How to collect emissions data from suppliers (Scope 3 in practice)

For most manufacturers and trading companies, the biggest block in the whole inventory is not the chimney — it is what you buy. The GHG Protocol calls it Scope 3 category 1 (purchased goods and services); under ISO 14064-1:2018 it lands in category 4 (see how the two standards map). In year one almost everyone estimates it with industry-average emission factors, and that is fine — the standards allow it. But the moment the inventory becomes an annual exercise, faces verification, or a customer starts asking questions, the direction is fixed: replace industry averages with supplier-specific data, one supplier at a time.

The hard part is not the arithmetic. The hard part is getting dozens of suppliers to actually return the form.

Decide which tier of data to ask for

Supplier data comes in three tiers. The higher the tier, the better the number — and the fewer suppliers can produce it:

Tier	What the supplier provides	What you do	Who can produce it
Product carbon footprint (PCF)	kgCO₂e per unit of product, ideally third-party verified	Multiply by your purchase volume	The few who have done an LCA / ISO 14067
Allocated emissions	Your share of their annual emissions, allocated by revenue or output (kgCO₂e)	Check the basis, then book it	Those with an organizational inventory
Activity data	Raw electricity, fuel, and similar figures	You pick the factors and convert	Those with nothing but utility bills

The practical answer is to collect mixed: one form that accommodates both the PCF and the allocated-emissions path, letting each supplier take the lane they can actually drive. Sending a PCF-only questionnaire is a classic mistake — most suppliers cannot answer it, and what you get back is not a downgraded reply but silence.

Whatever the tier, three metadata fields are non-negotiable: the reporting period (which year the number describes), the method (which standard it was calculated under), and whether it was third-party verified. Without these you cannot weight data quality, and you cannot defend the number in verification (see inventory vs verification).

Response rate is the real constraint

Your questionnaire sits below real work in the supplier’s queue — a large supplier fields dozens of these requests a year. Every unit of friction costs you responses:

• Requiring an account or a platform login is the single biggest unit of friction. The person filling it in must register, verify an email, and learn an unfamiliar interface — to enter seven or eight numbers. Cross-border chains make it worse: asking a Chinese supplier to sign up to an overseas SaaS platform frequently kills the request outright.
• A spreadsheet attachment, anyone can fill. If they can open Excel, they can respond — that is what zero-barrier collection looks like.
• Fewer questions. Ten or less; every extra question is an argument for giving up.
• Bilingual headers. When the chain crosses languages, do not expect the supplier to translate for you.
• A deadline and a named contact on your side. A request without a date has no priority.

Close the loop as a process

Collection is not an email — it is a stateful process. Threads scattered across a mailbox stop being reconstructable within a quarter:

• Keep a ledger: one row per supplier, four statuses — draft, sent, received, ingested. At any moment you can see who is still missing.
• Make overdue visible and chasing cheap: record the send date and the due date, flag what has slipped, and keep a reminder template ready (with the deadline and days overdue in it) so a nudge is one click, not a drafting session.
• Review before booking: returned forms do not go straight into the accounts. Check units (kg or t), plausibility of magnitude, and that the reporting period matches your inventory year — then confirm.
• Archive the original file: when the verifier asks where a number came from, the right answer is the supplier’s original file, not an untraceable manual entry.
• Keep the chain clickable: activity rows created from a disclosure should link back to that disclosure and its file — the same evidence-per-number principle from the seven-step inventory guide, extended to supplier data.

Common failure modes

• PCF-only questionnaires, answered by silence (see above: collect mixed).
• No reporting period on the form — 2023 numbers booked into a 2025 inventory.
• Units left open; kg and t are three orders of magnitude apart and nearly impossible to spot once mixed in.
• Double counting: the supplier’s PCF is cradle-to-gate and already includes upstream transport, and you count the same inbound leg again under the transport category. Ask for the PCF boundary explicitly.
• No verification status collected, so every number carries equal weight — until the verifier weighs them for you, line by line.
• Forms returned with restructured tables, or supplier A’s template filled in by supplier B, booked without any check.
• Numbers detached from their original files — one failure, but enough to undo everything else at the verification table.

Supplier data is one part questionnaire design, seven parts process discipline. Push the barrier to zero, keep the statuses in a ledger, keep the originals in the chain — response rates and auditability are not luck. They are designed.