10 Carbon Reporting Mistakes That Fail Audits in...

The Australian National Audit Office examined 545 NGER reports and found errors in 72% of them. Seventeen percent had significant errors. That was during the scheme's early years, but the fundamental causes haven't gone away. Manual data entry, stale spreadsheets, misunderstood emission factors, and reporting boundaries that nobody double-checks until an auditor asks.

And the stakes just got higher. NGER reporters are now automatically pulled into ASRS Group 2 for financial years starting 1 July 2026. The same emission numbers that go to the Clean Energy Regulator will sit in your annual report, subject to ASSA 5010 assurance. Under AASB S2, false or misleading climate statements carry penalties up to $16.5 million or 10% of annual turnover. Directors are personally liable.

These aren't theoretical risks. In July 2025, Beach Energy entered an enforceable undertaking for "inadvertent misstatements" across multiple NGER periods. Not fraud. Errors. The CER's fix: three years of externally commissioned reasonable assurance audits at Beach Energy's expense, plus a consultant to rebuild their control systems.

Here are the 10 mistakes that cause audit failures. We've seen every one of them across the companies we work with.

1. Using the wrong NGA Factors edition

This one is embarrassingly common. The NGA Factors workbook gets updated annually by DCCEEW. The 2025-26 edition changed state-based grid emission factors: Queensland dropped from 0.71 to 0.67 kg CO2-e/kWh, NSW fell from 0.66 to 0.64, Tasmania jumped from 0.15 to 0.20. It also added hydrogen combustion factors and updated flared gas methodologies.

Why it happens. Someone copies last year's spreadsheet and starts entering data. The emission factors from the 2024-25 edition are still sitting in the formula cells. Nobody updates them because nobody remembers to check whether DCCEEW published a new edition. Sometimes the new edition lands mid-reporting-period, creating confusion about which version applies to which months.

What it costs. For a company with 20 sites across multiple states consuming 10 million kWh annually, applying the wrong state factors can swing your reported Scope 2 emissions by 5-15%. Tasmania's factor jumped 33% between editions. A property portfolio concentrated in Hobart could see hundreds of phantom tonnes appear in the report purely from using a stale factor.

How to prevent it. Lock your emission factor library to a specific NGA edition with a clear effective date. When a new edition drops, load the updated factors separately and apply them only to activity data within the relevant reporting period. And keep the old version in place for prior-period calculations. An auditor will want to see which edition was applied to which data.

2. Skipping the energy content conversion

A gas bill shows 45,000 MJ. The NGA Factors workbook gives natural gas an emission factor of 51.53 kg CO2-e per GJ. Multiply 45,000 by 51.53 and you get 2,318,850 kg CO2-e. That's 2,319 tonnes from one building in one quarter. Sounds about right if you don't think about it too hard.

It's wrong by 1,000x. The bill was in megajoules. The factor is per gigajoule. Actual emissions: 2.32 tonnes.

Why it happens. Diesel is another common trap. The NGA Factors give transport diesel an emission factor of 69.9 kg CO2-e per GJ. But fuel records are in litres. To use the per-GJ factor, you first need diesel's energy content: 38.6 GJ per kilolitre, or 0.0386 GJ per litre. Skip that step and apply 69.9 directly to litres, and you've overstated by roughly 26x. We wrote a detailed breakdown of this problem because we see it so often.

What it costs. Order-of-magnitude errors. A 1,000x overstatement on gas or a 26x overstatement on diesel can push a facility above NGER reporting thresholds that shouldn't apply to it. Or trigger phantom Safeguard Mechanism obligations. Under AASB S2, it's a material misstatement in your annual report.

How to prevent it. Every calculation should pass through an explicit unit normalisation step. Gas in MJ gets divided by 1,000 before meeting a per-GJ factor. Diesel in litres gets multiplied by 0.0386 before meeting a per-GJ factor. If the source unit and the factor unit don't match, the calculation shouldn't proceed.

3. Duplicate entries from multiple sources

The same diesel delivery gets uploaded as a fuel card CSV, a supplier PDF, a photographed docket, and a line on the monthly statement. Four files. One delivery. Your Scope 1 figure is now inflated.

