How Australian Climate Compliance Outgrew the Spreadsheet in 18 Months
Between late 2024 and mid 2026, the shape of Australian climate compliance changed. NGER got more detailed. AASB S2 went live for Group 1 and is coming for Group 2. Safeguard baselines are now production-adjusted and declining. Scope 3 is moving from spend-based approximation to activity-based physical units. ASAE 3410 assurance is real. Here's why the spreadsheet that worked in 2023 won't survive 2026, and what audit-ready actually means now.
Eighteen months ago, an Australian sustainability lead could run climate compliance out of a shared spreadsheet. It wasn't elegant. It wasn't fun. But it worked. One analyst pulled bills into tabs, applied factors from the NGA workbook, rolled the numbers up, and the NGER submission went out on 31 October.
That workflow is done. Not in a "nice to modernise" way. In a "your auditor will not accept this" way.
The last year and a half has quietly rewired what Australian climate compliance demands from a reporting entity. NGER got more granular. AASB S2 went live for Group 1 for financial years starting from 1 January 2025, with Group 2 on FYs from 1 July 2026 and Group 3 from 1 July 2027. Safeguard baselines turned production-adjusted with a 1% minimum decline trajectory for TEBA facilities. Scope 3 expectations shifted from spend-based approximation toward activity-based physical units wherever the data exists. And ASAE 3410 assurance stopped being a theoretical conversation and started being a line-by-line evidence test.
The spreadsheet was designed for a different world. It's worth walking through what actually changed, why the tool no longer fits, and what "audit-ready automation" means in practical terms.
What actually changed between late 2024 and now
We'll keep this short, because most readers are living it. But it's worth naming the shifts together, because the cumulative effect is bigger than any single one.
NGER got more detailed. The Clean Energy Regulator has been tightening documentation expectations around per-facility breakdowns, per-gas reporting (CO2, CH4, N2O, HFCs, PFCs, SF6), and fuel combustion method elections. The 31 October deadline hasn't moved. Using AR5 GWP values under NGER hasn't moved either. What's moved is the quality of evidence expected to sit behind each number.
AASB S2 went from paper to practice. Group 1 reported against FYs starting from 1 January 2025. The first filings have now hit ASIC. Real auditors asked real questions. A lot of patterns that looked defensible in a draft turned out to be indefensible when someone asked "show me the source document for this kWh figure." AASB S2 also uses AR6 GWP, which is a different set of values to NGER's AR5. A single reporting entity can legitimately need both at once, for different frameworks, from the same underlying activity data.
Safeguard Mechanism baselines are now production-adjusted. The reform brought in a declining trajectory of at least 1% per year for TEBA facilities, with Safeguard Mechanism Credits (SMCs) earnable, bankable, and tradeable subject to a 7-year expiry. ACCU surrender kicks in if a facility exceeds its baseline by more than 30%. That's a compliance and financial problem at once, and it punishes anyone who only checks their Safeguard position annually.
Scope 3 started moving to physical units. Spend-based Scope 3 is still allowed, and in many categories still the only feasible method. But the direction of travel is unmistakable: where activity data is available, auditors and investors expect it. Physical units. Passenger-kilometres. Tonne-kilometres. Litres. Kilowatt-hours. Full-time equivalents. Room nights. Square metres. Hours. Spend-based estimates have a shelf life now, and that shelf life is getting shorter.
ASAE 3410 assurance is real. Limited assurance over Scope 1 and Scope 2 is in scope from day one of AASB S2 for the relevant group. Reasonable assurance phases in later. Either way, an independent practitioner is going to pull records and test them. The question "can you produce the source document for this line item?" is no longer hypothetical. It's the job.
Any one of these changes is manageable. The problem is that they arrived together, and they compound.
Why the spreadsheet breaks at this level
Spreadsheets are brilliant analytical tools. We still use them constantly. But the role they were asked to play in carbon reporting, the system of record, was always a stretch. At 2026 compliance levels, the seams have opened up.
Version control is the first thing to go. The moment two people are editing, or one person is working across quarters, the question "which file is the submitted version" becomes non-trivial. A file called NGER_FY25_FINAL_v4_reviewed_v2.xlsx is not a reassuring answer to an auditor tying a reported number back to its source.
Formula drift is next. Anyone who's maintained a reporting workbook across years has seen a factor get hard-coded somewhere, a sheet reference get pasted over, a conditional get broken when a row is inserted. These errors don't announce themselves. They produce numbers. Plausible, wrong numbers.
Then there's the provenance problem. A spreadsheet cell has a value. It does not, on its own, tell you which invoice produced that value, who extracted it, when, what factor was applied, and whether the factor's vintage matches the reporting year. An auditor under ASAE 3410 will ask those questions at line-item level. A spreadsheet answers them by pointing you at a folder of PDFs and asking you to match them up again. That's not an audit trail. That's a second audit.
Broken links and stale factor tables are the next layer. Whenever the NGA Factors workbook updates, every downstream spreadsheet that hard-copied values has to be re-pointed. Most never are. That's how you submit Scope 2 with last year's grid factor, a mistake that looks small until someone traces it.
And finally, the multi-framework problem. One entity in 2026 may need outputs under NGER (AR5, 31 October), AASB S2 (AR6, integrated with the financial report), Safeguard (baseline and compliance position), Climate Active (if certified), and GHG Protocol (for customer requests). Same activity data. Different factors, gases, scopes, boundaries, timings. Maintaining five workbooks for one business is how data diverges.
What "audit-ready automation" actually looks like
The phrase gets thrown around. Worth being specific about what it means in a 2026 Australian context, because most of what gets labelled automation is really just nicer dashboards on top of the same fragile workbook.
Audit-ready automation starts at the record, not the report. Every emission record in the system ties back to a single source document: an electricity bill, a fuel docket, a supplier statement, a waste manifest, a refrigerant service report. The record carries the source file, the extraction fields, the factor used, the factor's vintage, the GWP set applied, the person or process that created the record, the timestamp, and any subsequent edit history.
That chain has to be immutable in the auditable sense. Edits are allowed, but they're versioned and attributable. The original extraction and the current accepted value are both retrievable. When an ASAE 3410 practitioner samples a line item from your Scope 1 total and asks to see the evidence, the answer is a click, not a folder dig.
Per-gas breakdown is the second thing. NGER and AASB S2 both expect CO2, CH4, N2O, HFCs, PFCs, and SF6 to be identifiable in the underlying data. Aggregating straight to CO2-e and losing the constituents means you can't answer gas-level questions during assurance, and you can't switch GWP sets cleanly if a framework requires a different vintage.
Which brings up the third requirement. AR5 and AR6 at the same time. NGER uses AR5. AASB S2 uses AR6. If your system stores aggregated CO2-e with no way to re-apply a different GWP vector, you're locked into a single output. The records have to carry quantities by gas, and the reporting layer has to apply the correct GWP set per framework. That's not optional anymore. It's the cost of running both in parallel, which every NGER-reporting entity also falling into AASB S2 is now doing.
Fourth, variance narrative at the line level. When this year's diesel emissions are up 12% on last year, an auditor will ask why. "Activity grew" is a start. The better answer is a drill-down showing that a specific site's litres-per-month rose in Q3, tied to a known project ramp, with the underlying fuel dockets attached. Spreadsheets can produce that narrative if the analyst is willing to build it manually every quarter. Systems produce it structurally.
Fifth, Safeguard baseline tracking as a live view, not an annual reconciliation. If you're a Safeguard covered facility, knowing your position against your production-adjusted baseline at any point in the year is the difference between managing to the number and being surprised by it. TEBA declines of at least 1% per year compound, and the 30% ACCU surrender trigger is not forgiving.
Where Carbonly sits in this
We built Carbonly because we spent 18 years inside ASX-listed mining, resources, and energy companies, watching smart sustainability teams get buried in data entry. The maths was never the hard part. The hard part was pulling the activity data out of invoices, dockets, bills, supplier reports, emails, and shared folders, and holding it in a form an auditor could trust.
So that's what the product does. Our AI agents scan the activity as it arrives, through folder sync, email ingestion, supplier uploads, or direct drop-in, and create emission records with full provenance. Each record knows its source file, its extracted fields, its factor, its vintage, and its GWP set. The framework outputs (NGER under AR5, AASB S2 under AR6, Safeguard baseline position, Climate Active inventory, GHG Protocol view) all draw from the same underlying record layer. You configure the framework. The records don't move.
For construction specifically, which is the sector we see the most friction in, the pain isn't really fuel. It's material diversity. A single project site produces invoices for multiple concrete mixes with different embodied-carbon profiles, AdBlue, diesel at varying grades, hired plant with operator hours, sub-trade deliveries with mixed line items, and waste manifests by stream. Spreadsheets lose in that environment because the taxonomy is too wide and the documents are too inconsistent. Matching material lines to emission factors with five-tier logic that learns aliases over time is a different problem to reading a uniform utility bill.
We're currently working with a Tier 1 Australian construction company, starting at a single site, to prove out this approach before portfolio rollout. More on that when we and the customer are ready.
One thing we won't do is claim time-saving percentages. We think those numbers are mostly marketing. What we'll say, honestly, is that the tool moves clerical hours to analysis hours. The people who were transcribing numbers start reviewing them. The people who were reconciling workbooks start investigating variances. The work gets more valuable because less of it is typing.
What to do this quarter if you're Group 2 or an NGER reporter
If you're a sustainability lead or CFO inside a Group 2 entity, or an NGER reporter watching the Group 3 threshold approach, the useful thing to do in the next 90 days is not to buy software. It's to stress-test your evidence chain.
Pick five numbers from your most recent NGER submission. Any five. For each one, trace it back to a source document and ask:
- How long did the trace take? If it was more than a few minutes, that's your audit timing problem.
- Did the source document actually contain the value, or did it require re-derivation? If re-derivation, where's that working captured?
- What factor was applied? What vintage? Can you show it was correct for the reporting year?
- Has the value been edited since extraction? If yes, by whom, when, and why?
- If an ASAE 3410 practitioner asked for the same five records tomorrow, how would you deliver them?
The answers tell you how much runway you have before the 2026 workload hits. If the answers are "fine for now, painful at scale," that's the time to start looking at systems, because the scale is coming whether you're ready or not.
We're happy to walk through your specific situation. Email hello@carbonly.ai or join the waitlist at carbonly.ai. Per-project pricing (Small, Medium, Large, Enterprise) sized to scope, with a $100/month minimum workspace fee. We'll tell you honestly whether the problem you're solving is the one we solve.