overwritten on each recalculation destroys the record of what was claimed when.

What verification does not do

It does not certify that your footprint is good, or that your product is better than an alternative. A verified figure is a figure produced properly and reported transparently, which is a statement about process rather than about performance.

Presenting verification as an environmental endorsement is itself a claim that can be challenged, and it is one of the more common ways a technically correct disclosure becomes a misleading communication.