Stop Massaging OEE: Measure the Whole Truth or Miss the Real Opportunity
Overall Equipment Effectiveness can be one of the most powerful indicators in manufacturing, or one of the most misleading. When companies exclude planned downtime, changeovers, scrap, lack of demand, or short stops, OEE stops showing the real performance of an asset and becomes a customized comfort metric instead. This article makes the case for a more honest approach: keep the full data visible, make every exception explicit, and use an AI advisor such as the Maneva Orchestration Platform to reduce bias while preserving the information leaders need to find the improvement opportunities that convenient definitions tend to hide.
Why OEE can mislead as much as it helps
OEE can be a fantastic indicator or a terrible one. It can help you improve your process, or it can fool you completely, and you may not even realize which one is happening. On paper, the OEE calculation has a short, simple definition: availability multiplied by performance multiplied by quality. But the moment you start debating what to include or exclude, you have already started to compromise it. Every exclusion feels reasonable in isolation, a changeover here, a quality hold there, but each one quietly narrows what the number represents, until it stops describing your equipment and starts describing your assumptions.
A simple analogy: your investment return with exceptions
Imagine you go to the bank, and your manager promises you a 3% risk-free investment. You think, that's fantastic, and since it sounds correct, you accept the deal. A year later, when you come to collect your return, you discover you only earned 1.5%.
When you complain, your manager explains, with the most satisfied expression you have ever seen, that the rate was indeed 3% per year, but the bank doesn't operate on weekends or national holidays, he took a month of vacation, and 10% of the remaining time the branch had no customers at all. Add it up, and barely half the days were actually left for your money to grow. But during those days, it truly was 3% per year.
The danger of excluding reality from the metric
By now you see where this is going. OEE is supposed to measure the overall effectiveness of your equipment, not its effectiveness only on the days you happened to choose. It exists to tell you how much return your capital investment is generating relative to its absolute maximum potential. Every time you discount planned downtime, lack of sales, or any other exception, what's left is a distorted number that only makes sense inside your own company, on your own terms, and buried inside that distortion are the improvement opportunities you can no longer see.
When definitions hide the factory floor
I remember walking a customer's plant floor once and seeing scrap everywhere. I asked him: what's your OEE? He answered immediately: 90 to 95%. So I asked a different question: what's your scrap rate? He said 50%, but they had agreed to exclude rejects from the OEE calculation. I have met another customer who excluded any stoppage shorter than ten minutes, and another still who excluded every changeover entirely. Each exclusion, on its own, seemed defensible. Together, they made the number meaningless.
That is why I consider myself an OEE purist: nothing should be excluded. Vorne, whose world-class OEE benchmark sits around 85%, defines the metric against the full calendar of potential production time for exactly this reason. You are free to disagree, and most people do, especially when OEE feeds directly into their bonus.
But the real problem isn't the definition itself; it's that humans are emotional beings, unable to stay fully unbiased and fair, so our own definitions start to drift over time.
How an AI advisor keeps you honest
An AI advisor such as the Maneva Orchestration Platform can solve this problem, partially or completely. It is bias-free, no matter the day, and no matter how good or bad the result looks. More importantly, whatever exceptions you want to apply to your OEE calculation, you have to make them explicit, and those rules then get applied consistently, everywhere, every time.
That might sound like a small improvement, but it is a significant one. Even if you decide to heavily massage your OEE so your Christmas bonus is safe, you still have to spell out those rules to the AI. It will report the OEE the way you asked for it, but it will also keep track of every number you pushed under the rug, so the day you finally decide to tackle the problems you have been ignoring, all the data is already there and ready for you.
The Orchestration Platform goes a step further: it acts as a real advisor, proactively challenging you on the issues you have been ignoring and pointing out the gains you are leaving on the table. This is the hidden-data-gap problem Carl Tokarek described in what your OEE score is not telling you, now solved by keeping the complete data set preserved underneath whatever definition you choose. You get the best of both worlds: the freedom to define the rules that govern your OEE, and the certainty that nothing gets quietly buried. That is very different from a "humanized" interface, because humans tend not to keep track of the metrics nobody asked to see.
Conclusion
OEE was never meant to be a comfort metric. It was designed to answer one honest question: how much value is your equipment actually returning on the investment you made in it? Every exclusion you allow (planned downtime, changeovers, scrap, short stops) buys short-term peace of mind at the cost of long-term visibility. The opportunities do not disappear when you exclude them from the calculation; they simply move out of view.
You do not have to give up the flexibility to define OEE in a way that fits your business. What you do have to give up is the illusion that convenient definitions come without cost. Tools like the Maneva Orchestration Platform let you keep both: the freedom to set your own rules, and the certainty that nothing gets quietly buried. Measure the whole truth, keep every exception explicit, and the real opportunities will finally have somewhere to be found. Book a demo at maneva.ai to see honest OEE on your own line.
