Real OEE vs declared OEE: why your Excel doesn’t tell the truth

OEE: a number everyone watches, but few ever question

In most plants, OEE has become a ritual. It’s calculated every week, dressed up in a dashboard, discussed in the production meeting and compared month over month. Everything about the way it’s presented inspires confidence: clean percentages, trend lines, sometimes a target for each line. And yet, behind that air of rigour, there’s a fragility few teams dare to look at squarely – the quality of the data feeding the calculation. (OEE, or Overall Equipment Effectiveness, is the English name for what French manufacturers call TRS.)

Because the calculation itself is never the problem. The formula is universal and well known: OEE = Availability x Performance x Quality. What changes everything is not the equation but what you pour into it. And when that raw material comes from manual readings – shift sheets filled in by hand and consolidated the next day in a spreadsheet – the result inherits every limitation of human data entry. What you end up with is a declared OEE: a smoothed version of reality that is almost always more flattering than what actually happens on the line.

Real OEE vs declared OEE: the real difference

Real OEE is the performance a line actually delivers, measured continuously and to the second by a system that depends on no one to record what happens. Declared OEE, by contrast, is a number rebuilt after the fact from whatever operators had time to observe and jot down while they were running production.

This distinction is not a methodological footnote. It conditions the reliability of every decision that flows from the number: where to launch an improvement project, which line to prioritise, how to compare two workshops. Declared OEE and real OEE can diverge by tens of points on the very same line – and always in the same direction. Declared overstates; real brings you back to the ground truth.

That divergence is structural, not accidental. It doesn’t come from a one-off error you could fix by training teams better or buying a smarter spreadsheet. It comes from the very nature of manual logging, which simply cannot see certain losses.

Why declared OEE lies, without the slightest bad faith

A manual reading isn’t dishonest – it’s physically limited. An operator can only record what they see and what they materially have time to note, all while running the line, handling the unexpected and protecting quality. Long, visible stoppages get logged, because they impose themselves. Anything too short, too frequent or too discreet, on the other hand, inevitably slips under the radar.

Declared OEE therefore adds up the losses you saw and forgets the ones you didn’t. The resulting number is, by construction, more optimistic than reality. And it suffers from a second, equally awkward flaw: it arrives a day late. By the time the report is consolidated, the shift in question is over, the team has rotated, and it’s too late to act on what actually happened.

Micro-stops: the first invisible loss

The family of losses most systematically underestimated is micro-stops. A cell that clears itself, a jam fixed in a few seconds, a sensor that hesitates, a part nudged back into place by an operator who barely notices doing it. Taken on their own, none of these stoppages feels worth writing down.

But added up across a shift, they tell a very different story. On high-speed lines, micro-stops frequently account for several points of OEE – sometimes more than breakdowns themselves. The trouble is that they appear in no logbook: too short to be written down, too frequent to be counted, too mundane to be remembered at the end of the shift.

Speed losses after a changeover

The second invisible loss is speed loss. After every changeover, a line doesn’t instantly return to its nominal rate. There’s a ramp-up phase during which the machine is running but below target speed. This grey zone triggers no stoppage: the line is producing, it isn’t stopped, so nothing gets recorded.

Yet those minutes spent under rate are a permanent leak of capacity. The shorter the runs and the more frequent the changeovers, the more this loss weighs. In high-mix workshops, it can on its own explain the entire gap between the OEE you display and the OEE you actually achieve.

Unrecorded incidents and the hidden cost of ‘close enough’

The third family gathers all the small incidents that never surface: the adjustment quietly redone, the two-minute wait for material, the deliberate slowdown to avoid a quality defect, the drifting parameter corrected on the fly. These are perfectly normal realities of shop-floor life – and completely invisible to manual logging.

Stacked end to end, these three families of losses explain why declared OEE drifts so far from real OEE. And until you measure them to the second, you can’t recover them. You keep optimising against a blurry picture, launching projects against assumed causes, and waiting for performance points that never come.

How to measure real OEE, without an MES project

Closing this gap means giving up rebuilding OEE the next morning and measuring it at the source instead, continuously, while the line runs. That’s exactly the principle behind TeepTrak: a sensor fitted to the machine returns real OEE in real time – no MES project, no production stop, installed in under an hour, with usable data within 48 hours.

From there, the conversation changes completely. You no longer debate the reliability of the number, because it’s no longer arguable. You look at where the minutes actually go, machine by machine, and you act on them – in the right place, at the right time. The data stops being a meeting topic and becomes a decision tool.

What it changes in practice: the Hutchinson case

The payoff isn’t theoretical. Hutchinson improved its OEE from 42% to 75% with the same headcount and machines, sensor installed in under an hour. The gain didn’t come from a new machine or from extra hires – it came from visibility. Once the invisible losses were made visible, the teams could act on the real causes instead of on guesswork.

This pattern repeats from site to site. More than 450 plants across 30+ countries already monitor their OEE to the second with TeepTrak. When declared OEE and real OEE finally converge, three things unlock: performance reviews stop being about whether the number is trustworthy and start being about action; improvement projects target the real losses; and comparing lines and sites becomes legitimate at last, because everyone measures the same way.

Key takeaways

Declared OEE almost always overstates real OEE – by construction, not by bad faith. The gap comes from three losses that are invisible to manual logging: micro-stops, post-changeover speed losses and unrecorded incidents. Only real-time measurement at the source closes that gap.

And the gain is concrete. Hutchinson improved its OEE from 42% to 75% with the same headcount and machines, sensor installed in under an hour. The number didn’t improve because the line got faster; it improved because the team could finally see what the readings never showed.

FAQ

What’s the difference between real OEE and declared OEE?
Real OEE is measured automatically and continuously on the machine, to the second. Declared OEE is rebuilt after the fact from manual readings and stays almost always more optimistic, because it doesn’t capture short losses.

Why is my OEE calculated in Excel wrong?
The calculation isn’t the issue – the input data is partial. Manual readings don’t capture micro-stops or speed losses, which mechanically inflates the result and gives an OEE more flattering than reality.

How far can declared OEE drift from real OEE?
It depends on the process, but on high-speed lines with frequent changeovers the gap is frequently measured in tens of points.

How do I measure my real OEE?
With automatic measurement at the source that reads availability, rate and quality continuously. A plug-and-play layer like TeepTrak installs in under an hour, with no MES project, and delivers usable data within 48 hours.

Do I need an MES to get reliable OEE?
No. Measuring OEE is a standalone need that can be met by a dedicated layer fitted to the machine, independently of any MES you may or may not have.