Can carbon intensity fall while total emissions rise?

Yes. Carbon intensity can fall while total emissions rise when the activity used as the denominator grows fast enough. A business may emit less per finished unit while emitting more across the year. Report both results: the improving ratio does not establish a reduction in the total.

For example, an invented manufacturer moves from 10.0 to 8.8 kilograms of CO2 equivalent per finished unit, a 12% improvement. Its output grows by 25%, so its emissions rise from 100 to 110 tonnes CO2e. Both statements are true.

This explainer is for UK business teams preparing a carbon-performance update. The arithmetic applies more widely, but the example is not a complete inventory, a product carbon footprint, a reporting-law assessment or validation of a climate target.

Absolute emissions describe the quantity within a stated boundary and period. Intensity divides that quantity by a defined activity or economic measure. It therefore answers a different question: how much was emitted per unit of that measure?

The GHG Protocol Corporate Standard distinguishes absolute reporting from optional ratio indicators and warns that business context affects their interpretation. Its target guidance explains that an improving intensity result can coexist with rising absolute emissions. Source 1

The US EPA makes the same distinction: achieving an intensity target does not, by itself, establish that actual emissions fell. That is methodological context here, not a US rule being applied to UK businesses. Source 2

All figures below are fictional. Assume the same single facility, the same product specification and the same accounting methods in both years. The numerator includes the facility’s gross Scope 1 and location-based Scope 2 emissions, with no offset deduction. Scope 3 is excluded in both years. These are stipulated inventory inputs; no fuel, meter or emission-factor data was measured for this article.

The denominator is finished units passing the same acceptance criteria during each full calendar year. Count each finished unit once, including a reworked unit only when it passes. Keep emissions from scrap and rework in the numerator. Do not substitute units sold, production attempts or units shipped without redefining and checking the comparison.

Table units: emissions in tCO2e; output in accepted finished units; intensity in kgCO2e/unit.

Intensity = tonnes CO2e × 1,000 ÷ finished units.

For 2024: 100 × 1,000 ÷ 10,000 = 10.0 kgCO2e/unit. For 2025: 110 × 1,000 ÷ 12,500 = 8.8 kgCO2e/unit.

Three comparisons then describe the result:

• Intensity: (8.8 ÷ 10.0 − 1) × 100 = −12%.
• Output: (12,500 ÷ 10,000 − 1) × 100 = +25%.
• Absolute emissions: (110 ÷ 100 − 1) × 100 = +10%.

Reconcile them by multiplying the change factors: 0.88 × 1.25 = 1.10. Subtracting 12% from 25% would incorrectly give 13%. The factors multiply because total emissions equal intensity multiplied by output, using consistent units.

The reverse check also works: 8.8 × 12,500 ÷ 1,000 = 110 tCO2e.

For this example, the break-even output factor is 1 ÷ 0.88 = 1.13636… . A 12% intensity improvement therefore allows approximately 13.64% output growth before total emissions return to the original level.

At 10% output growth, the total would be 100 × 0.88 × 1.10 = 96.8 tCO2e, a 3.2% decrease. At the example’s 25% growth, it is 110 tCO2e, a 10% increase.

These are scenario calculations, not forecasts. The break-even result assumes the 8.8 kg rate remains applicable at the different output level. Fixed energy use, changed product mix or operating conditions may make that assumption unsuitable for planning. Use the calculation to understand the relationship; do not treat it as permission to increase emissions or evidence that a target has been met.

At the old intensity of 10.0 kg per unit, producing 12,500 units would correspond to 125 tCO2e. The 2025 total is 15 tonnes below that constant-intensity benchmark. It is still 10 tonnes above the 2024 total.

Calling the 15 tonnes “our emissions reduction” would change the comparison without telling the reader. It is a difference from a hypothetical benchmark, not a year-on-year inventory reduction. The arithmetic also does not show which action caused the lower intensity. Further evidence would be needed to attribute a saving to new equipment, operating changes or another intervention.

Keep any separate avoided-emissions comparison clearly distinct from inventory totals. A benchmark subtraction is not automatically a credit, an offset or a validated avoided-emissions claim.

Fictional record CI-018-v1 — annual facility performance

Comparison: 1 January–31 December 2025 against 1 January–31 December 2024; same facility and accepted-product definition. Both are full calendar years; the leap-day difference is not adjusted. This example does not claim equal operating hours.

Boundary: Gross Scope 1 plus location-based Scope 2 for the facility. Scope 3 excluded. No offsets subtracted. This ratio is not a cradle-to-grave footprint for an individual product.

Supported statement: “For the stated facility boundary, emissions per accepted finished unit fell 12%, from 10.0 to 8.8 kgCO2e. Output increased 25%, and total emissions increased 10%, from 100 to 110 tCO2e. Scope 3 is excluded from both years.”

Unsupported statement: “We cut our total carbon emissions by 12%.” The example does not support that claim.

Evidence to retain in a real record: The two versioned inventory totals, scope and method notes, production-register extracts, acceptance definition, reconciliation of excluded or duplicate units, calculation file and the exact proposed statement. None of these fictional records represents evidence collected from a real business.

Decision and owner: The sustainability coordinator presents the absolute increase alongside the intensity improvement. Operations investigates the drivers; the reporting owner checks the supporting evidence and wording before external use. An unexplained method or boundary change holds the performance conclusion.

If the product mix changes, a count of “units” may conceal a shift toward smaller or less energy-intensive products. Consider separately defined product groups or another defensible activity measure. Do not compare two companies merely because both publish kgCO2e per unit.

Revenue-based ratios require additional care. Price increases can enlarge the denominator without increasing physical output. The GHG Protocol discusses price, product-mix and inflation effects on monetary intensity metrics. Source 1 A revenue-based result should not be relabelled as improved production efficiency.

If an acquisition, outsourcing decision or material accounting change alters the comparison, assess the applicable base-year recalculation policy before interpreting the trend. The linked Canadian example explains a method; it does not establish a UK legal obligation.

If output is zero, the intensity ratio is undefined: report the emissions and explain the absence of output. Do not enter zero intensity. If the earlier intensity is zero, the percentage-change and break-even calculations above are also inapplicable.

Start with one reporting period. Keep the total, denominator, intensity and comparison basis together, then write a sentence that preserves all four. That gives the reader a usable result and gives the business a clear question to investigate: what explains the change in total emissions?

Source 1: WRI/WBCSD, GHG Protocol Corporate Accounting and Reporting Standard, revised edition 2004; chapter 9, printed pages 65–67, and chapter 11, printed pages 76–77. Relevant sections and the current Corporate Standard landing page read on 18 October 2026. This article uses its ratio guidance, not every requirement for a complete GHG inventory.

Source 2: US Environmental Protection Agency, Target Setting, section “Absolute versus Intensity-Based Targets”; page updated 26 February 2026, read on 18 October 2026. Guidance, not a determination of UK reporting duties.

Worked figures, reporting note and sensitivity checks are original QSE Advisory editorial analysis using fictional inputs. Prepared with AI-assisted research and drafting. Author and publisher: QSE Advisory, office@qseadvisory.com. Reviewer and Assessor: Laraib Naveed Ahmad.

Spotted an issue? Contact our editorial desk.