Two ways to build the same number
Most Scope 3 inventories start with a gap: a company knows how much it spent on purchased goods, contracted logistics, or leased equipment, but not the physical detail behind those transactions. The GHG Protocol's Corporate Value Chain (Scope 3) Standard gives companies a data quality hierarchy to work through that gap, running from spend-based estimation at the weakest end to supplier-specific primary data at the strongest, with average-data and hybrid approaches in between. Understanding what each method actually does mathematically — not just where it ranks — is what lets a measurement team decide where to spend limited data-collection effort.
Spend-based estimation: dollars times a factor
Spend-based estimation multiplies the amount spent on a good or service by an emission factor expressed per unit of currency — kilograms of CO2e per dollar spent on, say, industrial machinery or professional services. These factors typically come from environmentally extended input-output (EEIO) models that map national economic sectors to their average emissions intensity.
The appeal is obvious: if procurement systems already tag purchases by category, a full Purchased Goods and Services or Capital Goods inventory (Categories 1 and 2) can be built in days without contacting a single supplier. The cost is precision. Spend-based factors are averaged across an entire economic sector, so a company buying premium, low-carbon aluminum and a company buying standard aluminum get charged the same emissions intensity per dollar if they're coded to the same sector. Price changes also distort the number over time — if a supplier raises prices without changing production emissions, the spend-based estimate rises anyway, even though nothing about the physical footprint moved. This is why the GHG Protocol places spend-based at the weak end of the hierarchy: it is useful for establishing a baseline and for screening which categories are material, but it does not reward genuine supplier-level decarbonization.
Activity-based and average-data estimation: physical quantity times a factor
Activity-based estimation (the Standard groups this under "average-data method") multiplies a physical quantity — tonnes of steel purchased, liters of fuel combusted, kilometers of freight moved, kilowatt-hours of electricity used by a sold product — by an emission factor expressed per unit of that activity, drawn from industry averages, national statistics, or life-cycle databases.
This is a meaningful step up in accuracy because it strips out price volatility and reflects the actual scale of physical activity. A ton of steel is a ton of steel regardless of what a supplier charged for it. It's the natural fit for categories where physical quantities are already tracked for other reasons: fuel and energy-related activities (Category 3), upstream and downstream transportation (Categories 4 and 9), waste generated in operations (Category 5), employee commuting (Category 7), and use of sold products (Category 11), where energy consumption over a product's life is a physical quantity, not a spend.
The limitation is that the emission factor is still an industry or regional average, not specific to the actual supplier or process. Two suppliers producing the same tonnage of the same material can have very different actual carbon intensities depending on their energy mix and process efficiency, and average-data estimation collapses that difference to a single sector-wide number.
Hybrid and supplier-specific: closing the gap
The hierarchy's next rung, hybrid methods, combines spend-based and activity-based data where full physical or supplier data isn't available for every input — for example, using activity-based factors for the categories where a company has volume data and spend-based factors as a fallback for the rest of the same category. This lets a company avoid the false precision of applying one method uniformly when its underlying data is genuinely mixed in quality.
At the top of the hierarchy sits supplier-specific data: primary emissions figures reported directly by suppliers, ideally tied to the specific product or process purchased rather than a sector average. This is the only method that actually reflects a supplier's real operational choices — a lower-carbon energy grid, an efficiency retrofit, a different feedstock. It's also the hardest to obtain at scale, since it depends on supplier willingness and capability to measure and disclose their own emissions, which is precisely the gap that supplier engagement programs are built to close.
Choosing a method in practice
In a functioning inventory, all four methods coexist. Spend-based estimation is appropriate for materiality screening and for low-priority categories where the cost of better data isn't justified by the emissions at stake. Activity-based estimation should replace spend-based wherever physical activity data is already collected for other purposes — it is rarely more expensive to obtain and is meaningfully more accurate. Hybrid approaches are the honest answer for categories where data quality is uneven across suppliers rather than uniformly good or bad. And supplier-specific data collection should be prioritized for the highest-emitting categories and the suppliers whose actual performance a company most needs to influence, since that is the only method that will show measurable improvement when a supplier genuinely decarbonizes.
The practical takeaway is that upgrading data quality is not an all-or-nothing exercise. A company can move category by category, starting with spend-based totals to find where the emissions actually sit, then targeting activity-based or supplier-specific data collection at those hot spots first. That sequencing — screen broadly, then refine where it matters — is usually a better use of a measurement team's time than trying to gather primary data everywhere at once.