How to calculate Scope 3 emissions: the calculation methodology, the four methods, and the data each one needs

Last updated August 2026. Most companies do not get stuck on the formula. They get stuck two steps earlier, deciding what counts as activity data for a category they have never measured, and two steps later, defending a factor choice to an assurance provider. This page walks the whole calculation: the methodology, the four methods and when each is defensible, the data every one of the 15 categories needs, the US factor sets to use, and where a spreadsheet stops being enough. If you want to see the classification step run on your own records, the demo below takes an invoice or a supplier list and drafts the scope, the category and the factor line by line, which is the part our carbon accounting software automates.

Scope 3 emissions are calculated as activity data multiplied by an emission factor, expressed in tonnes of carbon dioxide equivalent. Activity data is the quantity of the thing that happened: kilograms of steel purchased, tonne-kilometres of freight, dollars spent with a supplier, kilowatt-hours consumed in a leased building. The emission factor converts that quantity into emissions.

The reason Scope 3 feels harder than Scope 1 and Scope 2 is not the arithmetic. It is that you rarely own the activity data. Your gas meter is yours. The energy your supplier burned making a component is not, so you either ask them for it, estimate it from the physical quantity you bought, or estimate it from what you paid. Those three answers are the methods, and the GHG Protocol expects you to be explicit about which one produced each number.

The standards that govern this are the GHG Protocol Corporate Value Chain (Scope 3) Accounting and Reporting Standard, published in 2011, and its companion Technical Guidance for Calculating Scope 3 Emissions, which works through methods category by category. The corporate-level rules those sit on top of are covered in our note on the GHG Protocol Corporate Standard.

The GHG Protocol describes four calculation approaches, most clearly for category 1 but applied by analogy across the value chain. They run from the most primary data to the most secondary. Accuracy improves as you move up the table, and so does the effort, which is why almost every real inventory is a mix.

One property of the spend-based method catches people out. Because the factor is emissions per dollar, the calculated number moves with price. Negotiate a 10% discount on the same physical goods and your reported emissions fall 10%, which is obviously not a real reduction. That is fine for screening and indefensible as a reduction claim, and it is the single reason companies migrate off spend for their material categories. The trade-off is worked through in spend-based vs activity-based emissions.

Each category has its own activity data and its own default method. The table below is the practical version: what you multiply, and what you multiply it by. It is the fastest way to find out which categories you can calculate this month and which need a data collection project.

For a US reporter, most of what you need is published free by federal agencies, and using government factor sets removes an entire argument with your assurance provider about provenance.

Two housekeeping rules matter more than they sound. Record the factor version against every number, because factor sets are revised and a figure calculated on version 1.2 is not comparable with one calculated on 1.3. And keep the global warming potential set consistent across the inventory: the EPA supply chain factors are built on AR5 values, while CDP and several other frameworks now expect AR6, where fossil methane is 29.8 and nitrous oxide is 273. Mixing them silently is a finding waiting to happen.

Yes, and for a first screening inventory a spreadsheet is the right tool. Export a year of accounts payable, add a column for the NAICS code, VLOOKUP the EPA spend factor, multiply, pivot by category. That produces a usable screen in a few days and tells you where to spend real effort.

Where the spreadsheet fails is the second year. Three things break at once. The classification is manual, so ten thousand lines take a person a week and the mapping decisions live in someone's head rather than in the file. Factor versions drift, because a new EPA release means re-running the lookup and nothing records which version produced last year's number. And the audit trail is missing: a total in a pivot table cannot be traced back to the invoice that produced it, which is precisely what limited assurance samples for. That last point is why so many inventories that are arithmetically correct still fail a first assurance engagement, and the difference between assurance levels is set out in limited vs reasonable assurance.

Accurate enough that the number would not change a decision if it were refined, and documented well enough that a third party can follow how you got it. No framework demands precision for its own sake. The GHG Protocol asks for the best data reasonably available and for transparency about what you used, which means a spend-based estimate clearly labelled as such is compliant, while a physical-unit calculation with an undocumented factor is not.

What the frameworks do care about is completeness and consistency. CDP wants base year emissions for every category you have marked relevant and calculated, plus real calculations for at least one, which is a screening requirement rather than a precision requirement. California SB 253 phases Scope 3 in from 2027 with limited assurance arriving from 2030. IFRS S2 requires you to consider all 15 categories and disclose which ones your figure includes. The requirement common to all three is that next year's number is built the same way as this year's, because a method change that moves the total is a restatement, not an improvement. Framework by framework detail is on Scope 3 emissions reporting.

A spend-based screen across all 15 categories takes days once you have a clean accounts payable export. A first full inventory with mixed methods typically runs one to three months, and most of that is data collection rather than calculation: chasing freight tonne-kilometres, running a commuting survey, getting the lease register into a usable state. Supplier-specific data for material suppliers is a multi-cycle project, because you are asking other companies to measure something on your timetable.

The part that compresses fastest is classification. Assigning a scope, a category and a factor to every purchase line is mechanical, repetitive and enormous, which is exactly the shape of work worth automating. Our approach reads the underlying documents, drafts the mapping line by line with a stated confidence, routes the low-confidence tail to a human, and keeps every figure attached to the record that produced it. The starting point for most companies is the method in calculating Scope 3 emissions from spend, and the wider picture is on Scope 3 emissions.