Abstract Background The IEA has described energy efficiency as the ‘first fuel’ in the clean energy transition. Although improvements in energy intensity (i.e. energy use per unit of economic output) averaged 0.8% per year between 2000 and 2010, and 1.7% between 2010 and 2020, rebound effects often negate these gains. While the literature suggests a much higher potential of 15–22%, these studies are either bottom-up analyses of individual industries or cross-sectional, top-down assessments. Systematic bottom-up assessments remain scarce, particularly in non-energy-intensive sectors. This study addresses this gap by analysing 106 standardised, on-site, technology-based energy assessments conducted between 2015 and 2024 in sixteen non-energy-intensive industrial sectors. Potentials were identified by comparing the equipment in place with the best available technology, providing a bottom-up estimate of the efficiency gap. These potentials were then grouped into categories to facilitate comparison with the literature. Results All audited sites showed an average potential for economic energy efficiency improvement of 17%, equivalent to 303 GWh per year of energy savings. The largest contributions came from heat recovery (4.2% points) and process optimisation (3.8% points). Heat recovery, either directly or via heat pumps, proved to be the most significant lever for reducing CO₂ emissions. Conclusions Our data suggest a larger economic energy efficiency potential than is generally recognised in the literature, and support various studies that have identified an ‘efficiency gap’. The economic analyses we performed for each potential option further support this. However, in line with the existing literature, we confirm that high discounting and a shortage of skilled personnel required for implementation are the two main hurdles. By conducting a more in-depth analysis than previous studies and comparing actual energy use with both minimum physical requirements and the best available technology, we expand the scope and depth of the literature’s findings.
Marx et al. (Wed,) studied this question.