Redefining KPIs for meaningful results

We work with industrial clients to improve how performance is measured and managed. Drawing on extensive experience in energy-intensive sectors, we often see that while Key Performance Indicators (KPIs) are widely used, they rarely lead to meaningful changes, especially in the case of Energy Performance Indicators (EnPIs).

All organizations use KPIs to monitor, evaluate, and improve their efficiency and operations. Within the framework of ISO 50001, KPIs are called Energy Performance Indicators, and they are supposed to serve as critical metrics that quantify energy performance relative to key variables such as production output, operating hours, or environmental conditions.

The problem with traditional metrics

The problem is that most EnPIs out there are just numbers for the sake of having numbers for the energy management system. They look nice in reports and dashboards, but when it comes to actually driving improvement, they often miss the mark. Too many companies track energy per unit produced or monthly kilowatt-hours without ever asking: “So what?”

If the data doesn’t lead to action, it’s basically a waste of time. A real improvement needs real insight—not just metrics that sit in spreadsheets gathering digital dust.

We who work with reporting energy efficiency tend to spend a significant amount of time explaining why the progress with our EnPIs is what it happens to be. This time could be used for doing something more productive.

We keep trying to define optimization and operational improvement using broad, global EnPIs—like total energy per ton produced—as if they hold all the answers. But the truth is that those high-level numbers are too vague to tell us what’s really going on.

In short, most EnPIs do not fulfill the criteria for a good KPI, which are that EnPIs should:

• Give answers by translating raw energy consumption data into information on what actions they should take;
• Be relevant for the person using the EnPI, which means that they must be able to influence it;
• Provide timely information so actions can be taken at the right moment;
• EnPIs need to be used.

In an ideal world, EnPIs should be defined at different organizational levels and serve different purposes according to the receiver of information and the type of measurement.

The reason for having different levels of EnPIs is that what is “relevant” depends on the role one has within an organization. Only EnPIs that one can actually influence are really “relevant”.

It is important to remember that there are also other stakeholders who need information on energy consumption—for example, for cost allocation and similar purposes—but these are not considered proper EnPIs and are therefore not addressed in this article.

Shifting the focus: smarter indicators

As discussed earlier, our four favorite criteria for a good EnPI are the following:

• Gives answers
• Relevant
• Timely
• Must be used

Let’s consider these four criteria one at a time, starting from the last one:

Must be used

It is easy to understand that even if we have the most perfect EnPI, it is of no use if nobody follows it.

Ways to achieve success:

1. Involve the “customers” of the EnPIs early on in the process.
2. Choose tools for both following up and reporting that are easy to use.
3. Establish routines on how EnPIs are followed up and how corrections to the process, operation, and/or EnPIs are executed.

From the start, it’s essential to involve the people who will use these indicators in their day-to-day decision-making. These internal stakeholders—whether in operations, sustainability, or quality management—understand the nuances of their work and can help shape indicators that are both practical and aligned with business goals. When stakeholders recognize that the indicators reflect their actual needs, they are far more likely to use them effectively and contribute to continual improvement.

Equally important is choosing tools that make EnPI tracking and reporting seamless. Simple, intuitive platforms encourage routine interaction with data, making it easier to detect trends and take action. It is needed that not only energy data are available in the tool but also influence factors.

Timely

Retrospective monthly reports are only that useful. It is difficult to achieve the best possible results if we get information of poor performance with an average of two weeks of delay especially.

There are two ways of making the information more useful:

1. Using a time machine and/or

2. Having the information available in real time.

The Technology Readiness Level (TRL) of IT tools showing process values in real time is much higher compared to the TRL of time machines, so we recommend starting with IT tools.

Relevant

EnPIs are not really relevant if they do not have a significant impact on the performance you are trying to measure and/or if you cannot influence the outcome. EnPIs need to be designed at different levels in the organization, so they are relevant to the receivers.

The vanilla versions of specific energy consumption (SEC), e.g., kWh/ton, fail to meet this requirement. Different products need different amounts of energy in order to reach the product specifications. It is highly unlikely that it would be acceptable to refuse production of a certain product in order to meet the energy efficiency targets.

Gives answers

As we all know, there are plenty of factors that influence the SEC, and this is the reason why we often fail to get any answers when monitoring the SEC. Instead of answers, we have plenty of questions: the SEC might have increased due to colder temperatures outside, lower Overall Equipment Effectiveness (OEE) on our processes, or any other of the factors that influence it.

One way to address this is to follow ISO 50006 (Measuring energy performance using energy baselines (EnB) and energy performance indicators (EnPIs) — General principles and guidance), identify the relevant variables, and normalize the KPI/EnPI accordingly (see ISO 50006 for details). Unfortunately, in our industry, this method often does not take us very far.

All too often, we lack the means to measure all the relevant factors, and if we cannot measure these variables, then we cannot truly model them by using statistical methods. Our statistical models are therefore not complete, which makes it extremely difficult to make well-informed decisions based on the data we can get from the typical EnPIs.

Should we give up using energy consumption as basis for monitoring energy efficiency? The short answer is no, but we need to be aware that just measuring specific energy consumption is typically not a good EnPI.

The good news is that there might be an easier way of crafting a good EnPI, e.g., by measuring wasted energy instead of consumed energy.

Looking ahead: key principles

How to measure wasted energy?

There are numerous methods available to measure wasted energy in industrial processes, ranging from straightforward calculations to complex data-driven models. However, it is often most effective to start with the simplest viable approach. In this context, that means beginning with first-principles models. These models are rooted in fundamental physical laws, such as mass and energy balances, and they offer a transparent, consistent, and theoretically sound framework for identifying inefficiencies. They are especially useful for gaining a deep understanding of the process and for pinpointing where and how energy losses occur.

Only in cases where first-principles models are insufficient – either due to limitations in available data, the complexity of the system, or the dynamic nature of certain processes – should one consider moving toward more sophisticated statistical or machine learning models. While these approaches can be powerful, they often require significant amounts of high-quality data, and their outputs may be more difficult to interpret or validate without a strong physical basis. Also, as discussed earlier, we often lack reliable sensors that can monitor significant variables affecting the EnPI we so desperately want to have.

Exhaust air from the Yankee hood is a good example of a process parameter that provides a clear indication of overall process efficiency. We can be sure that we are wasting energy if the humidity level and/or temperature level are off. It is often relatively easy to develop a Standard Operating Procedure for operators to follow so that the EnPI based on these parameters can be adjusted to a reasonable level.

Another best practice that tends to work well with EnPIs is to try to monitor savings potential in kWh and aggregate them from individual processes to department levels and so on. It becomes much easier to prioritize the operational improvement work when it is easy to see where we have the biggest losses in our processes.

We believe in combining engineering fundamentals with modern tools in a pragmatic and effective way. With decades of experience in energy-intensive industries, including pulp and paper, we help our clients improve energy efficiency by selecting the right methods for the task – prioritizing clarity and accuracy without unnecessary complexity.

Authors: César Nuez, Director Zaragoza, Global Division Industry & Tom Karlson, Group Manager, Global Division Industry

This article was first published in Tissue World Magazine's July/August 2025 issue.