Key takeaways
See the OEE calculation this hidden capacity uses.
OEE is a useful metric, but its math has structural blind spots. Each component, availability, performance, quality, is computed against definitions that exclude certain categories of loss by design. The losses are real; they just do not show up in the number. The result is that two plants reporting very different OEE numbers can have similar actual capacity utilisation, the difference is partly what each plant counts.
The four hiding places below are not exotic. They are the places losses go when the OEE measurement system is set up reasonably but not aggressively. The article on manufacturing KPIs covers the underlying components.
Most OEE systems count a stop only if it exceeds some duration, typically 60 seconds, sometimes 5 minutes. Stops shorter than that are invisible. On a packaging line that hiccups 8 times an hour for 20 seconds each, that is 160 seconds per hour of unreported loss, roughly 4% of the shift, gone without trace.
The fix is a 30-day diagnostic where the threshold is dropped to 5 seconds for one line. The data will look uncomfortable; the team will see hundreds of stops per shift that nobody knew about. That ugliness is the point, the unrecorded micro-stops were always happening. Now they are visible.
Most plants find that micro-stop accumulation alone is 6-12% of total available time on lines that have not done this diagnostic. The piece on production loss analysis covers how to attribute the recovered micro-stops to specific assets and causes.
Performance in OEE is calculated as "actual / ideal" cycle time. If the ideal is set generously, at the demonstrated capability rather than the design intent, the performance number stays comfortably high while real speed loss accumulates underneath.
A line that was designed for 60 units per minute and consistently runs 55 has a performance of 91.7% against the design ideal. The same line evaluated against a generous "demonstrated" ideal of 55 has a performance of 100%. Both numbers are mathematically correct; only the first reveals the 5-unit-per-minute capacity that is hiding.
The fix is an annual re-baseline of ideal cycle times against the original design intent, adjusted only for hard physical constraints (raw material limits, quality requirements). Marginal-product losses that have accumulated over years get reset rather than baked in. The article on work order management systems covers how the re-baseline triggers maintenance work to recover the gap.
Planned downtime. PMs, changeovers, breaks, is excluded from the OEE clock. That is correct in principle. The problem is that the "planned" budget tends to grow quietly. A 30-minute scheduled changeover becomes 45 minutes that gets included in planned time. A 90-minute PM window becomes 120 minutes. The plant's available time shrinks; the OEE number stays the same because the denominator shrank with it.
The fix is a quarterly audit of planned-downtime entries against their budgeted durations. Items consistently overshooting their budget are either real (the budget needs updating) or operational drift (the team is taking longer than they should). Either way, surfacing the gap is the first step.
Most plants find a meaningful share of planned-downtime entries are running over budget after the audit. Some of those are real; some are drift. The drift cases are usually the easier fix. The piece on the preventive maintenance schedule covers how to keep PM windows honest.
Quality in OEE is "good units / total units." Most plants count first-pass good. Some count final good after rework. The two numbers can differ by 5-10% on lines with significant rework volume.
The hiding happens when rework gets done off-line, doesn't generate a quality work order, and the reworked units flow back into the "good" count without being flagged. The OEE quality number stays high; the rework capacity (labour, time, material) is consumed invisibly.
The fix is to require a quality close-out on every reworked unit and to count rework as a quality loss in the daily OEE while continuing to count rework yield in the monthly view. Two numbers, two windows, both honest. The article on root cause analysis covers how rework patterns surface upstream quality problems.
A plant reporting 78% OEE that runs all four diagnostics often finds the honest number is materially lower. The honest number is the bad news. The good news is the gap between honest and reported is the hidden capacity, the throughput the plant could recover without buying anything.
Net hidden capacity in a typical mid-market plant: often a double-digit percentage of available time, recoverable over 6-18 months of focused work. None of it requires new equipment.
Run one diagnostic per quarter. Trying all four at once is too much team bandwidth and the data collection collides. The recommended sequence:
By year-end the plant has a honest OEE number and a concrete recovery roadmap for the next year. The piece on the work order management system covers how the recovered capacity gets allocated.
The diagnostics work in any OEE system, but the threshold flexibility matters.
Where a unified OEE + CMMS platform helps is that the threshold can be temporarily dropped on one line without breaking the comparability of the rest of the plant's numbers, and the recovered micro-stop data can be linked to work orders without manual reconciliation. Fabrico is built for that workflow.
To see what your honest OEE number would look like, book a demo .
The reported OEE drops, sometimes by 10+ percentage points. Plants that do this without preparing leadership often face uncomfortable conversations about "we got worse" when the operational reality is the same. The right sequence is: run the diagnostic, present the honest baseline, then permanently move to the lower threshold.
OEM ideals are usually the highest reasonable benchmark. If the team is comfortable performing below that for legitimate reasons (product mix, regulatory), document those reasons. If the gap is just operational drift, the ideal should not be lowered to accommodate it.
Yes, explicitly. Surprise diagnostics that drop the reported OEE without context create defensiveness. Framed as "we are looking for hidden capacity, not blaming anyone" usually lands well with mature teams.
The diagnostics produce visibility in months; the operational changes to capture the capacity usually take 6-18 months. The fastest payback is the micro-stop fixes, because they often link to specific failure modes that are addressable with PM tightening.
Reporting the recovered capacity as a headline OEE improvement when it is actually just better measurement. The recovered capacity is real, but the OEE number changes for two reasons, the calculation got more honest and the operation got better. Separating those two narratives matters for board-level reporting.
Login
Blog
