Scope 3 is not a data problem. It is a relationship problem.

Scope 1 and Scope 2 are laborious. You chase bills, you argue about boundaries, you learn which meter serves which building. It is tedious and entirely within your control.

Scope 3 is different in kind. It is the largest part of most inventories and the only part you cannot measure, because it happens inside other companies. Every technique for producing it is a way of dealing with that fact.

Why spend-based estimates became the default

Most companies report Scope 3 by taking what they spent with each supplier and multiplying it by an industry average emission factor. Take a year of steel purchasing, apply the steel factor, get a number.

It is defensible, it is fast, and it is what almost everyone does. It also has a property that should worry you: it cannot detect improvement. If your supplier decarbonises their smelter, your spend-based Scope 3 does not move. If you switch to a cleaner supplier at the same price, your number is identical.

So the metric you report is not measuring the thing you are trying to change. Fine for a first inventory. Useless as a basis for action, and increasingly hard to defend as frameworks tighten.

The alternative is primary data — asking suppliers for their actual figures. Everyone knows this. Almost nobody does it, and the reason is not technical.

The reason suppliers do not respond

Consider what you are asking. You email a mid-sized fabricator in Johor and request their greenhouse gas emissions attributable to what they sell you. From their side:

• They may not have an inventory at all
• If they do, it is company-wide, not allocated to you
• The person who receives your email does not know what Scope 1 means
• You are one of several customers asking, in different formats, at different times
• Nobody has told them what happens if they ignore it

Then you send them a portal link and ask them to create an account.

That last step is where most supplier engagement programmes quietly end. A supplier who does not have the data will not register for a system to tell you so. And the ones who do have it will not maintain a login for one customer’s annual request.

What we changed, and why

No account. A supplier gets a link, opens a short form, and submits. The link is the whole security model: a unique token, valid ninety days, tied to exactly one buyer and one reporting period. Nothing to remember, nothing to install, nothing to reset.

Let them answer what they can. The form offers two paths — their own emissions figure, or activity data we calculate from. A supplier who has never done an inventory usually cannot give you tonnes, but can tell you how much electricity their plant used. That is worth far more than a refusal, and it is a question they can actually answer.

Ask how they worked it out. Measured, calculated from records, or estimated. This is one dropdown and it changes everything downstream, because a metered figure and a guess arriving in the same field is what makes an inventory misleading. An estimate is perfectly acceptable — it just has to be labelled as one.

Send it from the buyer’s mail server. A supplier is far more likely to answer an email from a company they already invoice than from software they have never heard of. Our platform sends through the customer’s own SMTP, so the sender is a name the recipient recognises.

Chase, but stop. Maximum three reminders, at least three days apart, and only once a deadline is actually in sight. A supplier reminded every week forever stops reading anything from that address — including next year’s request.

Coverage is measured in spend, not headcount

Here is the metric that matters, and it is not the one most programmes report.

“We contacted 40 suppliers and 22 responded” sounds like 55% coverage. It usually is not. If the 22 who responded are your smallest vendors and the four that make up 60% of your purchasing are silent, you have a response rate and no coverage.

Track the share of addressable spend represented by responses. Ten small suppliers answering is worth less than one large one, and the programme should be aimed accordingly. The first year, target the ten suppliers that account for the majority of spend and ignore the tail entirely.

Nothing a supplier types becomes your number

This is the part that makes the whole thing survivable.

A supplier response does not enter the inventory. It lands as a draft, reviewed and approved by someone in your organisation, exactly like a figure typed in-house. The supplier gives you data. A person on your side decides whether it becomes a disclosure. Those stay separate, and the audit trail records which is which.

It matters more than it sounds. You are accepting numbers from a party with an interest in them, produced by methods you cannot inspect, into a document you sign.

What a realistic first year looks like

Not full coverage. Nobody gets that.

A realistic first year: the ten highest-spend suppliers asked properly, three or four responding with something usable, one of them with genuinely good data, and a much clearer picture of which relationships can carry this conversation and which cannot.

The second year is where it compounds, because you are no longer introducing the idea, you are following up. And because the suppliers who responded once have now done the internal work to answer, which is the expensive part for them.

The uncomfortable truth about Scope 3 is that the first year mostly produces better questions rather than better numbers. That is not failure. Spend-based estimates will still fill the gaps, and they will keep doing so for years. But every supplier who moves from estimated to measured is one line of your inventory that can finally register the thing you are trying to do.