Why it happens. Carbon accounting pulls data from more sources than accounts payable does. Site managers photograph receipts. Accounts teams forward supplier invoices. Someone downloads a CSV from the fuel card portal. Another person uploads the monthly consolidated statement. Nobody coordinates. The files have different names, different formats, sometimes slightly different amounts (different flow meters reading the same delivery).

What it costs. We typically observe duplicate rates of 3-8% in carbon datasets when companies first run proper deduplication checks. On a 120,000-tonne facility under the Safeguard Mechanism, a 5% over-report means 6,000 phantom tonnes. At current ACCU prices around $30-35 per tonne, that's $180,000 to $210,000 in unnecessary surrender obligations. For nothing.

How to prevent it. Hash-based duplicate detection catches exact file copies but misses the real problem: different documents describing the same event. You need fuzzy matching across supplier name, quantity (with a tolerance window), date range, site, and material type. This is one area where we're honest that it's genuinely hard to get right. Tolerance settings that are too tight miss real duplicates. Too loose and you flag legitimate separate deliveries.

4. Using national average instead of state-specific grid factors

Australia doesn't have a single electricity grid. The NEM covers Queensland, NSW, ACT, Victoria, Tasmania, and South Australia. Western Australia runs the SWIS. The Northern Territory has its own system. Each grid has a different generation mix and therefore a different emission factor.

Why it happens. Some carbon calculators and spreadsheet templates include a single "Australian average" electricity factor. It looks reasonable. It's easy. And it's wrong for NGER purposes. The NGA Factors workbook specifies Scope 2 emission factors by state and territory: Victoria at 0.78, NSW at 0.64, Queensland at 0.67, Tasmania at 0.20, South Australia at 0.22 (2025-26 edition). Using a national average when your facilities are concentrated in low-carbon states like Tasmania or SA will materially overstate your Scope 2 emissions. Concentrated in Victoria? You'll understate.

What it costs. A company with 5 million kWh of consumption across Tasmanian sites using a national average factor (roughly 0.59) instead of the Tasmanian factor (0.20) would report 2,950 tonnes instead of 1,000 tonnes. That's a 195% overstatement. Under NGER, the CER expects state-specific factors for location-based Scope 2 reporting. Under AASB S2, your assurance provider will test whether you've used the appropriate factors.

How to prevent it. Every site in your system needs a state or territory tag. Every electricity calculation should automatically pull the grid factor for that state from the correct NGA edition. If someone tries to apply a generic "Australia" factor, that should raise a flag.

5. Facility boundary errors

Under the NGER Act (sections 9 and 11-11B), a "facility" isn't a building. It's an activity or series of activities. A construction project with temporary sites, a mine with remote camps and processing plants, a logistics operation with depots and a vehicle fleet. The reporting boundary wraps around all emission sources under the operational control of the registered entity at that facility.

Why it happens. Companies define their facility boundary once during initial NGER registration and then never revisit it. Acquisitions happen. New equipment gets installed. A backup diesel generator goes in at a remote camp. Solar panels go on the roof (affecting net electricity consumption). Nobody updates the facility scope to include (or account for) these changes.

What it costs. Omitting a source means understated emissions. Including a source that belongs to another entity under operational control means either double-counting or misallocation. Both fail an audit. The CER's operational control supplementary guideline is clear: if corporate changes affect who has operational control, the reporting boundary must be reassessed.

How to prevent it. Annual facility boundary reviews, timed to the start of the reporting year. Every corporate change (acquisition, divestment, new joint venture, new contractor arrangement) should trigger a check on operational control boundaries. Document the outcome even if nothing changed. An auditor asking "did you review your facility boundaries this year?" wants to see evidence, not hear "we didn't think we needed to."

6. Unit confusion between invoices and factors

This is related to mistake #2 but different. Here, the conversion isn't skipped; it's done with the wrong assumption about what unit the source data is actually in.

Why it happens. A fuel supplier invoices diesel in litres. The fleet management system records it in kilolitres. Someone exports the data and assumes it's in litres because "diesel is always in litres." They apply the energy content factor for litres (0.0386 GJ/L) to a number that's actually in kilolitres (where the factor should be 38.6 GJ/kL). Result: emissions understated by 1,000x. Or the reverse happens: data in litres gets treated as kilolitres and emissions are overstated by 1,000x.

