Carbon Accounting Explained: How Companies Measure and Reduce Emissions

Microsoft’s electricity consumption nearly tripled between 2020 and 2024, from 11 terawatt-hours to approximately 30 TWh. Under carbon accounting methodologies, its reported emissions declined. Google’s data centre electricity use jumped 27% in a single year. Its sustainability report highlighted emission reductions. The divergence stems from choices embedded in how companies measure climate impact.

The divergence stems from choices embedded in carbon accounting, the discipline that has become the language of corporate climate performance. As mandatory reporting takes effect across 40 countries and investors controlling more than €65 trillion demand emissions disclosure, those choices increasingly carry legal consequences. In February 2025, Apple faced a class action lawsuit over carbon-neutral product claims. Delta Air Lines and British American Tobacco confront similar litigation. What companies count, and how they count it, now matters to securities regulators and trial lawyers alike.

What Gets Measured

Carbon accounting divides emissions into three categories, a framework that sounds straightforward until you attempt implementation at scale.

Scope 1 captures what you burn directly: fuel in company vehicles, natural gas in boilers, coal in owned power plants. A delivery van consuming 1,000 litres of diesel releases approximately 2,700 kilograms of CO2. Track fleet fuel consumption, apply the emission factor, and you have Scope 1 transport emissions.

Scope 2 covers electricity and heat purchased from external providers. You don’t burn the coal at the power station, but you’re responsible for the emissions created to generate the electricity you consume. The calculation requires knowing your consumption in kilowatt-hours and the emission factor for your grid. A factory in coal-dependent Poland faces different Scope 2 emissions than an identical facility in hydro-powered Norway, even with identical electricity consumption.

Scope 3 encompasses everything else. Raw materials arriving at your factory, business travel, employee commutes, contracted waste disposal, customer use of your products, eventual disposal. For most companies, Scope 3 accounts for 70% to 90% of total emissions. It’s also where measurement becomes genuinely difficult.

The Greenhouse Gas Protocol, established in 2001 by the World Resources Institute and World Business Council for Sustainable Development, provides the framework most organisations follow. Its persistence as the global standard stems from comprehensive industry coverage whilst maintaining consistency in reporting. Think of it as generally accepted accounting principles for emissions, though considerably younger and still evolving as regulators tighten requirements.

How the Calculation Actually Works

The mechanics follow a logic familiar to any financial analyst. Identify emission sources, collect activity data, apply conversion factors, aggregate results.

Emission factors are coefficients maintained in databases by organisations including the US Environmental Protection Agency, the UK’s Department for Environment Food & Rural Affairs, and the Intergovernmental Panel on Climate Change. These factors relate the quantity of pollutant released to an activity, expressed as mass of gas per unit of emissions-producing activity. They vary by geography, technology, and fuel type, and get updated as energy systems evolve.

A simple example: Your London office consumed 50,000 kWh of electricity last quarter. The UK grid’s emission factor averages around 0.23 kilograms CO2 per kWh. Multiply consumption by factor: 50,000 × 0.23 = 11,500 kilograms, or 11.5 metric tons of CO2. That’s your Scope 2 emissions from electricity.

Scale this across thousands of facilities, multiple fuel types, international operations with different grid intensities, and the spreadsheet chaos becomes apparent. A manufacturing conglomerate might track diesel consumption at a Texas plant, electricity use across facilities in 15 countries, natural gas for heating in Canadian warehouses, jet fuel for corporate travel, and shipping emissions from contract logistics providers. Each requires appropriate emission factors, often sourced from different databases, applied consistently across reporting periods.

This is why carbon accounting software has emerged as essential infrastructure rather than optional tooling. The market is projected to grow from roughly $9 billion in 2024 to over $33 billion by 2029, growth driven by regulatory mandates making manual processes untenable.

The Methodology Problem

Scope 2 presents a peculiar challenge. The GHG Protocol permits two calculation methods: location-based and market-based. This choice matters enormously.

Location-based methodology uses the average emission intensity of the local electricity grid. If you operate a data centre in Virginia where the grid derives power from coal and natural gas, your emissions reflect that reality regardless of what you claim to purchase.

Market-based methodology allows companies to claim credit for purchasing renewable energy certificates. These are paper instruments representing the environmental attributes of green power. A company can buy certificates for wind energy generated in one location whilst its actual facilities draw electricity from fossil-heavy grids elsewhere, then report emissions as if it consumed the renewable power directly.

