Introduction
As organisations intensify efforts on climate change, accurately measuring and managing greenhouse gas emissions has become a core commercial capability. Emissions are conventionally categorised into Scopes 1, 2, and 3 according to whether they are direct, energy-related, or spread across the value chain. Alongside these, a further concept has gained attention: avoided emissions, informally called “Scope 4”, which addresses emissions prevented elsewhere through the adoption of a product, service, or solution.
“Scope 4” is not an official scope. The GHG Protocol defines three scopes, and avoided emissions sit outside the corporate inventory entirely. They must be reported separately and can never be subtracted from Scopes 1, 2, or 3 to produce a lower net figure. A company reporting 500,000 tonnes of emissions and one million tonnes avoided still emitted 500,000 tonnes.
What are avoided emissions?
Avoided emissions are the difference between the emissions that occurred and the emissions that would have occurred had a reference solution been used instead. Where Scopes 1 to 3 measure emissions attributable to an organisation, avoided emissions quantify a comparative benefit occurring in someone else’s inventory — typically a customer’s.
The distinction matters for how the figure behaves. An inventory figure is attributional: it allocates real, measured emissions. An avoided emissions figure is consequential: it compares reality against a counterfactual that did not happen and therefore cannot be measured, only argued.
How avoided emissions differ from the three scopes| Aspect | Scopes 1, 2, and 3 | Avoided emissions |
|---|---|---|
| What is measured | Emissions that occurred | Difference against emissions that would have occurred |
| Approach | Attributional | Consequential |
| Where the effect occurs | Within the company or its value chain | Usually in the customer’s or society’s inventory |
| Verifiability | Verifiable against records and meters | Depends entirely on the reference scenario chosen |
| Reporting treatment | Reported as the inventory | Reported separately, never netted against the inventory |
| Target setting | Basis for science-based targets | Not accepted toward reduction targets |
Background and development
The three scopes were introduced by the Greenhouse Gas Protocol, developed by the World Resources Institute (WRI) and the World Business Council for Sustainable Development (WBCSD), which established the standardised approach organisations use to measure and report emissions consistently.
The avoided emissions concept developed later, driven by companies whose products reduce emissions in use — renewable technology, efficiency solutions, insulation, software — and whose corporate inventories therefore understate their climate contribution. WBCSD has published guidance on avoided emissions, and CDP and other frameworks have addressed how such claims should be presented.
There is currently no dedicated ISO standard for avoided emissions. The methodological reference points are ISO 14064-2 for project-level emission reductions, ISO 14067 and ISO 14040/14044 for the product footprints the comparison rests on, and the GHG Protocol Project Protocol for comparative quantification.
The mathematics
The core calculation is a difference between two life cycle footprints delivering the same function. Everything difficult about avoided emissions is contained in defining the second term.
AE — avoided emissions, in tonnes CO₂e. Ereference — life cycle emissions per functional unit of the reference solution. Esolution — life cycle emissions per functional unit of the solution being assessed. Q — quantity of functional units supplied in the reporting period.
Both terms must cover the full life cycle, not the use phase alone. A solution that reduces operational emissions while carrying a higher manufacturing footprint may avoid far less than its use-phase difference suggests, or nothing at all.
Production, use, and end-of-life for each option. Omitting the production term is the most common way an avoided emissions figure is overstated.
The attribution problem
A solution rarely delivers its benefit alone. Where several actors contribute — the technology provider, the installer, the financier, the customer who chose it — the same avoided tonne can be claimed several times over. An attribution factor addresses this.
α — the share of the outcome attributable to the reporting organisation, between 0 and 1. There is no settled convention for setting it, which is precisely why it must be disclosed and justified rather than assumed to be 1.
Additionality and counterfactual correction
Part of the market would have adopted a lower-emission alternative regardless. Claiming the full difference against the incumbent technology credits the reporting company with change that was already happening.
s — market shares of each alternative that would plausibly have been chosen, summing to 1. A market-average reference is more defensible than comparing against the worst available technology.
Time and durability
The reference emission factor is not constant. As grids decarbonise, Eref falls year on year, so an efficiency or electrification solution avoids progressively less over its life. Holding the reference factor fixed at year one systematically overstates the total.
Worked example: cloud migration
A technology provider migrates a customer from on-premise data centres to cloud infrastructure. Figures below are illustrative and rounded, to show the calculation structure.
Reference scenario — on-premise data centre
- Annual energy consumption: 1,000 GWh = 1 × 109 kWh
- Grid carbon intensity: 1,000 g CO₂e/kWh
Grams are converted to tonnes by dividing by 106, not multiplying.
