My first academic paper showed that the vast majority of companies publish emissions numbers without any communication of uncertainty. In short, as I argued in a previous post, most carbon accounting is currently a fugazi and, without uncertainty analysis, seems way more credible than it is.
The GHG Protocol, which sets the standards for how corporate carbon footprints are calculated, is in the middle of a once-per-decade update. Given that uncertainty analysis is still optional, I sent my findings to some of the people co-ordinating it. What followed was a long and useful email exchange which has left me fixated on one question: what do we all think GHG accounting is actually for?
Carbon accounting has largely been voluntary since its inception. Companies publish these numbers to publicly slap themselves on the back and signal that they are conscientious actors in the world. In organisational theory this is called “social licence to operate”; in the culture it’s called “virtue signalling”. It reflects the idea that organisations can only exist on condition of ongoing acceptance by wider society (I don’t fully subscribe to this idea; lots of organisations which are widely derided seem to be doing just fine). In this conception of carbon accounting, the numbers themselves are almost beside the point; it’s the act of reporting which creates value. Whether this thinking informed the founding mission of the GHG Protocol or ISO standards remains unclear to me, but in its woolly modern Mission and Vision statement the GHG Protocol says, “Our vision is that all private and public entities account for their GHG emissions, enabling an acceleration in reductions in line with the global warming limits required by climate science.”
That sounds good, but it doesn’t say how the numbers will enable those reductions, or answer the most fundamental question: how should these numbers be used?
The purpose of reporting these numbers should be to provide credible decision-making inputs. A consumer buying a new laptop, or a procurement team evaluating potential suppliers, might pick one over the other directly on the basis of the numbers which were reported. Not only the companies themselves, but investors, policy makers and consumers take millions of decisions like these, and we need them to organise our economies in the most carbon-efficient way possible. A series of well-informed decisions is the only way we can get from our current unsustainable society to a society that has limited global warming in alignment with the climate science, and nearly all good decisions rely on comparing the viable alternatives.
When I put this to the GHG Protocol representatives, I realised that my core conception of carbon accounting was not shared. In response to my argument that uncertainty analysis should be mandatory to enable comparison, they proposed tiers of requirements: if the numbers were only being presented in an informational context they shouldn’t require uncertainty analysis, but if they were being presented to underpin a specific claim (e.g. “we have reduced our carbon footprint by 50%”) they should.
While I welcome any change which introduces uncertainty analysis, this slightly baffled me. It recognised that uncertainty analysis would be valuable as evidence for backing up a claim, while implying that there are situations in which the reported numbers don’t matter and aren’t likely to be used to take decisions. This seems nuts to me. Why should we ask anyone to dedicate resources to producing the numbers in that case, and what impact will they have?
Product footprints are where this type of thinking breaks down most clearly. The laptop manufacturer doesn’t have to make any claims, but its reported numbers directly lead to a decision, a decision which may be incorrect if the numbers turn out to be incomparable on the basis of underlying methodologies, data and uncertainties. And this idea of carbon accounting numbers being reported in some sort of neutral informational space, not intended for external decision-making, is something I’ve come across before.
I’ve had arguments time and again with carbon accounting and LCA experts who reject comparability across organisations and products outright, and argue that the data is just an informational tool by which internal decision-makers can better understand their own impacts. That is a fairly unambitious vision for the discipline. It leaves emissions in the hands of a small group of people with a moral imperative for improving their own products and processes, who also have to consider their costs and timeframes, and who are impeded by that deep resistance to change that exists within most people. Can we honestly expect that to bring about the change we need? It also excludes everyone else. It excludes the consumers, who want to compare two products with the same function on a like-for-like basis and pick the one which they know to be of lower impact. It excludes investors, who want to screen thousands of companies for their carbon performance and direct their resources to the best-performing ones. It excludes politicians, who want to try and reduce emissions across a whole sector of their national economy. Without comparability, none of these decisions can be taken.
We have completely lost sight of any wider theory of the purpose of carbon accounting (if there ever was one). The discourse is now dominated by people arguing about incredibly niche methodological complexities. In the meeting notes emerging from the GHG Protocol’s update, scores of carbon accounting boffins appear to be engaged in a Hunger Games-like contest to have their chosen methodological ideas win: chain of custody models, recycling methodologies, electricity accounting, biogenic carbon, the list goes on. Where is the working group that is investigating if carbon accounting as it currently exists is fit for purpose? I am not trying to criticise individuals who care deeply about the consistency and rigour of the methods; I have often fallen into this trap myself (the irony of nesting this criticism in the middle of a treatise on the importance of uncertainty analysis is not lost on me). But I was alarmed to discover this formless conception of the purpose of carbon accounting right at the heart of the discipline. Most people involved, including the standard-setters, seem to care more deeply about the process than the outcome which we are collectively trying to achieve.
The goal is the system-wide reduction of emissions; the carbon accounting numbers are just a means. In that light, a good argument can be made for less complexity in the standards, not more, and more focus on genuine comparability between reported numbers. We should also be honest in admitting that carbon accounting is currently a quasi-scientific discipline, as, to some extent, is LCA. We’ve borrowed the aesthetics of science (precise numbers, standardised methods, technical working groups) while jettisoning one of its foundational commitments: honestly representing what you do and don’t know, and what you can and can’t infer from data. LCA is better at this than carbon accounting, but we still have a long way to go across the board.
That is why I care so much about uncertainty analysis. If one number said that a product produced 10 tonnes of carbon dioxide, and another number said 11 tonnes, we are likely to pick the first one, unless we understand that it is a very uncertain estimate and the real value could be anywhere between 5 and 20 tonnes. Uncertainty analysis improves comparability by acknowledging the things which you don’t know, and the limitations of your methods and data. If those principles aren’t at the core of the discipline, then carbon accounting is little more than a marketing tool.
We are in the middle of a rare opportunity to steer the discipline towards outcomes. For the past 20 years it didn’t really matter what the numbers were for because they were voluntary and not universal, but a slate of new legislation across the world, including CSRD, SB 253 and SECR, explicitly defines these numbers as inputs to external decisions. They have the right idea; our standards and best practices now need to deliver on it.
Uncertainty analysis is one small piece of this puzzle, alongside removing unnecessary flexibility, loopholes and optionality in the types of data and methods that you can use. The main thing we need is to be relentless in reforming the discipline in pursuit of a set of stated outcomes, which we all agree on and could explain clearly to a non-expert in 30 seconds without hesitation, repetition or deviation. Then people can easily do the things that are within their gift, like picking the right suppliers, deploying financing, or designing national policy which actually accelerates reductions in emissions.
If we can’t do that, then what exactly is carbon accounting for?
Update (August 2026): I followed this argument further in The measurement-industrial complex.