Bulk LPG is another trap. Some suppliers invoice in litres. Others invoice in kilograms. The NGA Factors give LPG emission factors per GJ, with energy content factors of 25.7 GJ/kL or 49.6 GJ/tonne depending on which starting unit you have. Apply the wrong energy content factor and your emissions are off by roughly 50%.

What it costs. Same as mistake #2: order-of-magnitude errors or material misstatements. But this version is harder to catch because the person doing the calculation thinks they did the conversion correctly. They just started with the wrong assumption about the input unit.

How to prevent it. Validate the unit at the point of extraction. When a value is pulled from an invoice or data export, tag it with the unit as written on the source document. Then let the system handle the conversion based on that tagged unit, not on an assumption.

7. Scope misclassification

A company-owned ute burns diesel on a construction site. That's Scope 1. But the sustainability analyst records it under Scope 3 (upstream transportation) because the fuel was purchased by a subcontractor who billed it back. Or the reverse: a contractor's excavator gets classified as Scope 1 because it's operating on the company's site.

Why it happens. The GHG Protocol boundary is clear in theory: if you own or operationally control the source, it's Scope 1. In practice, construction sites, joint ventures, and contractor arrangements blur the line. Under NGER, the operational control test asks who has authority to introduce and implement operating, health and safety, and environmental policies. A contractor's diesel excavator on your site might be your Scope 1 if you direct their operations and environmental policies. Or it might be Scope 3 if they operate independently under their own policies.

What it costs. Under AASB S2, Scope 1 and 2 carry full liability from year one of reporting. Scope 3 gets a one-year deferral. If you've been classifying company-owned vehicle emissions as Scope 3, you're understating the scopes that face immediate assurance and legal exposure. The penalties for misleading climate disclosures reach $16.5 million or 10% of turnover.

How to prevent it. Build a classification framework for your specific operations before you start collecting data. For every emission source, ask: who owns it, who controls its operating policies, and who controls its environmental policies? Document the answer. For contractor arrangements, review the contract. The scope classification flows from the control test, not from who pays the fuel bill.

8. Ignoring fugitive emissions

Refrigerant leaks from commercial HVAC systems. SF6 leaks from electrical switchgear. These are Scope 1 fugitive emissions. And most companies either forget them entirely or drastically underestimate them.

Why it happens. Nobody sees a refrigerant leak. There's no invoice for it. R-410A, the most common refrigerant in commercial split systems, has a global warming potential of 2,088. One kilogram of leaked R-410A equals over two tonnes of CO2-e. A typical commercial HVAC system losing 5% of its refrigerant charge annually can contribute 10+ tonnes CO2-e without anyone noticing. SF6 in high-voltage switchgear is worse: it has a GWP of 23,500. Even small leaks are significant.

Under NGER, all SF6 emissions are reportable from the operation of facilities across all industry sectors. And the NGER framework has known gaps: exemptions for small units, reliance on estimated data, and absence of mandatory leakage registers make accurate reporting harder than it should be.

What it costs. For a property portfolio with 100 split AC units, each containing 3-5 kg of R-410A, a 5% annual leak rate across the portfolio generates roughly 30-50 tonnes CO2-e. That's material for a mid-size company. Miss it entirely and your auditor will ask why your Scope 1 inventory doesn't include fugitive emissions. Because they always ask.

How to prevent it. Maintain a refrigerant register: equipment type, refrigerant type, charge quantity, date of last service, quantity topped up. The top-up method is the simplest way to estimate leakage. If a technician adds 2 kg of R-410A during servicing, you just had 2 kg of fugitive emissions at a GWP of 2,088. That's 4.18 tonnes CO2-e from one service call. For SF6, track installed quantities and reconcile annually against purchases and disposal records.

9. No audit trail from reported number to source document

An auditor picks a number from your NGER report. Say, 1,247 tonnes CO2-e for Scope 1 at Facility 3. They ask: "Show me the source documents and calculations that produced this number."

If the answer involves opening three spreadsheets, searching a shared drive for PDFs, and explaining from memory which invoices fed which cells, you've already failed.

