Accuracy is non-negotiable in carbon accounting. EcoLink's Spend-to-Carbon AI engine does not rely on generic global averages. Instead, every Xero transaction is automatically classified and matched against the latest National Greenhouse Accounts (NGA) Factors, published annually by the Australian Department of Climate Change, Energy, the Environment and Water (DCCEEW).
This localised approach guarantees that the kg CO₂e generated for electricity, transport, and procurement accurately reflects the Australian market context — providing a highly defensible and accurate carbon baseline. A unit of electricity consumed in Queensland has a different emission intensity than in Tasmania, and our engine accounts for this automatically.
EcoLink uses the NGA 2023–24 edition — the most current factors available from DCCEEW. When the 2024–25 edition is published, all calculations update automatically. Historical reports retain the factors that were current when they were generated.
For SMEs without direct meter readings, the most practical — and AASB S2-accepted — method is spend-based estimation: multiply the dollar amount of a purchase by an emission intensity factor for that category.
Transaction: "BP Station Sydney · $150"
AI classifies to: Transport — Petrol (Scope 1)
NGA Factor applied: 1.087 kg CO₂e per AUD $1
$150 × 1.087 = 163 kg CO₂e
EcoLink's AI reads the transaction description, identifies the correct NGA category using keyword matching and large language model classification, then applies the factor automatically. No spreadsheets, no manual lookups.
Keyword matching
Fast rule-based matching against known supplier names and transaction patterns (e.g. "BP", "Toll", "Origin Energy").
Groq llama-3.3-70b
If keyword matching is inconclusive, the transaction is passed to a large language model for semantic understanding.
Gemini fallback
A second AI model is used as a fallback for further disambiguation.
Confidence threshold
≥ 60% confidence → auto-classified. 40–59% → flagged for human review. Below 40% → escalated as unmatched.
| Scope | Category |
|---|---|
| Scope 1 | Stationary Energy |
| Scope 1 | Transport — Fuel Combustion |
| Scope 1 | Fugitive Emissions |
| Scope 2 | Purchased Electricity |
| Scope 3 | Business Travel |
| Scope 3 | Freight & Logistics |
| Scope 3 | Purchased Goods & Services |
| Scope 3 | Waste Generated in Operations |
Classified
AI confidence ≥ 60%. Auto-approved.
Needs Review
40–59% confidence. Human review queue.
No Factor
No NGA match. Manual assignment needed.