Can You Trust a Company's Carbon Footprint?

Companies are revising carbon footprints for years they have already reported. Some changes correct errors, while others reflect improved data or different calculation methods, complicating comparisons of corporate emissions performance.

 

In July 2026, Deloitte counted 59 FTSE 100 companies that had amended greenhouse gas figures for earlier reporting years. Of those, 44 had revised Scope 3 emissions, which include emissions associated with suppliers, products and other activities outside a company's own operations. Deloitte's review covered annual reports available by the end of 2025, and the adjustments had different causes, including errors, changes in methodology and changes to businesses themselves. Anyone comparing a company's latest emissions figures with what it reported a few years ago may find that even the older years now look different.

 

A paper published in Nature Sustainability in July 2026 shows how large such differences can be without a mistake in arithmetic. Researchers took the same purchases of chicken, milk, eggs and wheat for a hypothetical food and drink company and calculated the associated emissions using four established databases. One produced a footprint of about 1.7 million tonnes of CO2 equivalent; another produced 3.5 million. The quantities purchased were identical. Changing databases between reporting years could therefore create an apparent reduction of more than half, although nothing about the company's purchasing activity had changed.

 

A company does not usually publish its carbon footprint alongside a second version calculated with alternative data. Readers see a single total, often presented to the nearest tonne, and a percentage change from the previous year. Some parts of that total may come from fuel records and electricity bills. Others rely on supplier disclosures or broad estimates for goods and services bought elsewhere. Those different sources of information are combined into one inventory, and the accounting choices behind them can have as much influence on its movement as changes in the business itself.

 

Behind the number

 

For purchased goods and services, carbon accounting commonly starts with an amount of material bought, or the amount spent on it, and an emission factor. The factor is an estimate of emissions per unit of activity or expenditure. Databases can differ in the countries they cover, the production processes they assume and the years from which their underlying figures are drawn. A company using a national average for a material may arrive at a different answer from one with information specific to the factory supplying it. Both may be following an accepted method, but the resulting figures are not automatically comparable.

 

The researchers also examined reported emissions data from 401 European companies. Their model left 62.5% of the year-to-year variation in reported supply-chain emissions unexplained by the business, industry and emissions-intensity variables they considered. For direct emissions, the unexplained share was a little above 5%. The larger figure does not mean that 62.5% of Scope 3 reporting is wrong. It is a measure of what this particular analysis could not account for, and other influences may lie outside the model. The contrast still shows the limits of interpreting a change in value-chain emissions without knowing what changed in the calculation.

 

The difficulty grows with the distance between a business and the source of emissions. A manufacturer will generally have records of the fuel it burns on site; it may have much less information about the production of the steel, chemicals or components it purchases. Suppliers may use different reporting boundaries, and some will have no inventory available at all. In those cases, industry averages and spending-based estimates allow companies to report a more complete footprint than they otherwise could. Those estimates can identify where further investigation is needed, but they are a less certain basis for claiming that a particular supplier has reduced emissions.

 

Figures revised later

 

A study by Harvard Business School professors Lauren Cohen and Ethan Rouen, published in Nature Climate Change in January 2026, examined historical Scope 1 emissions disclosures from S&P 500 companies. Among the repeated disclosures the researchers could compare, 58% of originally reported figures were revised. The researchers found more understatements than overstatements, with previously understated emissions exceeding overstated emissions by more than two to one in absolute terms, and an estimated 135 million tonnes underreported in earlier disclosures. The study does not establish that companies set out to mislead readers, although it found that explanations for revisions were often missing.

 

One company has since disclosed a particularly large calculation error. In February 2026, Vishay Precision Group (VPG), which makes precision measurement equipment, revised its reported Scope 1 emissions for 2023 from 12,740.2 to 1,262.28 tonnes of CO2 equivalent. An internal audit had identified a unit-conversion error that caused an incorrect factor to be applied to diesel and propane combustion. The original figure was just over ten times the corrected one. VPG's restatement notice concerns a reporting error, not a reduction in the fuel burned in 2023. A reader who used the earlier disclosure to compare VPG with another manufacturer would have been working with a very different picture of the company's direct emissions.

 

VPG's notice also shows what a useful correction can contain. It identifies the emissions category affected, the original and revised values, the reason for the error and the work being undertaken to prevent a recurrence. Other historical revisions can be harder to reconstruct, especially when new data is incorporated into a table without much explanation of the previous figure. For users trying to track progress against a target, the problem is then less the existence of a new number than the missing account of how it was produced.

 

Supplier Data

 

A different sort of adjustment was reported by TeamViewer in June 2025. The German software company described replacing broad spend-based factors with supplier-specific emissions intensities for a selected group of IT and professional-services suppliers. Those new factors were calculated from the suppliers' verified Scope 1, 2 and upstream Scope 3 figures, together with revenue data. For the suppliers assessed, TeamViewer reported a median reduction of 75% in calculated Scope 3 emissions. It was a change in the estimated footprint of that group, rather than evidence of a 75% fall in TeamViewer's entire value-chain emissions or a corresponding physical reduction at the suppliers.

 

For a buyer comparing suppliers, one whose footprint relies on an industry average may appear to have higher emissions than one with direct information from its operations, even where the underlying processes are similar. In other cases, replacing an average might increase the reported figure. That complicates attempts to rank bidders or assess progress across a purchasing portfolio, particularly when the emissions numbers come from different reporting years. The quality and coverage of supplier data matter as much as the total on the first page of a disclosure, and a buyer needs to know which parts of an inventory are estimates.

 

A reason to revise

 

There is a good case for companies correcting their historical data. Deloitte's July 2026 analysis found that errors accounted for 20% of adjustments to FTSE 100 sustainability metrics, compared with 26% in the previous review. Changes to methodology, business structure and reporting processes accounted for much of the rest. These findings do not establish that every new figure is more accurate. They do suggest that a rise in restatements should not automatically be taken as a rise in mistakes. A company that keeps using an outdated factor after securing better evidence would preserve an unbroken series at the expense of the best available estimate.

 

The GHG Protocol requires companies to establish and disclose policies for recalculating their base-year emissions when significant changes would undermine comparisons across time. The triggers include acquisitions and divestments, material errors, changes in calculation methods and significant improvements in the accuracy of emission factors or activity data. Companies set their own significance thresholds; the standard does not prescribe a universal 5% figure. Recalculating an earlier year can be necessary to assess whether emissions have genuinely fallen from a consistent starting point. It also demands an explanation, particularly where the revised total differs sharply from the number on which previous claims were based.

 

Independent assurance can help readers judge the information, although its scope needs careful attention. Deloitte reported that 86 FTSE 100 companies obtained assurance over sustainability metrics, but that figure does not establish that every part of those companies' greenhouse gas inventories was assured. A limited-assurance engagement also gives a different level of confidence from reasonable assurance, and neither removes the uncertainty inherent in estimates made with incomplete supplier data. What matters to a reader is which figures were examined, the criteria used and any significant exclusions. The word 'assured' on its own is an incomplete description of the work performed.

 

What the figures are good for

 

Carbon accounts are still useful for finding where a company's emissions are concentrated. A retailer may learn that much of its impact is associated with the goods it purchases, while a manufacturer may identify one material or process that deserves closer attention. Inventories also give businesses a basis for setting targets and discussing performance with lenders, investors and customers. They become less dependable when companies are compared to the nearest tonne despite differing data quality, or when a reduction is attributed to operational progress without examining whether the accounting method has changed.

 

A more useful disclosure would make the choices visible: the sources of emission factors, the use of primary supplier data, the assumptions that matter most and the reasons for significant restatements. Where an earlier figure has been changed, readers should be able to see both values and understand what led to the correction. That still leaves uncertainty, especially in Scope 3, but it gives the reported trend a meaning that a percentage alone cannot provide. Without that explanation, even a sharply lower carbon footprint remains difficult to interpret. A company's carbon footprint can be trusted about as far as its method can be checked.

 

Sources: Nature Sustainability (2026), Nature Climate Change (2026), Deloitte UK (2026).

Additional references: Vishay Precision Group (2026), TeamViewer (2025), GHG Protocol.