Microsoft’s location-based Scope 2 emissions more than doubled from 4.3 million metric tons CO2 in 2020 to nearly 10 million in 2024. Their market-based figure shows a decline. The entire divergence comes from renewable energy certificates. Replace the market-based Scope 2 figures with location-based ones and add Scope 3, and Microsoft’s actual emissions in 2024 would be around 25.2 million metric tons CO2, not the 15.5 million they report using market-based methodology.

This isn’t fraud. Both methods are legitimate under current frameworks. Most companies report the more favourable market-based number prominently whilst burying location-based figures in appendices. But the gap between methodologies creates two fundamentally different pictures of environmental impact.

Google’s sustainability reports highlight data centre emission reductions, claims made plausible through market-based accounting. Location-based figures tell a different story. Data centre electricity use increased 27% between 2023 and 2024 alone, and location-based emissions climbed 22% from baseline.

When Accounting Becomes Litigation

In February 2025, Apple was sued in a putative class action claiming its carbon-neutral Apple Watch assertions were false and misleading. The complaint centres on whether offset projects Apple relied upon delivered genuine emissions reductions or provided convenient accounting cover. Similar litigation against Delta Air Lines and British American Tobacco signals that carbon accounting has crossed from voluntary disclosure into securities law territory.

Get the methodology wrong, misrepresent your footprint, and you’re potentially violating consumer protection statutes. The regulatory landscape reinforces this shift. California’s Climate Corporate Data Accountability Act, enacted in October 2023, imposes strict disclosure requirements. The European Union’s Corporate Sustainability Reporting Directive mandates location-based reporting alongside market-based figures. France already requires location-based methodology by regulation.

Financial institutions face particularly sharp scrutiny. The Partnership for Carbon Accounting Financials now counts over 300 banks and asset managers representing more than €65 trillion in assets. These institutions must disclose financed emissions, the carbon intensity of loan portfolios and investment holdings. A mortgage book’s climate risk now sits alongside its credit risk in board presentations.

For CFOs, carbon data increasingly carries the same audit risk as financial statements. The tolerance for creative accounting is narrowing whether businesses prefer it or not.

The Scope 3 Labyrinth

Measuring Scope 3 emissions forces companies into uncomfortable dependency on suppliers who may lack measurement capability entirely. You’re asking a Vietnamese steel provider to quantify carbon intensity when they might not track electricity consumption by production line.

The challenges extend beyond data collection into questions of what counts as legitimate reduction. Shell’s carbon offset projects in China illustrate the problem. An investigation revealed questionable accounting in rice farming initiatives claiming credit for convincing farmers to use sustainable flooding methods. Studies showed the practice was already used in around 41% of rice paddies around China in 2018, the year the offsetting projects began.

This exemplifies the additionality problem haunting carbon accounting. Did the project cause emissions reductions that wouldn’t have occurred otherwise, or did it document existing practices and claim credit? It’s not a question of dishonesty but of counterfactuals impossible to prove definitively.

Research on 26 avoided-deforestation projects across six countries in the UN’s REDD programme found that most projects have not significantly reduced deforestation. For projects showing reductions, actual impact was substantially lower than claimed. The pattern repeats across offset categories: renewable energy projects in regions already transitioning away from fossil fuels, efficiency improvements that would have occurred regardless, forest preservation in areas facing no realistic threat.

An airport expansion campaign failed to include emissions from flights in its environmental accounting. Britain’s advertising regulator banned the ads in June 2024. The complaint came from Adfree Cities, whose co-director compared the omission to advertising cigarettes as healthy.

The 2024 Corporate Climate Responsibility Monitor examined leading corporations’ climate strategies and found not one received unqualified approval. Companies routinely exclude Scope 2 and 3 emissions when setting reduction targets whilst leaning on offsets rather than operational changes.

Software as Strategic Infrastructure

Persefoni exemplifies the enterprise approach to carbon accounting. Launched in 2020 for financial institutions managing Partnership for Carbon Accounting Financials requirements, it expanded to serve corporations facing complex regulatory demands. The platform uses AI to detect anomalies in reported data, errors that compound through annual reports and eventually attract regulatory attention.

For finance teams processing thousands of transactions across multiple jurisdictions monthly, automation transforms three-week quarterly exercises into overnight calculations with audit trails satisfying external verification. The value isn’t speed but accuracy and defensibility when regulators or litigants scrutinise disclosure.

Watershed treats measurement as inseparable from reduction planning. After acquiring VitalMetrics and its emissions factor database, the platform strengthened its ability to survey suppliers systematically, addressing Scope 3 data gaps. When organisations can see where emissions concentrate across operations, reduction becomes strategic rather than symbolic. The integration of collection with decarbonisation strategy appeals to companies seeking operational intelligence rather than mere compliance documentation.

