The Scope 3 number your customer needs is sitting in your OT data

From the financial year starting 1 January 2026, Straits Times Index constituents on SGX must disclose Scope 3 emissions, on top of the Scope 1 and 2 reporting already mandatory for all listed companies from FY2025. For most of them, a material share of that Scope 3 number is Category 1: the emissions embedded in what they buy, raw materials, contract manufacturing, freight. That number has to come from somewhere, and the GHG Protocol is explicit about where it comes from when a supplier does not provide it: a spend-based estimate, built from an industry-average emission factor applied to the invoice value. That is the least accurate tier the Protocol defines, and it behaves oddly in ways a verifier will ask about, negotiate a discount with a vendor and the reported emissions fall, with nothing about the physical activity having changed.

• FY2026

  the year STI-listed companies must start disclosing Scope 3 emissions

• 4

  calculation methods in the GHG Protocol's accuracy hierarchy — spend-based is the least accurate

The Protocol's own technical guidance ranks four calculation methods in increasing order of accuracy, and where a supplier sits on that table is not the customer's decision to make.

• Spend-based Least accurate

  Invoice value × an industry-average emission factor — the sector's average intensity per dollar, not this supplier.

• Average-data Better

  Industry-average activity data applied to the supplier's known energy, fuel and material use for the order.

• Hybrid Better still

  Supplier-specific data where it exists, blended with average data to fill the gaps — a defensible, auditable number.

• Supplier-specific Most accurate

  The supplier's own verified emissions, allocated to the specific product — the most granular figure the Protocol recognises.

Moving up that table does not require a new disclosure programme. It requires the energy draw, fuel consumption and throughput a plant's automation and metering layer is already recording, the same figures that would go into that supplier's own Scope 1 and 2 inventory, to be logged at a granularity that can be allocated to a specific product run rather than only summed at the site. That is an instrumentation and retention decision, made at the point the metering is designed, not a reporting-season task. A gate meter that only totals monthly site consumption can support a spend-based estimate. It cannot support an activity-based one, because there is nothing in it to allocate to one customer's order rather than another's.

1. 01

   Sub-metering

   At the process or line level, so consumption can be allocated to a product rather than only totalled for the site.

2. 02

   Fuel logs at the point of use

   Generators, forklifts, site vehicles — captured directly, not reconstructed later from purchase invoices.

3. 03

   Retention that outlives the reporting cycle

   A verifier asks for the source record, not the summary figure.

4. 04

   A named owner for the meter data

   An unowned sensor drifts out of calibration and nobody notices until an auditor asks for evidence it did not.

Four things have to be designed in rather than audited in afterwards: sub-metering at the process or line level so consumption can be allocated to a product rather than only totalled for the site, fuel logs captured at the point of use rather than reconstructed later from purchase invoices, retention that outlives the reporting cycle it was collected for, and a named owner for the meter data, because an unowned sensor drifts out of calibration and nobody notices until an auditor asks for evidence it did not.

A verifier does not ask whether a number is plausible. As a GHG Lead Verifier, the question I ask first is where the figure came from, the meter, the log, the date, and who can produce it again if asked twice. A spend-based estimate cannot answer that question by construction, it was never built to. For a supplier, the choice is not whether to do carbon accounting. Regulation is deciding that upstream of them, on their customer's timeline. The choice is whether the number their customer reports about them is one they controlled, or one an industry average produced on their behalf.