Primary vs spend-based emissions data: the trade-off explained

What is the difference between primary and spend-based emissions data?

Primary data is an emissions figure reported by the organisation the activity belongs to: a supplier, a portfolio company or a group entity publishing its own Scope 1 and Scope 2, or a product carbon footprint for the item you bought. Spend-based data estimates the same activity by multiplying the money spent by an emissions factor for the relevant industry sector. One figure is measured by someone, the other is modelled by you.

The GHG Protocol Scope 3 Standard names four calculation methods for Category 1, purchased goods and services: the supplier-specific method, the hybrid method, the average-data method and the spend-based method. Primary and spend-based sit at the two ends of that list. The hybrid and average-data methods sit between them, taking reported figures where they exist and physical activity data or sector averages where they do not. Framing the decision as two options rather than four is what leads teams to treat a whole category as all or nothing.

The same four methods apply to Category 2, capital goods. Category 15, investments, follows PCAF instead, where the equivalent question is which data quality score a holding can support.

What counts as primary emissions data?

Primary data is anything the reporting organisation did not model itself: a counterparty's disclosed Scope 1 and Scope 2, a verified inventory, a CDP response, an EPD or a product carbon footprint for a specific line item. It carries the one thing an average cannot, which is the performance of that particular organisation.

Two organisations selling near-identical products at near-identical prices can have very different footprints, depending on their electricity contracts, their process efficiency and where they manufacture. Primary data separates them. A spend-based factor cannot, because the price paid is the only input that varies.

ISO 14064-1:2018 is where the documentation burden shows up. It requires an organisation to identify and quantify its significant indirect emissions, explain the criteria used to decide what counts as significant, document the quantification methodology and the reason it was selected, and address uncertainty. A mix of primary and modelled figures is compliant. A mix nobody documented is not.

The constraint on primary data is availability. It needs the other organisation to have measured, to be willing to share, and to share on a boundary you can use. Early in a programme, most of a supplier or portfolio list has not done that yet, which is why coverage and accuracy pull in opposite directions at the start.

How does the spend-based method work?

The spend-based method multiplies the amount spent in a category by an emissions factor expressed per unit of currency. Those factors come from environmentally extended input output models, which trace the emissions embedded in each sector's output across an economy. CEDA and the US EPA's supply chain emission factors are the sets buyers name most often, and both publish the sector and currency basis they apply to.

Physical activity factor sets are a different job. DEFRA's UK conversion factors, Ecoinvent and GaBi express emissions per tonne, per kilowatt hour or per unit produced, which makes them the inputs to the average-data and hybrid methods rather than the spend-based one. Mixing the two families without saying which is which is a common source of double counting in a Category 1 calculation.

Spend-based coverage is its advantage. Procurement data already exists, so a full Category 1 estimate can be produced across an entire vendor list without asking anyone for anything. It is the honest way to get a first complete picture, and the GHG Protocol treats it as a valid method rather than a shortcut.

Two properties limit what it can then be used for. It reflects a sector average, not the organisation you actually bought from. And its only sensitive input is money, which means the picture changes when prices change and stays still when performance does.

Is this just a trade-off between accuracy and coverage?

Accuracy against coverage is the usual framing and it is incomplete. The sharper question is which decision the number has to support, because the two methods fail at different jobs.

Spend-based figures are adequate for shape. They show which categories and which parts of a vendor list dominate a footprint, which is what a team needs to decide where to spend its attention. A number that is directionally right across all spend beats a precise number covering a tenth of it.

Primary figures are what a reduction claim rests on. Tracking a supplier's improvement, rewarding a lower-carbon option in a sourcing decision, setting a target on a category or evidencing progress against a science-based target all require a figure that moves when the underlying organisation changes. Spend-based figures do not move for that reason.

So the methods are not competing estimates of the same thing. Spend-based data prioritises, primary data proves. Most disagreements about which to use are actually disagreements about what the number is for.

Can you use primary and spend-based data together?

Yes, and the GHG Protocol has a name for it. The hybrid method combines supplier-reported figures where they exist with sector averages or physical activity data everywhere else, inside a single Category 1 total. Nothing in the standard requires one method per category.

The practical sequence is to model everything first, then replace. Start with a spend-based estimate across the full list so the total is complete, rank by modelled contribution, and pursue reported data on the organisations at the top. Each replacement improves the total and narrows the uncertainty around it, without ever leaving a gap in coverage.

What this requires is a record of which figure came from where. An inventory carrying both methods needs the method, the source and the period recorded line by line, both because ISO 14064-1:2018 asks for the quantification approach to be documented and because next year's comparison is meaningless without it.

