Why the source matters as much as the number
Most Scope 3 inventories run on secondary data: a mass of purchased goods, fuel, or waste multiplied by an emission factor pulled from a published database. The factor itself often gets treated as a neutral constant, something you look up once and move on from. It isn't. Three of the most commonly used sources — the UK's DEFRA conversion factors, the US EPA's emission factor publications, and the Swiss-based ecoinvent lifecycle database — are built on different assumptions, different geographic scopes, and different update schedules. Swapping between them mid-inventory, or using them inconsistently across categories, can shift your totals in ways that have nothing to do with actual emissions performance.
This matters more as disclosure regimes mature. Under frameworks like CSRD/ESRS E1 or California's SB 253, inventories increasingly face external assurance. An assurance provider will ask not just "what number did you report" but "where did this factor come from, and why." If the methodology notes say "various sources," that's a finding waiting to happen.
Regional specificity
DEFRA's conversion factors are built around UK activity data — UK grid electricity, UK fuel mixes, UK waste treatment pathways, UK commuting and travel modes. They're a strong fit for a company reporting primarily on UK operations, but applying a DEFRA road-freight factor to a supply chain segment running through Southeast Asia introduces a geographic mismatch that no amount of careful spend-data collection will fix.
EPA's factors are similarly anchored to the US context — electricity grid factors by region (eGRID), US fuel combustion factors, US-specific waste and transport data. They work well for US-centric inventories but carry the same mismatch risk if applied elsewhere.
ecoinvent is structured differently. It's a lifecycle inventory database covering a wide range of materials, processes, and products, with many entries offering regional variants (global average, EU, specific countries) rather than being anchored to a single national context. That breadth is useful for multinational inventories and for Category 1 (Purchased Goods & Services) and Category 11 (Use of Sold Products) work where the product or material in question doesn't map cleanly to one country's dataset. It also means more decisions get pushed onto the user: which regional variant, which production route, which system boundary assumptions.
Update cadence and what it implies
DEFRA publishes updated conversion factors on a regular annual cycle, which means inventories built on them need a documented policy for which year's factors apply to which reporting period — using the factor set current at the time of activity, not the time of calculation, is the more defensible convention, but it has to be stated.
EPA updates its emission factor resources periodically as underlying government data (fuel mixes, grid composition) is refreshed, though not necessarily on the same fixed annual rhythm as DEFRA.
ecoinvent releases are versioned rather than annual, with periodic major updates that can include revised background data across many processes at once. A big version jump can shift calculated emissions for reasons that have nothing to do with a company's actual supply chain changes, purely because the underlying process models were revised. That's a legitimate methodology event, but if it isn't flagged in your year-over-year narrative, it looks like an unexplained swing in performance.
Cost and access
DEFRA's conversion factors are published freely by the UK government. EPA's emission factor resources are similarly free and publicly accessible. ecoinvent is a licensed, paid database, typically accessed through LCA software platforms, which is one reason smaller companies and early-stage inventories lean on DEFRA or EPA factors even outside their home geographies — not because the fit is better, but because the access cost is lower. That's a reasonable starting point, but it should be an explicit, acknowledged tradeoff rather than a silent default.
Where this sits in the data quality hierarchy
None of this changes the underlying GHG Protocol data quality hierarchy: spend-based estimates sit at the weak end, average-data and hybrid methods sit in the middle, and supplier-specific primary data sits at the strong end. Choosing between DEFRA, EPA, and ecoinvent is mostly a decision within the average-data tier — it affects which average you're using, not whether you've moved up the hierarchy. Companies sometimes treat a database switch as a methodology improvement when it's actually a lateral move. Genuine improvement still comes from engaging suppliers for primary activity and emissions data in the categories that matter most to your footprint.
The practical takeaway
Treat emission factor source selection as a methodology decision, not a lookup convenience:
- Document which source is used for which Scope 3 category, and why that source is the appropriate regional or sectoral fit.
- Record the factor set version or publication year alongside the reporting period it was applied to.
- Flag any database version changes in your year-over-year narrative so reviewers and assurance providers can distinguish real performance change from methodology change.
- Avoid mixing sources within the same category across reporting years unless the switch is documented and, ideally, the prior year is recalculated or footnoted for comparability.
The emission factor is doing real analytical work in your inventory. It deserves the same documentation discipline as your activity data.