Traditional methods of estimating corporate carbon emissions are dramatically underperforming when it comes to Scope 3 emissions, according to a study by UK-based climate intelligence firm Carbon Responsible.
The company’s research shows that commonly used Environmentally Extended Input Output (EEIO) models can overstate Scope 3 emissions by as much as 2,480% compared to verified data. The findings are based on a benchmarking analysis using 2023 verified emissions data from a sample of FTSE 100 companies.
Carbon Responsible compared these figures against outputs from Ada, its AI-powered emissions engine. While EEIO estimates diverged wildly from real data, Ada reduced inaccuracies to just 80%—making it 30 times more accurate than traditional models and achieving a 97% improvement in precision.
“This represents a step-change in emissions measurement capability,” said Matthew Paver, COO of Carbon Responsible. “When you’re 97% more accurate than the industry standard, you’re no longer in the realm of estimation – you’re capturing investment-grade data.”
The EU’s Corporate Sustainability Reporting Directive (CSRD) and the US SEC’s climate rule are both mandating traceable, auditable emissions data—pushing firms away from outdated or opaque methods and raising the stakes for data accuracy.
Scope 3 emissions, which encompass indirect emissions such as those from supply chains, investments and product usage, can account for over 80% of a company’s total carbon footprint. Yet they remain difficult to track. EEIO, though endorsed by frameworks like the GHG Protocol and PCAF, relies heavily on dated, spend-based data that can lead to flawed reporting, according to the providers.
26% of European asset managers fail to disclose Waci
“We’re seeing emissions disclosures based on proxies from six years ago – and regulators are starting to notice,” said Paver. “What’s being accepted in ESG reports today would never pass audit in a financial statement. Procurement, risk modelling and capital allocation – all now demand a firmer grasp of emissions data. Even if it feels like attention is shifting away right now, the CFO’s office can no longer afford to treat carbon as a compliance footnote.”
“There’s a widespread misconception that Scope 3 data is just a best guess, but if you’re setting net zero targets, allocating capital, or filing regulatory disclosures, guesswork isn’t good enough,” he added.
Ada was launched this month to address what Carbon Responsible describes as a “crisis of credibility” in Scope 3 reporting. Built for investment-grade analysis, the platform claims 90% accuracy across portfolios and supply chains, using a proprietary dataset of over 14,000 verified company records. All insights are generated in real time, using machine learning and excluding emissions data older than two years.
“You wouldn’t invest based on a credit rating built from an industry average in 2017,” said Paver. “So why are firms making climate decisions based on the same approach?”
The Ada platform prioritises evidence-based reporting, using primary data wherever possible and transparent proxy logic where necessary. Each data point includes audit trails and methodological clarity—aiming to eliminate the ambiguity that plagues many ESG datasets.
“What we’re witnessing is the end of ESG’s ‘honour system’,” said Paver. “When your AI solution is 30 times more accurate than traditional methods, it transforms not just reporting, but the entire approach to climate risk management and decarbonization strategy.”