Why it happens. Spreadsheet-based reporting doesn't create audit trails. It creates snapshots. When someone types a number into a cell, the link between that number and the invoice it came from exists only in the analyst's memory (and maybe a filename scribbled in a comment). When the analyst leaves, so does the trail. We've talked to sustainability managers who inherited a predecessor's NGER spreadsheet and couldn't explain a single number in it.

What it costs. The CER's record-keeping requirements mandate retaining source data for five years from the end of each reporting period. Under ASSA 5010, your assurance provider must obtain sufficient and appropriate evidence. If they can't trace a sampled figure back to its source, that's an assurance finding. It calls into question every other number in the report.

The Beach Energy enforceable undertaking is instructive here. The CER didn't just flag wrong numbers. They flagged the absence of internal controls that would have caught those errors. An unverifiable audit trail is exactly the kind of control weakness that leads to enforceable undertakings.

How to prevent it. Every reported emission figure needs a traceable chain: source document (with file reference), extracted data points (with extraction method), unit conversion steps (with factors and versions), emission factor applied (with NGA edition and table reference), and final calculated value. If any link in that chain is "someone typed it in," the trail is broken.

10. Overriding system calculations without documenting why

Automated calculations get overridden. Sometimes for good reasons. An AI extraction misread a smudged fuel docket. A meter was faulty and the billed amount doesn't reflect actual consumption. A one-off event (generator testing, emergency operations) created an anomalous reading that needs context.

But when the override isn't documented, it looks like data manipulation.

Why it happens. Someone reviews a calculated figure, decides it "doesn't look right," and changes it. Maybe they're correct. Maybe they've introduced an error. Either way, the system now shows a number that doesn't match the calculation logic, and there's no record of who changed it, when, or why.

What it costs. Under ASSA 5010, assurance providers test the integrity of data processing. If they find a calculated value that's been manually overridden without documentation, they have to treat it as a potential integrity failure. That triggers additional testing, extended timelines, and higher assurance fees. Worst case, it becomes a qualification in the assurance opinion. Under NGER, undocumented overrides in historical data make it nearly impossible to defend prior-period calculations if the CER asks questions during a re-audit.

How to prevent it. Overrides are fine. Undocumented overrides are not. Every manual adjustment should require a reason (free text, minimum 20 characters), a timestamp, and the identity of who made the change. The original calculated value should be preserved alongside the override so both are visible. This isn't bureaucracy. It's the difference between "we corrected an extraction error" and "we can't explain why this number doesn't match."

The pattern behind all 10 mistakes

If you read through this list carefully, there's a common thread. Every one of these errors thrives in environments where data entry, calculation, and review happen in disconnected steps. An invoice arrives. Someone types numbers into a spreadsheet. Someone else (or the same person, weeks later) applies emission factors. A third person reviews the totals. Nobody checks the connections between those steps because the connections aren't recorded.

That's the environment where wrong factors persist for a full reporting year. Where gas consumption in MJ meets a per-GJ factor without conversion. Where the same diesel delivery gets counted from three different source files.

We built Carbonly specifically to close these gaps. Documents get processed when they arrive, not in a September panic. Unit validation happens at extraction. Emission factors are version-locked to the correct NGA edition. Duplicates are flagged before they enter the dataset. And every number carries a traceable path from source document to reported figure, because we know from 18 years of building enterprise data systems at BHP, Rio Tinto, and Senex Energy that traceability is what separates defensible data from guesswork.

We won't pretend the tool catches everything. Facility boundary assessments still need human judgment. Scope classification for complex contractor arrangements requires understanding the actual contractual relationship, not just the data. And we're still improving our fugitive emissions tracking, because refrigerant registers are one of those data sources that barely exist in structured form at most companies.

But the mechanical errors, mistakes 1 through 4 and 6? Those shouldn't happen anymore. Not when the October deadline carries both NGER and AASB S2 consequences.

If your team is preparing for the 2025-26 NGER submission or getting ready for ASRS Group 2 reporting, we'd rather help you fix these problems before your auditor finds them. Reach out at hello@carbonly.ai for per-project pricing.

Related Reading: