Materiality Thresholds for Emissions Errors Under...

The error lands in your inbox on a Tuesday. A site engineer has noticed the diesel meter at one of your facilities was reading in litres while the spreadsheet template assumed kilolitres. The misstatement is 1.8% of total Scope 1. Do you restate? Do you flag it in a note? Do you do nothing and fix it in the next period?

Most sustainability teams in Australia can't answer that question with confidence. Not because they're sloppy. Because nobody told them where the line is.

AASB S2 doesn't give you a number. The 5% rule of thumb from financial reporting doesn't translate cleanly to a tonne of CO2-e. ASIC has now made clear in RG 280 and the Corporate Plan 2025-26 that they expect documented materiality judgements before the error appears, not after. And the ANAO performance audit found 72% of NGER reports submitted to the Clean Energy Regulator contained errors, with 17% flagged as significant. The materiality question is not theoretical. It's the most common audit finding in Australian emissions reporting.

This is the decision rule we recommend reporters document and operationalise.

Why financial materiality doesn't translate

In financial reporting, materiality is conventionally anchored to a percentage of profit, revenue, or net assets. The unspoken 5% rule has decades of audit practice behind it. Practitioners know what it means. Auditors know how to apply it.

Emissions don't work the same way. A 2% error at a Safeguard facility holding a declining baseline can be more material than a 5% error in a small office portfolio. A 1% error in Scope 2 that masks a deteriorating trend can be more material than a 10% error in Scope 3 Category 7 employee commuting.

AASB Practice Statement 2 Making Materiality Judgements is the methodology AASB S2 paragraph 64 points to. Practice Statement 2 is explicit that materiality is entity-specific and that quantitative thresholds alone are insufficient. The judgement has to consider qualitative factors and the information needs of primary users of the report.

For climate disclosures, primary users are capital providers. The question becomes: would this error change a reasonable investor's view of the entity's climate-related prospects? That's a different question to "is this 5% of EBITDA?".

Three numerical anchors that work in practice

Practice Statement 2 doesn't forbid numerical anchors. It just refuses to set them for you. The reporters we see surviving early ASIC surveillance use three quantitative tests in combination, then layer qualitative overrides.

Scope-level threshold. 5% of total reported Scope 1, Scope 2, or Scope 3 individually. Not 5% of combined emissions, which lets a Scope 1 error hide behind a large Scope 3 footprint. Scope-level testing forces visibility on the categories investors care about most.

Facility-level threshold. 5% of any single reported NGER facility's emissions. This catches operational errors that wouldn't move a corporate-level number but would change a facility's standing under the Safeguard Mechanism or change the picture of a specific asset's transition risk.

Threshold proximity test. Any error that could push the entity over or under an NGER threshold (50 kt CO2-e corporate, 25 kt CO2-e facility, 100 TJ / 200 TJ energy) or a Safeguard baseline. An error of 800 tonnes is immaterial in isolation. The same 800 tonnes at a facility sitting at 24,600 tonnes is the difference between in-scope and out-of-scope. That's material at any size.

Pick which one you're using. Document it in your basis of preparation. ASIC will read it.

The qualitative override

Numerical thresholds are necessary and insufficient. Practice Statement 2 paragraph 49 makes clear that an item small in size can be material because of its nature or the circumstances surrounding it. For emissions, the qualitative overrides we'd treat as material regardless of size include:

The error changes the direction of a disclosed trend (emissions reported as decreasing are actually increasing, or vice versa)
The error affects a Safeguard Mechanism baseline calculation or a Safeguard Mechanism Credit position
The error affects a sustainability-linked loan KPI calculation
The error affects a stated public target or SBTi commitment metric
The error was made knowingly or after the question had been flagged
The error aggregates with other individually-immaterial errors across the same disclosure

Each of these has a regulatory or contractual consequence that makes the dollar or tonne size of the error secondary. The 1% error that flips a trend line is the kind of finding ASIC has signalled it will look for.

NGER and AASB S2 thresholds are not the same

This is the trap most first-time reporters fall into.

NGER materiality is largely binary at the facility level. Either the facility crosses the 25 kt CO2-e or 100 TJ threshold and triggers reporting, or it doesn't. The Clean Energy Regulator's audit framework and the National Greenhouse and Energy Reporting (Audit) Determination 2009 contemplate restatement when an error is identified that affects compliance with the Act. There is no soft 5% wraparound. If the facility crossed the line and you didn't report, that's a contravention regardless of percentage.

AASB S2 uses entity-wide materiality, includes Scope 3, and is filtered through Practice Statement 2's information-needs lens. A 4% Scope 1 error sitting below your AASB S2 threshold can still be a NGER reporting failure if it flips a facility's threshold status. The opposite also holds: a Scope 3 Category 1 purchased goods error that is immaterial under NGER (because Scope 3 isn't in NGER) can be material under AASB S2 if it changes the entity's climate-related prospects picture.

You need two materiality policies sitting alongside each other, with one decision tree that routes an identified error to the right framework. That's the part most basis-of-preparation documents we've reviewed during pilot work don't yet address.

The basis of preparation problem

Most Group 1 reporters we've seen during pilot work and assurance-readiness reviews either didn't document a numerical threshold for emissions errors, or documented "follows AASB Practice Statement 2" without operationalising what that means.

