A carbon footprint comes down to a multiplication. On one side, activity data: 42,000 kWh of gas, 180 metric tons of steel purchased, 95,000 euros of IT services. On the other, an emission factor that converts that data into kilos of CO2 equivalent.
Meeting discussions almost always focus on the first term, the one you have to dig out of invoices. The second is accepted without scrutiny, even though it shapes the result just as much. An order of magnitude can flip depending on the factor chosen, without a single line of activity data moving.
What an emission factor is
It is a conversion coefficient, expressed in kg CO2e per unit: per kWh, per kilometer, per metric ton of material, per hotel night, sometimes per euro spent. It compresses an entire production chain into one number: extraction, processing, transport, losses.
Behind every coefficient sit assumptions. The factor for a kilo of steel depends on the route (blast furnace or electric arc), the country of production, the share of recycled steel, the local electricity mix. Two factors carrying the same label can differ by a factor of three.
The databases we use
Base Empreinte, published by ADEME, is the French public reference. It is free, documented, tuned to the national context, and its factors are the ones expected in a BEGES. It is the backbone of any assignment run under the Bilan Carbone method.
Ecoinvent is an international life cycle assessment database, commercial, going down to unit process level. It becomes necessary as soon as you move from a corporate footprint to a product footprint, an LCA or an environmental declaration.
Then come sector references, such as those published by ASTEE for water and wastewater services, and above all primary data supplied by vendors. One verified primary data point beats any average, and it is the only route that later lets you measure a genuine improvement at a supplier.
Physical or spend-based: the choice that changes everything
A physical factor applies to a measured quantity: kilos, kilowatt hours, ton-miles. A spend-based factor applies to an amount of money, in kg CO2e per euro, by purchasing family.
The spend-based ratio has one real merit: it lets you map thousands of accounting lines in a few days, without contacting every supplier. On a first assessment, it makes scope 3 reachable at all.
Its limits are just as clear. It cannot tell apart two suppliers in the same family, one of whom has decarbonized production and one who has not. It moves with prices: a tariff increase raises your calculated emissions while nothing has changed physically. And it rewards no real effort, which discourages the buyers who are actually trying.
The rule we apply: spend-based to map and rank in the first exercise, physical data and supplier data on the major items from the second one onwards.
Uncertainty is not a flaw, it is information
Every factor carries an uncertainty, often documented in the database itself. It is low on fuels, moderate on common materials, and high on spend-based ratios, where it routinely exceeds half the value.
That rules out a very widespread habit: reporting a result to the last kilo. Writing 12,347 tCO2e implies a precision the method cannot deliver. Writing about 12,300 tCO2e, with close to two thirds on purchasing, says the same thing and holds up in front of an auditor.
Uncertainty also helps prioritize collection. An item that is both heavy and uncertain deserves extra data effort. A marginal item does not, even if its factor is shaky.
How we choose a factor
Three questions drive the selection. Does the factor match the real geography of the flow, rather than the country of the head office? Does it match the technology in use, rather than an industry average? Is its publication date compatible with the year under study?
Every choice is logged in the processing workbook, with its source and date. That traceability is not bureaucracy: without it, redoing the assessment next year becomes impossible, as does answering an auditor's question or explaining why one item doubled.
What happens when databases are updated
Databases move. The French electricity mix changes from year to year, material routes are reassessed, some factors disappear in favor of finer entries.
When a major factor changes, the rise or fall you see in your footprint does not come from your actions. It comes from the database. The only way to keep a readable trajectory is to recalculate the base year with the new factors, then compare like with like. It is tedious, and it is what separates credible monitoring from a decorative dashboard.
The same mechanics apply to a corporate footprint and to a product footprint. We set out how boundaries are split in our article on scopes 1, 2 and 3, and the real cost of an assignment in the one on the price of a carbon footprint assessment.
Frequently asked questions
ADEME's Base Empreinte is public and free. It covers most of what a corporate assessment needs in France and documents its sources and uncertainties. For a detailed product footprint or an environmental declaration, you have to move to an LCA database such as Ecoinvent, which is commercial.
Yes for a first mapping exercise, no for steering a trajectory. Spend-based ratios cannot separate two suppliers and move with prices, which makes measuring real progress impossible. From the second assessment onwards, the major items need to move to physical or supplier data.
Usually because an emission factor was updated, typically the electricity mix. The variation comes from the database, not from your activity. To keep a readable trajectory, recalculate the base year with the new factors.
A carbon footprint gives orders of magnitude and a ranking of items, not accounting to the nearest ton. That is enough to decide where to act, which is the point of the exercise. A result shown with five significant digits mostly signals that uncertainty was never handled.