The shift also changes the work. Modelling is an internal exercise with procurement data. Replacing it means dealing with the organisations themselves, which is a relationship task rather than a calculation one.

What goes wrong with primary data?

Primary data is more accurate about the organisation reporting it and harder to use in aggregate. The problems are consistent enough to plan for.

• Boundaries differ. Two suppliers can report to different consolidation approaches, include or exclude the same subsidiaries, and both be correct under their own standard.
• Scope 2 comes in two versions. Market-based and location-based figures are not interchangeable, and a supplier reporting only the market-based number is reporting the flattering one.
• Periods do not line up. A financial year, a calendar year and a disclosure cycle rarely match yours, so a current-year inventory often carries a prior-year supplier figure.
• Allocation needs assumptions. A corporate total tells you what the organisation emitted, not what your purchase caused. Splitting it by revenue, by spend or by mass introduces uncertainty that the reported figure looked like it had removed.
• Verification status varies. A reported number may be assured to a recognised standard, internally reviewed, or neither, and the three carry very different weight with an auditor.

None of this argues for staying with averages. It argues for validating and normalising reported figures on arrival, and recording what was done, rather than treating a supplier disclosure as finished data.

Where does spend-based data fall short?

Spend-based estimates fail wherever the answer depends on a specific organisation rather than a sector. Three consequences matter.

It cannot separate performers. Every supplier in a category shares a factor, so the one running on renewable electricity and the one running on coal produce the same footprint per pound spent. A sourcing decision made on spend-based data is a decision made on price with extra steps.

It does not respond to improvement. When a supplier cuts its emissions, or when you move volume to a cleaner supplier at the same price, the spend-based figure holds still. Reported emissions and actual progress then drift apart, which is uncomfortable to explain in a disclosure.

It responds to things that are not emissions. Price rises, inflation and currency movements all raise a spend-based total with no change in physical activity, and negotiating a discount lowers it. Anyone reviewing a year-on-year movement has to separate that effect out before reading anything into the trend.

The conclusion is not that spend-based factors are unsound. It is that they answer "where are our emissions" and cannot answer "did they fall".

The four GHG Protocol Category 1 methods side by side

How do you move from estimates to data you can act on?

Reducing reliance on spend-based figures is an engagement problem, not a modelling one. The figure improves when the organisation behind it tells you something, so the work is making that easy enough to happen at scale.

Three things move the share of reported data:

• Ask the ranked list, not the whole list. A modelled footprint already tells you which organisations carry the total. Requests aimed at the top of that ranking convert better and matter more than a questionnaire sent to everyone.
• Accept what already exists. Many organisations have a CDP response, an EcoVadis scorecard or a published report sitting in a file. Taking those as inputs collects reported data without anyone completing a form.
• Reuse rather than re-ask. The organisation being asked is usually being asked by several customers and investors at once. A figure it can publish once and point everybody at gets shared far more readily than one it has to retype per requester.

DitchCarbon provides verified emissions data for over 2 million organisations, so procurement, sustainability and finance teams can measure and act on supply chain and portfolio emissions from one source. Entity resolution against DUNS, LEI and ISIN identifiers is what connects a vendor line or a holding to the organisation behind it, and every figure carries its source and change history, so a reader can see which method produced it. See how the calculation works on the methodology page, or see what the data covers.

Will your calculation method stand up to an auditor?

That depends on whether the method is documented, consistently applied and checkable by somebody outside your team. Buyers ask this as "how easy is it for a third party to audit it", and the answer is about the calculation, not the total.

An auditable Category 1 figure needs four things on the record: which of the four GHG Protocol methods produced each line, which factor set and vintage was used, where any reported figure came from, and what changed since the last cycle. ISO 14064-1:2018 asks for the quantification methodology and the reason for its selection, so the documentation is a requirement rather than good practice.

Verification of the tool doing the calculating is separate from verification of your inventory, and worth checking when you buy. The DitchCarbon Portal calculator is verified to ISO 14064-3, limited assurance, by UL Solutions, renewed annually, and our data capture is annually assured by UL Solutions. Our emission factor methodology was independently assessed by Globus Thenken in August 2025, covering the industry emission factor computational methodology for spend-based Scope 3.1 and 3.2, and assessed as compliant with the GHG Protocol Scope 3 Standard and ISO 14064-1:2018. Both reports are downloadable from the trust centre. We are the only specialist Scope 3 tool with third-party assurance of its calculation methodology, and our benchmark, Who's really verified? A reality check on carbon software assurance, lists every vendor we checked, verified or not, with the verifier for each.

That combination is what audit-ready means in practice: a figure an auditor can follow back to its method and its source, produced by a calculator that has itself been verified to a standard.