Measured vs Estimated Carbon Data: Why the Difference Is Commercial, Not Academic

Ask most organisations for their carbon footprint and you will get an estimate: annual energy bills multiplied by an average emission factor, allocated across activities by assumption. For a first disclosure, that is a reasonable starting point. As a basis for decisions — or for a border carbon charge — it breaks down quickly.

Three ways estimates fail

• They cannot localise. A site-level estimate tells you the total but not which line, machine or behaviour caused it — so it cannot tell you what to fix.
• They ignore time. Grid carbon intensity varies substantially through the day; an annual average factor misprices every kilowatt-hour, in both directions.
• They cannot be audited. An allocation assumption is an argument, not evidence. When a customer, regulator or border mechanism asks how a number was produced, 'we apportioned the bill' is a weak answer.

Where it becomes commercial

Carbon border mechanisms make data quality a line item. Under CBAM, embodied emissions determine the charge — and where actual data is unavailable, conservative default values apply. Supply-chain reporting has the same shape: manufacturers are increasingly asked by their customers for product-level emissions, and suppliers who can evidence measured figures are simply easier to buy from.

Reduction economics point the same way. Every site CarbonMeter has measured consumed 12–22% of its electricity outside operating hours — waste that is invisible in an annual estimate and obvious in a measured overnight profile. You cannot fix what you cannot see, and estimates, by construction, do not see.

What measurement looks like in practice

Measured carbon accounting means energy readings at asset level, joined to verified emission factors — official DEFRA factors and live grid carbon intensity — with every figure traceable to a reading and a timestamp. It does not require ripping out infrastructure: it connects to the meters and systems a site already has, extended with optional third-party sensors where more granularity is needed.

The principle is the one we build the whole platform on: you cannot manage what you estimate. Measured beats estimated — in accuracy, in auditability, and increasingly in pounds.