ISO 14083 sorts the data behind every logistics emissions figure into three types. Primary data is measured, such as the fuel a truck actually burned. Modelled data is calculated from the specifics of the operation, such as the vessel, its route and its load. Default data is an industry average, used when nothing better exists. The standard prefers them in that order, and your report has to say which type you used, where, and why. So every number carries two questions: how big is it, and what is it made of?

The three data types in plain terms

Primary data is the "quantified value of a process or an activity from a direct measurement or a calculation based on direct measurements". Think fuel receipts, energy bills or meter readings. The GLEC Framework calls it the type of data transport buyers should aim to collect from carriers for Scope 3 accounting.

Modelled data comes from a model that takes into account primary data or the parameters that drive emissions in a transport or hub operation. For a ship, that means the vessel's engines, its actual speed and its distance sailed. For a truck, it means vehicle type, load, empty runs and road gradient.

Default data represents average industry operating practice, for example a GLEC default intensity for a mode or vehicle class. GLEC calls it the last resort. It still has a job: defaults give a general indication of emissions, show hotspots and tell you where to collect better data first.

Modelled and default data together are known as secondary data.

What ISO 14083 asks your report to show

The rules are short and strict. According to the CLECAT guide to ISO 14083:

• Use primary data first. Secondary data is allowed only where primary data is not available, and its use must be justified and documented.
• Prefer modelled data over default values.
• Show which data types were used for what, and why.
• Never pick a data type because it gives a lower emissions result.

GLEC Framework v3.2 adds the reporting detail. A report should show the share of primary and secondary data, and split the secondary share into modelled and default. Where default intensities are used, the report must name their source and justify them. Where modelled data is used, it must state the type of model and the parameters applied.

One shipment, three data types

Take a typical door-to-door container move. The trucking company shares fuel records for the pre-carriage, so that leg is primary. The ocean leg is identified by vessel and modelled from its engines and tracked route. The final delivery leg is known only by origin and destination, so it runs on a default factor.

One shipment, three data types. CLECAT expects exactly this. Because many transport service providers do not control the entire transport chain, "many calculations will be based on a mixture of different data types". A single label for the whole shipment hides that mix. The useful view is leg by leg, and ideally parameter by parameter.

Why the label alone is not the quality

Here is the twist. CLECAT notes that "the mere indication of the extent to which primary or secondary data has been used does not make it possible to assess the quality of the data."

Primary data covers a wide range. GLEC says it runs from precise fuel receipts to aggregated values that reflect a year's worth of vehicle movements. So a carrier's annual fleet average can count as primary, and still tell you little about the vessel your container sailed on. The CLECAT authors add that modelling built on high-quality primary inputs can reach comparable quality, provided all relevant parameters are covered.

The better questions are specific. Primary data from what measurement, over which period? Modelled from which vessel, aircraft or truck? Default from which source?

What to ask for

Whether you buy freight or sell it, the same checklist applies:

• The data type for each leg, rather than one label per shipment.
• The source of every default factor.
• For modelled figures, the model and its inputs: vessel, flight number or truck type, and actual distance.
• A plan to move the largest default-based legs up to modelled or primary data.

Shippers can put this list straight into requests to logistics providers. Logistics providers should expect to see it in their customers' tenders and audits.

How Lune handles data types

Lune follows the order the standards set. Where primary fuel data is available, Lune calculates emissions from it directly. Otherwise it models fuel from the specific vessel (around 16,000 in its database, with AIS-tracked distance and speed), the aircraft behind the flight number, or the truck type, load and route. Industry averages are the fallback when nothing better exists.

For every leg, Lune shows a step-by-step explanation, the emission factor and its source, and a data quality score for the emission factor, route and load. The methodology is certified by Smart Freight Centre as GLEC Framework and ISO 14083 compliant.

If your auditor asked tomorrow which of your numbers are measured, which are modelled and which are averages, could you answer leg by leg?

See how Lune calculates logistics emissions

What is primary data under ISO 14083?

Primary data is the quantified value of a process or activity from a direct measurement, or a calculation based on direct measurements, such as fuel receipts or energy bills. ISO 14083 prefers it over all other data types (GLEC Framework v3.2).

What is the difference between modelled and default data?

Both are secondary data. Modelled data is calculated from the specifics of the operation, such as vessel, route, load and vehicle type. Default data is an industry average for a mode or vehicle class. ISO 14083 prefers modelled data over default values (CLECAT guide to ISO 14083).

Do I need primary data to comply with ISO 14083?

No. Secondary data is allowed where primary data is not available, as long as its use is justified and documented. A data type must never be chosen because it gives a lower result.

What should a logistics emissions report say about data quality?

GLEC Framework v3.2 says a report should show the share of primary and secondary data, split secondary data into modelled and default, name the source of any default values and state the model and parameters behind any modelled data.