ASIC's RG 280 expects more. The regulatory guide issued on 31 March 2025 is explicit that directors are responsible for establishing systems to identify, assess and monitor material climate-related risks, and that the materiality judgement process itself needs to be documented sufficiently to support the disclosure. RG 280 doesn't set a number. But it does set the expectation that you set one and explain it.

A defensible basis of preparation needs at minimum:

The numerical anchor (which scope-level, facility-level, or threshold-proximity test applies)
The qualitative override criteria (the list of trigger conditions that make small errors material)
The aggregation rule (how individually-immaterial errors are summed and tested)
The decision tree mapping discovered errors to one of three outcomes: restatement, supplementary disclosure, no action
The documentation requirement showing the judgement was applied to each identified error

This is the materiality policy. Pair it with the restatement playbook we walked through earlier this month and you have the closed loop assurance providers ask for.

What auditors will ask

ASSA 5010 assurance practitioners scope their own materiality at the engagement level. The first question they ask is whether your materiality policy aligns with the materiality they've scoped for assurance. If the assurance provider's threshold is tighter than yours, every error you treated as immaterial will be re-tested at their threshold.

Expect specific questions:

What is your quantitative threshold and how was it set?
What qualitative overrides have you defined?
How do you aggregate individually-immaterial errors?
Show the documentation of judgement for the three largest identified errors this year
How does your materiality threshold align with the assurance materiality we've scoped?
Has the threshold changed from Year 1, and if so, is that disclosed as a change in accounting estimate?

The reporters who answer these quickly have written it down. The reporters who don't end up reconstructing the reasoning during the assurance fieldwork window, which is the worst possible time to be making materiality judgements for the first time. The evidence pack assurance providers want is built around exactly this documentation trail.

The aggregation trap

Five individually-immaterial errors that all land on the same side of the ledger sum to a material misstatement. We've seen this pattern in real submissions: meter conversion error, factor version mismatch, missing decimal in a contractor invoice, unit confusion in a refrigerant log, transposition in a fuel docket. Each one is well below threshold. The combined understatement is material.

Without a basis-of-preparation aggregation rule, this gap is invisible until the auditor finds it. The rule we'd recommend documenting: identified errors are tested individually against the numerical threshold, then aggregated by scope and tested again at the same threshold. If the aggregate exceeds threshold, the cluster is material even if no individual error is.

This is mechanical, but it doesn't work without an audit trail that captures every identified error with its tonne impact, the affected scope, and the direction of the misstatement. Spreadsheet-based reporting tends to record fixes by overwriting the previous number, which destroys the evidence needed to aggregate later. That's a documentation architecture problem, not a methodology problem.

Year-2 implications

Once you've set a threshold and applied it in Year 1, changing it in Year 2 is itself a disclosable change. AASB 108 paragraphs 32-38 treat changes in accounting estimates as a current-period adjustment with disclosure of the nature and amount. A change in materiality threshold for emissions falls into the same category.

The practical implication: don't set a threshold you'll need to revisit. The reporters who lifted the threshold in Year 2 because Year 1 generated too many "material" findings end up disclosing that they relaxed the standard. That's a worse outcome than the over-reporting they were trying to avoid.

Set the threshold conservatively in Year 1, with the qualitative overrides doing the work of catching genuinely material small errors. Hold it through Year 2.

What makes the threshold operational

A materiality threshold is only useful if you can identify and aggregate errors against it. The architecture has to support the policy. Without an audit trail that captures source documents, emission factor versions, and the calculation chain, you can identify an error but you can't quantify its impact at the scope or facility level, and you can't aggregate it with other findings.

Carbonly's document AI engine reads supplier invoices, fuel dockets, meter reads, and refrigerant logs across PDF, CSV, Excel, Word, PPT, RTF, and image formats, with the source document linked to every emission record through the audit trail. Period locking prevents silent overwrites of submitted figures. The five-tier material matching keeps factor selection auditable. AR5 and AR6 GWP values are stored side-by-side so the NGER and AASB S2 reconciliation is explicit, not hand-built. None of this changes what materiality is. It changes whether your policy can be applied consistently across thousands of records and survive the question "show me the working".

The restatement playbook we published earlier this month assumes the materiality policy is in place. This piece is the upstream decision rule that feeds it.

Where to start

Three steps, ordered.

First, write the numerical threshold down. Pick the scope-level, facility-level, and threshold-proximity tests that apply to your portfolio. Don't combine them into a single corporate-wide percentage. Investors care about the components.

Second, write the qualitative overrides down. Trend direction, Safeguard impact, SLL KPI impact, target metric impact, knowing-error, aggregation. These are the small errors that will sink the disclosure if you don't catch them.

Third, write the aggregation rule and the decision tree down. The decision tree should route every identified error to one of three outcomes: restate, supplementary disclose, or fix forward with no current-period disclosure. Document the reasoning for each.

Pair this materiality policy with the restatement procedure and the risk register practice, and you have a basis of preparation that holds up to RG 280-level scrutiny.

If you want a conversation about how an audit trail architecture supports this policy in practice, including the period locking and source-document chain that make aggregation testing possible, the team is at hello@carbonly.ai or /waitlist.

Materiality Thresholds for Emissions Errors Under AASB S2 and NGER: The Decision Rule Australian Reporters Need to Document