The energy-savings paradox – they don’t exist, but we still have to measure them

Energy savings are essential for the sustainable transformation of buildings. They directly decrease greenhouse gas emissions, reduce operating costs, increase asset value and resilience. We need them more than ever – but in fact they do not exist.

They are among the most frequently quoted achievements in sustainability reports, yet they remain one of the least tangible. Unlike a new photovoltaic panel or insulation layer, savings cannot be seen or touched. They are, in essence, the energy that would have been consumed, but wasn’t. This logical twist makes them both powerful and paradoxical: how can we measure something that never existed?

What ‘energy savings’ really mean
When we say a building ‘saved’ energy, we mean that its energy consumption was lower than it would have been without certain actions – such as upgrading HVAC systems, retrofitting lighting, or improving controls. However, we never actually observe that ‘alternative reality.’ Once the upgrade is in place, we can only measure the new consumption, not the old one under unchanged conditions.

This is why, in strict physical terms, energy savings do not exist. There are only differences between two scenarios: the actual one (after the improvement) and the hypothetical one (what would have happened without it). In technical language, these differences are ‘avoided consumption’, sometimes referred to as negawatthours.

The common market practice
In the real estate market, ‘energy savings’ are often presented as a simple before-and-after comparison. If a building used 1,000 MWh last year and 800 MWh this year, one might say that 200 MWh were saved. The problem is that buildings are dynamic systems. Energy use changes continuously due to occupancy, weather, operating hours, or internal loads. Without adjusting for those factors, we risk attributing all the change to the retrofit, when in fact it may stem from milder winters, fewer tenants, or operational changes.

To compensate, consultants or energy efficiency measures providers often rely on simplified benchmarks or historical averages. These can be useful for a first estimate but are not accurate measurements. In other words, such ‘savings’ are claims, not verified results.

How proper measurement should be done
Reliable quantification of savings is only possible through a structured process known as measurement and verification (M&V). The most widely recognised frameworks are: ASHRAE Guideline 14 – 2014: Measurement of Energy, Demand, and Water Savings, and IPMVP (International Performance Measurement and Verification Protocol). Both standards describe how to define the ‘baseline’ (expected consumption without changes) and how to compare it with post-retrofit performance.

If a building has a good system of energy meters, the process is relatively straightforward. We collect pre- and post-implementation data, correct it for relevant variables – most often outdoor temperature, occupancy, and operating hours – and then model the baseline using regression analysis. For example, if heating consumption correlates strongly with heating degree days (HDD – measure of how much and for how long the outside temperature was cold enough to require heating) the baseline model will use that relationship to estimate what energy would have been used in the post-retrofit period under the same weather conditions. The difference between that modelled baseline and actual use gives us the verified savings, complete with confidence intervals and uncertainty.

When metering is limited
In many existing buildings, sub-metering is minimal or non-existent. Only total gas or electricity bills are available. In such cases, M&V is still possible, but the level of uncertainty increases significantly. Without granular data, it becomes harder to isolate the impact of the efficiency measure from other influencing factors.

Here, the concept of error becomes crucial. Presenting a single value, such as ‘15% energy savings,’ is misleading when data is incomplete or the baseline model is weak. Every calculation has an associated statistical uncertainty that reflects how confident we are in the result.

For example, an analysis might yield a saving of 15% ± 8%. This means the true value likely lies within the range of 7% and 23%, assuming a 68% confidence interval. The larger the uncertainty, the less reliable the figure. Both ASHRAE Guideline 14 and IPMVP emphasise that savings results should always include an estimate of uncertainty. Reporting a single percentage without an error range is poor practice, as it hides the reliability of the measurement.

The best practice is to make sure that the calculated savings are at least twice as high as the error. Verified savings are outside of the error range and they should be greater, to give confidence that they really occurred. Some sources and industry practice state that the savings should be even bigger – four to five times greater than the error.

When savings’ calculation did not include statistical error, claimed savings should be classified as ‘unverified’. Such unverified savings may serve educational or indicative purposes, but they must not be used for financial contracting, energy performance guarantees, or ESG-linked reporting. Without verification, there is no evidence-based foundation for payment, credit or recognition.

Summary: resolving the paradox
The paradox of avoided energy consumption lies in the fact that we celebrate something that never physically occurred. Energy savings are not measurable entities; they are the difference between the real and the hypothetical. We cannot touch them, yet we depend on them to prove progress. The paradox is only resolved when we stop treating ‘savings’ as physical units of energy and start treating them as statistical constructs that must be verified through evidence and uncertainty analysis.

To improve, we must do three things:

Accept uncertainty as inherent. Energy savings are estimates, not facts. Every reported value must include an error range showing how reliable it is. A single percentage without an uncertainty interval is meaningless. Always state whether the savings are verified or unverified.

Upgrade metering and data quality. Without sufficient sub-metering, savings verification remains guesswork. Buildings need detailed energy metering – by system or function – to build credible baseline models and reduce uncertainty.

Follow recognised verification standards. Apply ASHRAE Guideline 14 or IPMVP to define baselines, adjust for external factors (weather, occupancy, schedules), and calculate confidence intervals. Only savings verified under these protocols should support financial decisions, energy performance contracts, or ESG reporting.

The path forward is methodological, not rhetorical. We cannot ‘create’ savings, but we can measure avoided consumption with enough rigour to make it actionable. In doing so, we turn an abstract paradox into a foundation for responsible energy management and transparent climate action.