On 26 August, a glacier collapsed on the Nepal-Tibet border, triggering flash floods and mudslides that displaced over 100 million cubic metres of ice, rock and debris, destroying villages, roads and hydropower sites. Nepal’s confirmed death toll now stands at 903, with more than 4,200 still missing, one of the region’s deadliest disasters in recent memory. Scientists studying events like this point to the same mechanism: a warming atmosphere destabilising glaciers and releasing water with catastrophic force.
It is against this backdrop that companies publish sustainability reports on their carbon footprint, assuming that measuring the problem is the first step to solving it. Regulators read these reports, investors price risk against them, and customers base sourcing decisions on them.
Yet for most of what a typical company emits, the numbers are not reliable enough to act on, a fact that should trouble anyone who has staked a net zero commitment, a trade strategy, or a compliance budget on it. Actually, it should trouble all of us.
The part of the footprint nobody can see
Most of what a company emits does not come from its own factories, but from everything upstream and downstream: raw materials, components, freight. These are Scope 3 emissions, typically the majority of a company’s footprint, much of it several tiers deep in the supply chain, beyond the suppliers a company deals with directly.
Almost nobody measures this accurately. Standard practice is to apply an industry-average emissions factor, a generic number assigned to an entire product category, regardless of who made it or how, like assuming every car gets the same fuel economy and building energy policy on that assumption.
In research published recently in One Earth, with colleagues from Cambridge, Sheffield, Birmingham, Tsinghua and UC Santa Barbara, we measured how far these averages drift from reality. We looked at ethylene, one of the simplest, most heavily produced industrial chemicals in the world, because if any product should be easy to estimate accurately, it is this one.
Across 63 real production processes sampled, actual emissions factors ranged from 0.23 to 7.3 kilograms of CO2 per kilogram, against an industry average of 1.4 to 1.6, a gap that can misrepresent a supplier’s true footprint by up to 600%, for the simplest material we could have chosen. For steel, cement, or electronics, the uncertainty only compounds.
In practice, a procurement officer working from an industry average cannot distinguish steel made from iron ore in a coal-fired blast furnace from steel made from recycled scrap in an electric arc furnace running on renewables. On paper, the two look identical. In reality, one can carry several times the emissions of the other.
Why faster reporting, including AI-powered reporting, hasn’t fixed this
Faced with the real cost of measuring emissions properly, many companies have turned to faster tools, including AI-assisted platforms that match purchases against a database and generate a figure in seconds.
This looks like progress. It is not: an AI tool handed poor data does not produce better data, only the same average delivered instantly, with a false sense of precision. Speed was never the constraint holding back credible accounting. Accuracy was, and remains, the constraint.
The consequences are no longer theoretical. Scope 3 emissions make up roughly 92% of the average European company’s reported footprint, yet decarbonisation efforts reach only about 37% of it, and just 3% of companies use real, supplier-specific data as their baseline.
This gap helps explain why 93% of companies with declared net zero targets must double their rate of decarbonisation by 2030, simply to meet commitments already made. The physical means to cut industrial emissions largely exist. What’s missing is data precise enough to show where to apply them.
What would actually solve this
Alongside colleagues who have contributed to the IPCC’s Sixth Assessment Report, and who advise the US Government, the Greenhouse Gas Protocol, ISO, and the Science Based Targets initiative, we set out three conditions every credible emissions measurement now needs to meet. We call them the CSA Principles.
Data must be credible: built from real, process-level information from the actual supplier, not a generic average, and verifiable, with security robust enough that suppliers are willing to share it.
It must be scalable: fast and affordable enough to reach thousands of suppliers, since a months-long assessment cannot cover a supply chain of tens of thousands.
And it must be actionable: specific enough to tell a procurement team which supplier to switch, with the ability to track whether that change reduced emissions.
In our work with the Japanese aerospace manufacturer AeroEdge, a full measurement built on real, supplier-specific data, completed in three weeks rather than the months a traditional assessment takes, identified changes that cut potential emissions by 41.4%, improved energy efficiency by more than 80%, and delivered several million euros in annual savings, alongside an 8% rise in profitability.
None of this required new equipment or lower output; it required knowing precisely where the waste was. Excess emissions are almost always a symptom of inefficiency hiding in a supply chain, and inefficiency costs money whether or not anyone is measuring its carbon.
Why this matters for Thailand specifically
For Thai exporters, this is more than an academic argument. The EU’s Carbon Border Adjustment Mechanism is already in effect, and the UK is preparing a comparable mechanism, requiring importers of steel, aluminium, cement, fertiliser, hydrogen and electricity to report, and eventually pay for, embedded emissions.
A shipment backed by real, verifiable data can demonstrate a genuinely lower footprint and pay a lower carbon adjustment cost. A shipment backed only on an average has no way of proving it is cleaner than a competitor’s, even when it is. In a global market for green goods forecast to reach $7 trillion by 2030, that distinction will decide who wins the contract.
Encouragingly, the same data that reduces exposure to carbon border costs also tends to cut costs more broadly, since the inefficiencies driving up emissions, wasted energy, wasted material, are usually the same ones driving up the bill. For Thai manufacturers competing on tight margins against subsidised low-carbon producers elsewhere in the region, finding and fixing that waste is not really a compliance exercise. It is a competitiveness one, and arguably the more urgent of the two.
Carbon accounting was set up to answer a simple question: how much is a company emitting. On that measure, it has done a poor job. On the question that matters, whether emissions come down, it has mostly failed. Fixing that will not come from writing more detailed reports. It will come from measurement systems precise enough to show companies where to act, and credible enough to prove it.
The views expressed in this article are those of the author and do not necessarily reflect the official position of Bangkok Tribune.
Founder and CEO, Neutreeno, a pioneering Cambridge-born cleantech venture.