Carbon reporting in manufacturing: what a factory has to be able to prove

The boundary decision determines the number

Before any calculation, someone chooses what counts: which legal entities, which leased premises, which joint ventures, whether a contract manufacturer running your tooling is inside or outside. A plant that buys a component finished elsewhere reports differently from one that machines it in-house, even though the product is identical. Those choices are legitimate but they have to be written down, applied consistently and disclosed, because the year-on-year comparison is what readers actually use. Changing the boundary quietly is the fastest way to lose credibility with an assurance provider, and restating prior periods after an acquisition is normal practice rather than an admission of error.

Three layers of data with three different reliabilities

Fuel burned on site comes from invoices and meter readings and is the firmest part of the picture. Purchased electricity and heat are nearly as solid, though the choice between a grid average and a contractual instrument can move the figure materially. Everything upstream and downstream — bought-in materials, inbound and outbound freight, the use and disposal of what you sold — rests on supplier data where it exists and on generic factors where it does not. For most manufacturers that third layer dominates the total and is the weakest evidenced. Say which parts are measured and which are estimated rather than presenting one confident total.

Building an evidence chain a checker can follow

Assurance work traces backwards: from the published figure to the workbook, from the workbook to the emission factors and their vintage, from the activity data to the invoice, meter log or weighbridge ticket. Gaps appear in predictable places — a site that estimated gas use for a period when the meter failed, a factor applied from a source nobody can now identify, a spreadsheet edited after sign-off. Lock the calculation file, record who approved each input, and keep the source documents with the calculation rather than in whichever inbox they arrived in. Reconstructing this two years later costs several times more.

Customer requests arrive earlier than statutory duties

Large buyers, particularly in automotive, electronics, retail and construction supply chains, ask for plant or product footprints during qualification and re-tendering, sometimes with a methodology of their own and a portal to fill in. These requests are contractual, not regulatory, and they rarely align with whatever a disclosure regime expects. A supplier answering several buyers finds itself maintaining parallel numbers on incompatible boundaries. Deciding once how you calculate, then mapping that to each request, is more sustainable than answering each questionnaire from scratch and hoping nobody compares them. Ask early which method a buyer follows, because retrofitting a boundary to match theirs is painful.

Where the reporting duty is written, and who checks it

Disclosure obligations differ sharply by country and by company profile, are frequently amended, and are enforced by financial regulators, environment agencies or listing authorities depending on the regime. Trade measures that price embedded emissions add another layer for exporters. This page describes the shape of the exercise and is not advice on whether any of it applies to you; that determination sits with the relevant authority and with advisers who know both the jurisdiction and your corporate structure. Environment and energy agencies publish the methodological references most schemes build on.

Frequently asked questions

Our customer wants a footprint per part. How is that different from a site figure?

A site figure divides everything the plant emitted across a period. A part figure allocates a share of that to one product, which forces choices about how to split shared services, scrap, rework and idle time, plus upstream data for the materials in that part. Two defensible methods can give different answers for the same component. Ask the customer which allocation approach they expect, document yours, and keep the site total as the anchor the part figures reconcile to.

Can we use industry average factors instead of supplier data?

Generic factors are how almost every value-chain estimate starts, and they are accepted in many contexts provided the source and vintage are disclosed. The limitation is that they cannot show improvement: switching to a lower-emitting supplier changes nothing in a calculation driven by an average. Where a material dominates your total, primary data from that supplier is usually the first place effort pays back, both for the number and for the conversations it starts with procurement.

Who signs off the figure internally?

In practice it belongs with finance more often than with engineering, because the discipline needed is accounting discipline: consistent boundaries, retained source documents, a controlled calculation and a named approver. Sustainability teams hold the method and the domain knowledge; controllers hold the audit habits. Whoever signs should be able to explain any single input without opening a spreadsheet, which is a reasonable test of whether the process is documented well enough to survive that person leaving.