Primary data share is the fraction of a product carbon footprint that is built on data from your actual value chain rather than from industry-average databases. It has quietly become the number sustainability teams are judged by, because it is the one number that says how much of your footprint describes your supply chain instead of a statistical stand-in for it.
What counts as primary
A datum is primary when it comes from the specific suppliers, sites, and processes that make your product: metered electricity at a supplier's plant, weighed material inputs, a supplier-calculated PCF for the exact component you buy. Secondary data is everything pulled from databases and literature: sector averages, region averages, proxy factors.
The strict version, used by the PACT and Together for Sustainability methodologies, is stricter than most teams expect. A PCF contribution only counts as primary when both the activity data and the emission factor behind it are primary. A supplier's real production volumes multiplied by a database factor is still secondary. This single rule is why self-assessed primary data shares tend to fall on first review.
Why the number is rising in importance
Two pressures converge on it.
First, disclosure. ESRS E1 under the CSRD asks reporters to describe how scope 3 emissions were calculated, including the extent to which primary data was used. The regulation does not demand a high share. It demands that you state the share, which turns a methodology footnote into a comparable, quotable figure. A buyer reading two suppliers' reports can now see that one footprint is 60 percent supplier-specific and the other is a database exercise.
Second, decision usefulness. A footprint built on averages cannot show progress. If your supplier cuts their process emissions by a third and you model them with a sector factor, your reported footprint does not move. Primary data is what makes reductions visible, which is what makes the footprint worth managing.
The part the definitions skip: a primary datum is a claim
Here is the failure mode we see. A team spends a year collecting supplier PCFs, proudly reports a high primary data share, and then assurance asks three questions:
- Where did this number enter your system, and from whom?
- What methodology and factor vintage sits behind it?
- Can you show me the document?
If the answer trail is an inbox and a spreadsheet tab, the primary data share does not survive contact. The datum was primary; the evidence for it was not managed. Assurers do not grade intentions, they grade the chain from reported figure back to source document.
So treat every supplier-provided figure the way you would treat an invoice in a financial audit:
- Bind it to its source document. The PCF report, the utility bill, the weighbridge ticket. Store the document, not a transcription of it.
- Version it. Suppliers restate. When the March figure is corrected in September, both versions and the switchover date should be visible.
- Record the methodology. ISO 14067, GHG Protocol Product Standard, PACT conformance, factor vintages. A primary number with an unknown method is a future finding.
- Label the gaps. Where you fall back to secondary data, say so in the record itself. A footprint that marks its own estimates reads as controlled. One that hides them reads as lucky.
How to raise your share without boiling the ocean
Chasing primary data from every supplier is a volume game that stalls by summer. The teams that move the number do it narrowly.
Start from contribution, not from convenience. Rank purchased goods by their share of the footprint. In most physical-product companies a small set of materials and components carries most of the emissions. Ten well-chosen suppliers can move the share more than two hundred polite requests.
Ask for something specific. "Please send your carbon data" produces PDFs you cannot use. Ask for a product-level PCF for the exact item you buy, with methodology, boundary, factor vintage, and the primary share of their own calculation. Suppliers who cannot provide it yet can usually provide activity data, which still upgrades your model.
Make the request part of procurement, not sustainability. A data clause in the buying relationship outperforms an annual survey from a team the supplier has never traded with. Renewal conversations are the natural moment.
Meet suppliers where they are. A supplier's first submission will be imperfect. Take it, label its quality, and improve it next cycle. A labeled estimate that improves year over year is worth more than a stalled demand for perfection.
Keep one GWP vintage per reporting period. Mixed factor vintages are one of the quietest ways to make good primary data unusable, because the comparison baseline shifts under the number.
What good looks like
A defensible primary data share statement reads something like: measured share, method for classifying primary versus secondary, coverage by emissions contribution rather than by supplier count, and the evidence location for each major primary input. If you can produce that paragraph and back every clause of it with a document, you are ahead of most of the field, whatever the percentage says.
This is the lens we build CarbonSKU around: every number in a footprint, primary or secondary, bound to its source evidence, versioned, and replayable when someone asks a year later. If your primary data share is about to become a disclosed figure, the time to give each datum a paper trail is while it arrives, not the week assurance starts.