Primary data
Data measured at its source — your meter, your supplier's actual production — as opposed to an industry average.
Primary data comes from the process being described. Secondary data is a stand-in: an average, a proxy, a database entry for something similar.
The difference is not academic. A footprint built on averages cannot detect that you changed supplier, switched fuel, or improved a process, because none of those change the average. It is a number that cannot be managed.
Most inventories are a mixture, and there is nothing wrong with that. What matters is that the mixture is visible — that a reader can see which parts of a figure are measured and which are assumed, rather than being handed one number with a uniform appearance of precision.
Related