Microsoft Sustainability Manager offers native integration for businesses embedded in that ecosystem. The tool pulls data from Azure applications and uses optical character recognition to extract emissions data from paper documents. It addresses multiple environmental metrics simultaneously, though analysts note it works best for organisations already committed to the Microsoft technology stack.

A systematic literature review published in Green Finance examined 27 journal articles and found that utilising robust carbon accounting platforms positively influenced the sustainable development index amongst various international companies. Evidence suggests proper measurement infrastructure drives operational changes rather than merely documenting emissions.

Regulatory Convergence

In July 2024, the Task Force on Climate-related Financial Disclosures was replaced by Sustainability Disclosure Standards aligned with International Financial Reporting Standards Foundation guidelines. The shift signals integration of climate data into mainstream financial reporting rather than parallel sustainability documentation.

For accountants, this convergence matters. Emissions data now sits alongside revenue and profit in annual reports. Investors query carbon intensity per production unit the same way they examine operating margins. Understanding how environmental performance shapes valuation has migrated from corporate social responsibility departments into core financial analysis.

The convergence brings standardisation but exposes contradictions embedded in competing methodologies. As litigation proliferates and regulators demand transparency, room for divergent reporting approaches contracts.

Implementation Reality

According to TINA.org, just 6% of US greenwashing cases involve false or misleading carbon reduction claims, a percentage reflecting evidentiary challenges in offset litigation. Proving a forest preservation project in Indonesia wouldn’t have occurred without corporate funding requires counterfactual analysis courts struggle to evaluate.

Climate intelligence technology is changing how attorneys approach these cases, providing satellite data and verification tools withstanding judicial scrutiny. Companies building sophisticated measurement capabilities do so partly to identify efficiency opportunities but increasingly to establish defensible positions against litigation.

Third-party verification has become standard for credible reporting, adding cost but providing assurance that matters when legal teams assess disclosure risk. Data quality remains the constraint. Advanced platforms cannot solve supplier measurement gaps. If suppliers in developing markets lack basic infrastructure for tracking energy consumption, even sophisticated software relies on industry averages and estimates.

The methodology becomes only as good as the weakest data point, and Scope 3 chains stretch across jurisdictions with wildly varying measurement capacity. Securing value chain engagement requires sustained effort and often financial support for suppliers lacking measurement infrastructure.

Growth Versus Reduction

Microsoft’s Scope 3 emissions increased 30.9% from their 2020 baseline, driven primarily by data centre construction and embodied carbon in semiconductors, servers, and building materials. The company maintains its commitment to becoming carbon negative by 2030. One industry analyst questioned how achieving carbon negativity by 2030 represents a practical possibility, particularly considering the company is at the start of its AI infrastructure expansion.

Companies can purchase offsets, invest in removal technology, and deploy renewable energy certificates. Whether these mechanisms represent genuine climate action or sophisticated accounting manoeuvres depends on methodology, verification, and regulatory tolerance for divergent approaches.

Total energy-related carbon dioxide emissions rose 0.8% in 2024, reaching an all-time high of 37.8 billion metric tons. The trajectory suggests measurement, however sophisticated, achieves little without operational changes reducing actual atmospheric loading.

What This Means for Finance Professionals

Carbon accounting represents territory where technical competence creates competitive advantage. As reporting requirements tighten across 40 countries and stakeholder scrutiny intensifies, organisations building measurement capabilities navigate regulatory complexity whilst identifying efficiency opportunities that reduce both emissions and operating costs.

The discipline continues evolving rapidly. Emerging guidance addresses avoided emissions, forward-looking metrics, and integration of climate risk into financial planning. The question facing businesses isn’t whether to invest in carbon accounting infrastructure but how quickly they can build capabilities meeting regulatory requirements whilst delivering strategic value.

The frameworks are changing. Regulatory tolerance for the gap between market-based claims and location-based reality is narrowing. The diesel van calculation remains straightforward: consumption multiplied by emission factor. Scaling that process across global operations, satisfying regulators in 40 jurisdictions, and avoiding securities litigation requires infrastructure, expertise, and increasingly, legal advice.

Microsoft consumed nearly three times the electricity whilst claiming emissions declined. Google’s location-based data centre emissions climbed 22% whilst reported figures showed improvement. These divergences stem from legitimate accounting choices permitted under current frameworks. Those permissions are expiring.