Product carbon footprints: primary versus secondary data explained
Primary data comes from a supplier's own measurements and secondary data from industry averages. Most apparel footprints mix both, and the uncertainty should be shown.
KEY TAKEAWAYS Summary by the editors
- Primary data is measured, supplier-specific data such as the energy a particular factory used for a process, while secondary data is an industry average from a database.
- The Fashion Industry Charter guidance says that ideally apparel and footwear companies would have supplier-specific data for all tiers, but that this will take years, so most companies rely on secondary data for parts of their inventory, especially tiers 2 to 4.
- The same guidance cites the GHG Protocol in warning that supplier data may be less accurate than industry-average data, depending on granularity and allocation methods.
- Secondary factors can disagree widely: one analysis cites databases giving 28 to 142 kg CO2e for 1 kg of raw merino wool.
- Spend-based estimates can help initial screening but are discouraged for tracking year-on-year progress against targets.
What is the difference between primary and secondary carbon data?
Primary data is measured and specific to a supplier, process or product, for example the electricity a particular dyehouse used for a batch. Secondary data is an average taken from a database, such as ecoinvent or the Higg Materials Sustainability Index. Most apparel product footprints combine the two, and the share of each should be reported.
This guide explains how to choose between them, what each costs and how to present the result honestly.
Why do most apparel footprints rely on secondary data?
Garments pass through several tiers: fibre, yarn, fabric, dyeing and finishing, cut and sew, and trims. A brand usually has a direct relationship only with the final assembly site. The Fashion Industry Charter guidance states that ideally companies would have supplier-specific data for all tiers, but that reaching this will take years and major changes in supply chain visibility. It adds that most companies rely on industry averages for parts of their inventories, especially tiers 2 to 4.
Measurement itself is a barrier. One analysis describes primary data as the ideal but often unavailable, because factories frequently lack the submetering needed to collect energy use for a specific production step.
Is primary data always more accurate?
No. The Fashion Industry Charter guidance, citing the GHG Protocol, warns that supplier data may actually be less accurate than industry-average data depending on granularity and allocation methods. A factory-wide energy bill divided by output is primary data, but it is a coarse allocation to a single product.
Primary data does have a clear advantage for tracking change. It can show actual reductions, whereas secondary data does not change from year to year in a way that reflects a brand's own actions, and annual updates to average datasets can shift results for reasons unrelated to the brand.
| Aspect | Primary data | Secondary data |
|---|---|---|
| Source | Measured at supplier, process or product level | Industry averages from databases |
| Strength | Can show real supplier-level change | Available now for most materials and processes |
| Weakness | Costly to collect; allocation can be coarse | Generic; may not reflect the actual mill or energy mix |
| Use in targets | Better for tracking progress | Weaker for tracking; shifts with database updates |
| Typical tiers | Final assembly, sometimes fabric | Fibre, yarn and upstream tiers |
| Uncertainty | Depends on method and meter quality | Can be wide where factors disagree |
How big can the differences between datasets be?
Considerable. One analysis from a carbon accounting software supplier reports that for 1 kg of raw merino wool, the databases it cites range from 28 to 142 kg CO2e. It adds that some factors require inputs such as yarn thickness that brands rarely track. This is a single illustration from a supplier, but it shows why the choice of database and the declared uncertainty matter as much as the headline number.
The same source proposes prioritising data collection where it reduces the most uncertainty, for example one material accounting for 55% of total uncertainty in its worked example. That logic is general: collect primary data first for the materials and processes that dominate both emissions and uncertainty.
How should a brand combine the two?
- Start with secondary data to screen the portfolio and identify the materials and processes that matter most.
- Rank hotspots by emissions and by uncertainty, not emissions alone.
- Request primary data from the suppliers behind the largest hotspots, with a defined method and boundary.
- Keep a record of data source and quality for every input, so mixed footprints can be explained.
- Update the footprint when better data arrives and report the change in method separately from the change in performance.
The Higg Product Module, launched by the Sustainable Apparel Coalition and Higg Co, illustrates the structure. Its first edition covered impacts from resource extraction to finished product assembly, drawing on materials data from tier 2 manufacturers and Higg MSI data to calculate material impacts.
A practical question is who should collect primary data and at what cost. Asking every supplier for detailed process data on every style is unrealistic, and many suppliers receive similar requests from several customers. A focused request, limited to the few processes that dominate a product's footprint and based on a method agreed in advance, is more likely to produce usable answers. Shared formats and common questionnaires reduce the burden on suppliers and improve comparability.
Data quality also needs managing. A primary figure should be accompanied by the measurement period, the boundary, the method for allocating energy to products and, where possible, evidence such as meter records or utility bills. Without this, a primary figure cannot be distinguished from an estimate. It is sensible to rate each input for quality, for example by measured, calculated or default, and to improve the weakest inputs among the largest contributors first.
The role of AI in this area is mostly clerical. Models can extract values from supplier forms and utility documents, map materials to database entries and flag outliers. They do not make a secondary factor more accurate, and they cannot replace measurement. If a tool assigns a database factor automatically, the match should be visible and reviewable, because a wrong match to a similar-sounding material can change the result materially.
What are the limits for claims and regulation?
Product footprint figures are estimates with error margins, and any consumer claim should reflect that. The Fashion Industry Charter guidance discourages spend-based estimates for year-on-year tracking, though it notes they can help with initial screening. Looking ahead, the Ecodesign for Sustainable Products Regulation, in force since 18 July 2024, establishes digital product passports that can hold materials, origins and lifecycle environmental impacts. One compliance update expects the textiles-specific act in 2027 and advises defining product-level data requirements and engaging suppliers early. Records that distinguish primary from secondary data will be easier to adapt to whatever format is finally required.
Frequently asked questions
What is primary data in a product carbon footprint?
It is measured data specific to a supplier, process or product, such as the energy a given factory used for a production step. It contrasts with secondary data, which comes from averages in a database.
Is secondary data acceptable for a product carbon footprint?
Yes, for most apparel inventories it is unavoidable, especially in upstream tiers. It should be labelled as such, with the database named and the uncertainty acknowledged.
Why can two footprints for the same garment differ so much?
Different databases, assumptions and system boundaries give different results. One analysis cites databases ranging from 28 to 142 kg CO2e for 1 kg of raw merino wool.
Should I use spend-based data for carbon targets?
It is useful for early screening, but the Fashion Industry Charter guidance discourages it for tracking year-on-year progress against targets, because it does not reflect real changes in supplier performance.
One edition every weekday morning. Read in five minutes. Free for industry professionals.
SOURCES
- UNFCCC: Fashion Industry Charter guidance for measuring greenhouse gas emissions
- Carbonfact: Uncertainty metric, how to scale your LCA efforts
- Just Style: New Higg PM assesses environmental impacts of finished products
- McKinsey: Fashion on Climate
- Renoon: Latest compliance update on Digital Product Passport and ESPR timelines
