OEE (Overall Equipment Effectiveness) is a fundamental KPI and one of the most widely used in industrial production environments.

Its formula is simple: OEE = Availability × Performance × Quality and, on paper, it seems to offer a clear view of a plant’s actual performance. 

But there is one problem that is rarely discussed: OEE is also one of the KPIs that we tend to misrepresent the most, sometimes consciously… and many other times unintentionally.

As in a game of solitaire, it is possible to “cheat” to make the game look better than it really is. Adjustments here, generous interpretations there… and suddenly we have a very high OEE that does not reflect the reality of production.

Precisely to avoid this, we must be as accurate as possible when analyzing it (regardless of whether we use the classic OEE formula or a variant), so that we avoid biased interpretations that can give a false sense of efficiency.. 

✍️ In this article, we explore the most common traps of OEE, how to spot them, and why measuring it inconsistently can be worse than not measuring it at all. 

👉 First things first: OEE is not unique

The first thing to bear in mind with regard to OEE is that there is no universal standard in companies; there are almost as many formulas as there are factories.

And although it is common to use the same formula, the interpretation of each factor can vary greatly from one plant to another.

The conclusion is clear. Two factories with the same OEE will not necessarily have the same level of efficiency. Even within the same company, different plants may measure it differently. 

This makes OEE a very powerful KPI… but also very easy to misinterpret if you don’t know how it is calculated.

If you want to learn more about how to calculate OEE, visit our blog on the subject. 

🧪 Once the formula is set, the trap is set

Getting down to business, below we explain some of the most common traps you may encounter when calculating production OEE.

We can usually differentiate between two types of traps:

• Modification of any of the variables, which means that certain efficiency losses are not taken into account: redefining the planned time, using a theoretical cycle time that is not real… 

• And other traps that shift inefficiencies from one OEE component to another: if I do not record certain stoppages, it will not be a loss of availability, but it will affect performance. 

♥️ Availability Traps: 

Availability measures how long a machine is operational compared to the total planned time.

Availability is defined as:
⚙ Availability (%) = Operating time (without unplanned stoppages) / Total planned time

Simple in theory, prone to inaccuracies in practice. Let’s look at some examples: 

🔹 Creative reclassification of downtime 
This occurs when we modify the way downtime is classified, reducing losses without actually improving anything.

Among the most common reclassifications are: 

• Label unplanned stoppages as planned: this reduces planned time and therefore improves availability. 
  
• Consider all setup / changeover / break times as planned: it is true that a break or start-up time is a planned stoppage, but it should have a theoretical stoppage time, which, if exceeded, implies a loss of availability.
  
  Example: if a reference change should take 20 minutes but actually takes 30 minutes, those extra 10 minutes should be considered a loss of availability. It is common not to do this. 
  
• Ignoring ‘different’ shifts: there are shifts in which overtime is worked or which are longer (or shorter) than usual. It is a common mistake to always assume that our planned time is the same and not to consider these special cases.  

🔹 Not having the actual times of stops 
It is very common for stops to continue to be reported or measured manually. This means that the amount of stop time is not completely reliable data. As a result, we deceive ourselves about availability by not seeing the whole picture: estimated stoppages, small stoppages that are not counted, human errors, etc. 

🔹 Ignoring micro-stoppages and insignificant downtime  
Small downtimes, such as minor adjustments or cleaning, are often underestimated, but they can add up significantly. A few seconds multiplied by thousands of cycles can add up to hours of lost time. In this case, ignoring these downtimes does improve availability, although it should penalize performance.

 

These little tricks can inflate availability by 5% to 15%… without it actually being the case. 

♦️ Performance Traps:

Performance compares actual speed against ideal speed. Here, the trap is usually more technical.

Performance is defined as: 
⚙ Performance (%) = (Ideal cycle time (reduced speed) x Total parts produced) / Operating time (without unplanned stoppages) 

Some of the most common traps: 

🔹  Using a theoretical speed reference that is lower than ideal 
This is the industrial version of lowering the bar. If we adjust the ideal cycle time speed downwards, it will be easier to meet it. 

• Taking the cycle time on a good day as the ideal. Ese día ha tenido microparadas, ajustes y pequeñas ineficiencias. Si se usa como referencia, se convierte en un estándar artificialmente bajo. 
• Maintaining obsolete standards if a machine has been optimized, modernized, or new tooling is used. 

 🔹 Not segmenting production cycles well according to the reference  
It is not uncommon to find the same theoretical cycle time on a machine. If this occurs on a machine that manufactures different references, it is very possible that, depending on the product being manufactured, a better or worse value will be obtained and the resulting performance will not accurately represent reality.  

♣️ Quality Traps:

Quality considers manufactured parts that do not meet quality standards (including those that need reworking).

Quality is defined as: 
⚙ Quality (%) = OK parts (excluding defective parts and rework) / Total parts produced 

Quality should be the most difficult aspect to manipulate… but this is not always the case:

🔹  Hidden or unregistered defects  
By hidden defects, we refer to those parts that apparently meet specifications and are classified as OK parts by the machinery, but once outside the process are classified as NOT OK by the quality department.
This subsequent discrepancy will affect the actual quality of our production, and if these defects are not recorded, we will have an inaccurately inflated quality rating.

