Ask two analysts to calculate the carbon footprint of the same steel coil. Despite the same product, same ISO 14067 norm, same auditor sign-off, the two numbers can land more than 35% apart. And both can be defended as correct.
The spread is structural. Life Cycle Assessment (LCA) was built with that much room for the practitioner to choose. For procurement leaders, cost engineers, and CFOs whose decisions increasingly depend on a single carbon figure, that property has become a commercial problem. Carbon is no longer only a reporting line. Through the EU Emissions Trading System and the Carbon Border Adjustment Mechanism (CBAM), emission costs are already moving through European supply chains. A 35% variance in the number on the report is also a 35% variance in the cost being passed along.
In a recent episode of the Beyond Cost podcast, Prof. Dr. Jana Backes, Junior Professor for Safety, Security and Sustainability Evaluations in Foresight Research at SAF RWTH Aachen University, explained why the same product can carry such different footprints, and how the industry should respond.
Life Cycle Assessment (LCA), the methodology behind every product carbon footprint, sits inside the ISO 14040 family of norms. The standard defines how an assessment should be structured but leaves the practitioner room to choose data sources, set system boundaries, and select an impact assessment methodology. Those choices, all defensible under the norm, are what produces the spread.
As Dr. Backes put it on the podcast:
Three factors drive the spread. First, data: few manufacturers hold primary data for every step of their value chain, so gaps are filled with commercial secondary databases, and two analysts can select different entries for the same process and both be compliant. Second, system boundaries: whether a footprint runs cradle-to-gate or extends through end-of-life determines which emissions appear in the model at all, and a recycling credit can offset a meaningful share of upstream emissions (further varying in amount of credits, depending on the assessment approach). Third, impact assessment methodology: different frameworks count different greenhouse gases and weight them differently, so the final figure can shift without a single physical fact changing. Of these, Dr. Backes is clear that data selection carries the most weight.
What you are left with is a figure that is defensible under the norm, but not directly comparable to the figure a supplier or competitor would produce for the same product under different assumptions.
For years, carbon was a reporting obligation handled outside procurement. But that is changing now. With CBAM in force and ETS prices passing through European supply chains, every ton of CO₂ in a bill of materials carries a cost that suppliers are starting to price in. Geopolitical developments and broader economic shifts compound this further, adding pressure on emission-related costs that go beyond regulatory timelines alone.
Dr. Jana Backes' recommendation for a head of purchasing is direct:
The same logic applies to forecasting. Cost projections for next year and five years out should include emission costs across new technologies, new designs, and end-of-life options. For companies that have never measured the carbon footprint of their purchased goods, the starting point looks much like any cost engineering exercise. The data tends to already exist inside the company, in the form of, e.g., energy contracts and process documentation. The hard part is mapping the value chain past the first-tier supplier, which is where transparency breaks down. A practical baseline begins with, for example, the most expensive purchased goods and works outward from there, tracing the main energy sources, processes, and geographies that drive the footprint.
That supply chain map serves more than the carbon calculation. Resource criticality, the question of how much of a given material can still be extracted and whether circularity options exist, depends on the same transparency. So does any serious assessment of the social dimension: the working conditions and sourcing practices behind materials that travel through mining and processing across multiple geographies before reaching the factory floor.
A significant portion of an LCA practitioner's time goes into matching the items in a bill of materials to the “right entries” in an emission database. Research at RWTH Aachen is working on automating that matching step. What Dr. Backes’ group is doing sits closer to machine learning and language training than to AI in the broader sense. As she described it:
One of the benefits is consistency. When two analysts use the same model on the same bill of materials, the variance introduced by individual data selection drops. Expert review of the output remains essential, but the most repetitive part of the assessment becomes faster and more standardized.
A second application Dr. Jana Backes sees as promising is optimization and forecasting for new product designs. Current LCA practice is largely status quo assessment: what is the footprint of what already exists. A model capable of generating and comparing future design scenarios, showing how a change in material, process, or geography would affect the carbon footprint, cost, and social value chain simultaneously, would shift the tool from measurement to decision support. That work is at an earlier stage, but it is the direction her group is moving toward.
Real supply chains resist clean numerical representation. Every gap in the data forces a judgment call, and those calls accumulate across a long bill of materials. That is where machine learning earns its place: handling the pattern-based choices consistently, and flagging where a case falls outside what the model recognizes.
So far, CO₂e (the carbon footprint) has dominated sustainability reporting because the data is most available and the regulatory pressure is highest. Dr. Jana Backes expects this to change – or, to put it more accurately, that there will be additions.
She draws a useful historical parallel: the ozone layer represented a comparable global sustainability challenge, one that took decades to address at policy and industry level but is now largely on track. CO2 and climate change are following a similar arc, and the indicators that dominate the next phase will not necessarily be the same ones that dominate today. For example, biodiversity could be one of the topics gaining traction in that transition.
Dr. Backes’ wider expectation for the next generation of engineers is to read sustainability through all three pillars, environmental, economic, and social, rather than collapse it into a single number. She also raises a fourth dimension worth integrating: technical quality, understood as long product lifetime and durability, which carries its own sustainability logic and should not be treated as separate from the environmental and economic trade-offs.
The remedy for the variance is not more indicators, but tighter rules. "The key word is standardization," Dr. Backes said, "especially in the industrial and application perspective."
Frameworks like the Environmental Product Declaration (EPD) and the Product Environmental Footprint (PEF) build on ISO 14040 but fix more of the variables, which makes cross-company comparison meaningful for the first time. CBAM extends that logic outward. Producers outside Europe are likely to adopt European rules, because the cost of inconsistency across a global supply chain is higher than the cost of compliance.
For cost engineers, procurement leaders, and CFOs in manufacturing, the practical reading is this. The carbon number on the report is not the final answer. It is the output of a methodology that needs to be understood, questioned, and increasingly priced into every cost decision.
This blog article draws on a longer conversation between the host Jakob Etzel, VP Customer Success at Tset, and Prof. Dr. Jana Backes on the Beyond Cost podcast.
Listen to the full episode to know more, including how Dr. Jana Backes teaches her engineering students to interrogate every published number, the one EU regulation she would simplify tomorrow, and what she believes every CFO in a manufacturing company should do this week to take emission costs seriously.