Solution scenario — cloud infrastructure
- Annual energy consumption: 200 GWh = 2 × 108 kWh
- Effective carbon intensity after renewable supply: 100 g CO₂e/kWh
- Supporting infrastructure, embodied hardware, and logistics: 180,000 t CO₂e / year
| Scenario | Annual emissions (t CO₂e) |
|---|---|
| Reference — on-premise data centre | 1,000,000 |
| Solution — cloud infrastructure | 200,000 |
| Avoided emissions | 800,000 |
The 800,000 tonnes is a comparison against one chosen reference. Change the reference — a modern efficient on-premise facility rather than an ageing one, or a grid intensity of 400 g rather than 1,000 g — and the result falls substantially. The number is a statement about the comparison, not a measurement of the world, and it does not reduce the provider’s own reported emissions by a single tonne.
Practical examples across industries
Energy
A provider installs solar generation displacing grid electricity. Avoided emissions are the difference between the grid emission factor and the life cycle factor of the solar generation, multiplied by MWh delivered.
The solar factor is not zero — panel manufacturing, mounting, inverters, and end-of-life all carry emissions, typically an order of magnitude below fossil generation but not nil.
Transport
An electric vehicle replaces an internal combustion vehicle. The comparison must be over the full lifetime and must include battery manufacturing, which loads the EV heavily at production and is repaid over distance driven.
d — lifetime distance. c — electricity consumption per km. The breakeven distance is where the EV’s higher manufacturing footprint has been repaid, and it varies widely with grid intensity.
Agriculture
Precision farming reduces fertiliser application, fuel use, and passes across the field. The nitrogen effect dominates, because fertiliser carries both a production footprint and field N₂O emissions.
Buildings
Insulation, heat pumps, and glazing reduce operational heating and cooling demand. Avoided emissions accrue over decades, which makes the declining grid factor over time material to the calculation.
Industrial equipment
A high-efficiency motor, compressor, or heat exchanger reduces energy per unit of output against the equipment it replaces or against the market-average alternative.
P — useful power output. h — annual operating hours. η — efficiency of each option.
Materials and chemicals
A lower-carbon cement, steel, or polymer displaces a conventional grade at equal performance. The functional unit must reflect performance equivalence — comparing per tonne rather than per unit of structural function will mislead if the materials differ in strength.
Software and digital services
Route optimisation, remote monitoring, or predictive maintenance reduce fuel, travel, or material use. These are the hardest claims to substantiate, because the counterfactual behaviour is not observable and rebound effects are common.
Where these claims go wrong
- 1Netting against the inventory. Presenting emissions minus avoided emissions as a net figure. This is the most serious error and invalidates the whole disclosure.
- 2Choosing a flattering reference. Comparing against the worst available technology rather than the realistic market alternative.
- 3Omitting the production footprint. Counting use-phase savings while ignoring the emissions of making the solution.
- 4Ignoring rebound effects. Efficiency that lowers cost can increase consumption, returning part of the saving.
- 5Double counting. Manufacturer, installer, financier, and customer each claiming the same avoided tonne.
- 6Holding factors static. Assuming a decarbonising grid stays at today’s intensity for twenty years.
- 7Claiming avoided emissions while growing absolute emissions. A company can genuinely enable reductions elsewhere and still be increasing its own footprint. Both must be visible.
How to report avoided emissions credibly
- Report separately from the inventory, with the Scope 1, 2, and 3 figures presented alongside and never adjusted.
- State the reference scenario explicitly and justify why it represents the realistic alternative.
- Use full life cycle footprints for both options, prepared under ISO 14067 or the GHG Protocol Product Standard.
- Disclose the attribution factor and its basis.
- Present a sensitivity analysis showing how the result moves with the reference scenario and key factors.
- Seek third-party verification, drawing on ISO 14064-2 and ISO 14064-3 practice.
- Do not use avoided emissions toward science-based targets, which are set on the inventory alone.
Future developments
- Methodological standardisation. Consistent rules for reference scenarios, attribution, and additionality are the missing piece; work is under way but not settled.
- Position within disclosure frameworks. How avoided emissions sit alongside mandatory reporting under CSRD, IFRS S2, and CDP is still being defined.
- Investor use. Climate solution revenue and avoided emissions are increasingly used to identify transition-aligned companies, which raises the stakes on comparability.
- Verification. As claims grow more commercially significant, independent assurance is likely to become the expectation rather than the exception.
Conclusion
Avoided emissions capture something real that the three scopes cannot: a company whose products reduce emissions in customers’ hands has a climate contribution its own inventory understates. Quantifying that contribution is legitimate, useful, and increasingly expected by investors.
It is also the easiest climate figure to overstate, because the comparison scenario is chosen rather than measured. The discipline is straightforward: full life cycle boundaries on both sides, a defensible reference, a stated attribution factor, sensitivity analysis, and separate reporting that never touches the inventory. Presented that way, avoided emissions strengthen a company’s position. Presented as a net reduction, they undermine it.
Not sure which service fits your project?
Tell us your scope, timeline, and target outcome — DEISO will point you to the right service or scope the work directly.
