IT scope 3 emissions: what to measure and how

The GHG Protocol distinguishes three boundaries. Scope 1 covers direct emissions from sources owned or controlled by the organisation. Scope 2 covers emissions from purchased energy, mainly electricity. Scope 3 covers other indirect value-chain emissions, upstream and downstream.

The GHG Protocol Corporate Value Chain (Scope 3) Standard splits that boundary into 15 categories. Not all of them concern IT: the point is to identify those that genuinely do, and avoid paying for pointless data collection.

Purchased goods and services (category 1) and capital goods (category 2) usually carry most of it: laptops, monitors, phones, servers, network equipment, and also software and cloud services bought from third parties.

Then come upstream transportation and distribution (category 4), waste generated in operations (category 5), and end-of-life treatment of sold products (category 12) for organisations that distribute hardware. Remote work falls under employee homeworking and should be treated explicitly in your boundary.

A useful rule: the electricity your devices consume is scope 2, while manufacturing those same devices is scope 3. That is why a measurement limited to electricity structurally underestimates the footprint of an IT estate.

Two approaches coexist. The spend-based approach applies an emission factor to a purchased amount: fast and cheap, but barely sensitive to real decisions — extending a laptop's lifetime hardly changes the result.

The physical inventory approach starts from actual equipment, its count, type and age. It requires a clean inventory, but it is the only one that makes the levers visible: lifetime, reuse, sizing, dormant hardware.

Whichever you choose, document the source and vintage of the emission factors used, along with exclusions. Without that, no year-over-year comparison holds up.

Double-counting a device that appears both in the inventory and in the year's purchases. Forgetting dormant hardware, unassigned yet paid for. Treating cloud as out of boundary because the provider publishes no usable data — missing provider data does not remove the emission, it must be flagged as missing.

Finally, changing method or factors between two exercises without saying so: the variation you observe is then a change of method, not a change of footprint.