🎴  Special mention: Reworks and rejects

It is very common to be unsure about how to deal with this type of process, as we do not know whether a reworked part should be considered good or bad, and how this affects the process.

The truth is that it will depend on the process. We must not forget what OEE is: it is the measure of machine/line efficiency. Therefore, a part that comes out bad and is put back in for reworking should appear in the OEE calculation as two separate parts (one pass that was good and another that was bad).

Example: let’s assume a machine that has no downtime (100% Availability) and that works perfectly at theoretical cycle time (100% Performance). And that after producing a part, we have to put it back into the machine for reworking. If we only consider that there was one good part, we will get an OEE of 100%, when this is not true, because the quality was actually 50%: one part was bad and the next was good. 

That is why you must be aware of the following traps:

🔹 Mistaking rework with good production  
If a reworked part is only counted as good, you alter the Quality value and the OEE no longer represents reality.  

🔹 Overestimating First Pass Yield  
If minor defects are omitted, the indicator may show a higher quality than it actually is.   

♠️ Other less obvious traps

🔹 Lack of automatic data capture systems
When data depends on manual records, new possibilities arise for manipulating information, whether consciously or not:   

• Human errors: The person taking notes may make mistakes, especially if they are under pressure.
• Biases: The operator records data when they remember or when they think it is significant. And humans have a number of biases in our perception that affect us. We tell you more about this in our blog on Problems in data projects.  
• Temptation to adjust: This can be conscious (due to bonuses, pressure) or unconscious (desire to show good performance). 

Manual measurements can give very high OEE values… until an automatic system is installed, and the values drop significantly.  

🔹 Lack of context or segmentation 

• Using a plant’s overall OEE is using an incomplete indicator. An overall OEE is useful for getting a general picture of the plant, but not for implementing improvement actions.
  The reality is that, by averaging different realities, the overall OEE can hide operational problems such as bottlenecks or where real losses occur. A highly efficient line can mask another that is malfunctioning, one shift can compensate for another, very different products can distort average performance, etc. 
• Segmenting OEE by line/machine/shift/product will provide useful information for improvement. By breaking down OEE, we will be able to see which parts of the process are working correctly and which need improvement.
  Segmentation thus turns OEE into a very useful diagnostic tool: it will give us the availability of each of the different machines, the performance of each of the shifts, whether there are any particularly difficult references, etc.
  This allows us to act with precision: allocate resources, prioritise investments, adjust speeds, improve training or optimise processes.  

🧠 So… what is a good OEE?  

Real-world factories are not perfect. Therefore, these values are usually more reliable: 

Whenever we see an OEE > 95% in a manual plant, with a variety of products and multiple shifts… it is worth checking where the trap has been set.  

Once we have refined our OEE calculation by eliminating traps and inaccuracies that distort the reality of this KPI, we must remember one fundamental thing: the most important thing, even above the OEE value we have, are trends.
This is a key idea that is often overlooked. 

🚀 A poorly compared OEE is useless; a consistent OEE is valuable

What do we mean by this? Even if a factory calculates OEE differently than usual (as long as it is honest and consistent), what really matters is that the calculation does not change over time. 

Because absolute comparisons of OEEs between companies will not tell us anything. Internal comparisons will. 

 In fact, if you always calculate OEE in the same way: 

• You will detect rapid drops in any of its values.
• You will easily see improvements after an action.
• You will better identify patterns and correlations.
• You will be able to cross-reference it with stoppages, shifts, products or lines, obtaining other valuable indicators.. 

It’s like weighing yourself on a scale that is 2 kg off: as long as it shows the same bias, it can be used to see if you are gaining or losing weight. Therefore, rather than obsessing over a perfect OEE, it is better to focus on ensuring that the OEE is well measured, well explained, and always calculated in the same way. 

📊 What about Grafana for OEE? The power of visualization 

Once the OEE information has been correctly captured and contextualized (with solid, well-structured data), we will need to add a visualization layer that allows the team to quickly interpret what is happening, detect patterns, make comparisons, and make quick decisions.

This visualization can be supported by tools such as Grafana, which is increasingly present in industrial environments. And while it is true that Grafana does not calculate OEE itself, it can become an excellent showcase for displaying it: dynamic dashboards, comparisons by shift or line, historical trends, visual alerts, panels accessible from any device… all in a flexible and customizable environment. 

Grafana is therefore a flexible and powerful way to democratize OEE, make it visible, and turn it into a real working tool, not just a number in a report.

🧩 Conclusion: An honest OEE or nothing

OEE is a powerful indicator… but only if it is always measured in the same way. Not measuring this KPI consistently is almost like not measuring it at all: it hides problems, slows down continuous improvement, and creates a false sense of success..  

The most competitive plants are not those that report perfect OEE, but those that have reliable data, without cheating or creative interpretations, and use that value to improve every day.
Furthermore, we must not forget that what will truly bring us value is seeing how our OEE evolves over time: whether it rises, falls, stagnates or responds to improvement actions

As in solitaire, you only win if you don’t fool yourself.
In industry, it’s the same: an inaccurate OEE doesn’t measure, it deceives. So it’s worth the extra effort to be consistent and systematic so that the trends tell us the truth. 
 

Need help with your OEE? 👇