How is the count metric calculated in OEE?
The count metric in OEE is calculated by dividing the number of good units produced by the total units started during a given production run, then multiplying by 100 to express it as a percentage. This figure forms the Quality component of the OEE formula, which multiplies Availability, Performance, and Quality together to produce an overall equipment effectiveness score.
To break it down practically: if a machine starts 1,000 units in a shift and 940 pass quality inspection on the first attempt, the good count percentage is 94%. Units that require rework, are scrapped, or fail inspection are excluded from the good count, even if they are eventually corrected and shipped.
This distinction matters. OEE does not reward rework. A unit that fails the first pass and gets corrected later still counts as a defect in the Quality calculation. This is intentional. The metric is designed to surface the true productive output of a machine or line, not the output after human intervention has patched the result.
The full OEE formula works as follows:
- Availability: Actual run time divided by planned production time
- Performance: Actual output rate divided by the theoretical maximum output rate
- Quality: Good count divided by total count (units started)
- OEE score: Availability × Performance × Quality × 100
A machine running at 90% availability, 95% performance, and 94% quality delivers an OEE score of roughly 80.4%. Each component compounds, so a dip in good count pulls the final score down significantly.
What counts as a ‘good count’ versus a reject in OEE?
A good count unit is any unit that meets all defined quality specifications on the first production pass, with no rework, no secondary processing, and no inspection failure. A reject is any unit that falls outside those specifications at any point during the production run, regardless of whether it is later corrected or scrapped entirely.
The line between good count and reject is defined by your quality standard, not by what eventually ships to the asset owner. This is a common source of confusion. Manufacturers sometimes conflate shipped yield with OEE quality, but OEE is stricter. It measures right-first-time output.
Rejects in OEE typically fall into two categories:
- Startup rejects: Units produced during machine warm-up or after a changeover, before the process has stabilized. These are expected but should be tracked and minimized.
- Production rejects: Units that fail quality checks during a stable production run. These signal a process or equipment issue and carry more diagnostic weight.
Both categories reduce your good count percentage. Separating them in your data helps maintenance and process engineering teams identify whether the root cause is a machine calibration issue, a tooling problem, or an upstream input quality problem.
What is a good OEE count percentage by industry?
A good OEE quality (good count) percentage is generally above 99% in high-volume discrete manufacturing, while process industries and food production often target 95% to 98%. World-class OEE quality performance is typically cited at 99.9%, meaning fewer than one defect per thousand units started.
Industry benchmarks vary meaningfully based on product complexity, regulatory requirements, and the nature of the production process. Here is a practical reference by sector:
- Automotive and precision machining: 99%+ is the expectation. Defect rates directly drive warranty costs and customer contract penalties.
- Electronics and semiconductors: Quality targets are extremely tight, often above 99.5%, given the cost of components and the complexity of assemblies.
- Food and beverage: Targets typically range from 95% to 98%, with startup rejects during line changeovers accounting for a meaningful share of losses.
- Industrial machinery and OEM manufacturing: 96% to 99% is a realistic target range, depending on component complexity and batch size.
- Oil, gas, and process manufacturing: Quality losses are often expressed differently (yield, not unit count), but equivalent good count targets sit around 97% to 99%.
If your good count percentage sits below 95% consistently, that is a signal worth investigating at the equipment level, not just the process level. Worn tooling, miscalibrated sensors, and inadequate preventive maintenance (PM) cycles are frequent contributors to chronic quality losses in asset-heavy manufacturing environments.
Why does a low good count hurt more than a low availability score?
A low good count hurts more than a low availability score because quality losses are invisible until they reach the asset owner, while downtime is immediately visible on the production floor. Defects consume machine time, raw materials, and labor without producing sellable output, meaning the cost of a quality loss is higher than the cost of equivalent idle time.
Consider the compounding effect. When a machine is down, everyone on the floor knows it. Maintenance is called, the work order is opened, and the clock is running visibly. When a machine is running but producing rejects, the line may look productive for hours before the quality check surfaces the problem. By then, a significant volume of material has been consumed, and the defective units have to be reworked or scrapped.
There is also a downstream cost that availability losses do not carry in the same way. A second technician visit for a service issue on a complex asset eats the margin on the service contract. The same logic applies to production quality: a rework cycle consumes labor that was not budgeted, and scrap consumes raw material that cannot be recovered. Neither shows up in the availability score.
From an OEE perspective, the math reinforces this. Because OEE multiplies all three components together, a 5-percentage-point drop in quality has the same mathematical impact as a 5-point drop in availability. But the operational cost of the quality drop is often higher because it is harder to detect, harder to attribute, and slower to correct.
How can field service teams improve good count on production assets?
Field service teams improve good count by ensuring that production assets are maintained to specification, calibrated correctly, and serviced before degradation affects output quality. Preventive maintenance (PM) discipline, accurate asset history, and first-time fix rates on service work orders are the three levers that matter most.
When a machine drifts out of calibration or a wear component degrades gradually, quality losses accumulate before the machine triggers a downtime event. The asset appears available and running at speed, but the good count percentage is quietly eroding. Field technicians who have access to full asset history, including previous PM records and known failure patterns, can catch these drift conditions before they become quality problems.
Specific actions that drive good count improvement include:
- Completing PM checklists fully and on schedule, rather than deferring tasks when the asset appears to be running normally
- Recording calibration data and tolerance readings at each service visit so trends are visible over time
- Resolving work orders on the first visit by arriving with the right parts and documentation, reducing the window during which a partially serviced asset continues producing rejects
- Flagging quality-related fault codes in the asset record so engineering teams can correlate maintenance history with production quality data
The connection between first-time fix rate and good count is direct. Every return visit for the same issue extends the period during which the asset is either down or running below quality specification. Improving first-time fix rates reduces that window and protects the good count percentage.
How Gomocha Helps Improve OEE Good Count
Unplanned equipment failure and chronic quality losses share a common root cause: field service teams that lack the tools, data, and process structure to maintain assets to specification, consistently, at scale. That is the problem we built Gomocha to solve.
Our field service platform gives manufacturing service teams the capabilities that directly protect good count performance:
- Offline-capable mobile app: Technicians access full asset history, PM checklists, calibration records, and safety documentation on the plant floor, even without connectivity. Teams using offline access to complete asset documentation have seen a 19% improvement in first-time fix rates.
- No-code Workflow Designer: Operations teams configure PM checklists and quality inspection forms by asset type without waiting on IT. When a process changes or a new asset class is added, the workflow adapts in days, not months.
- Guaranteed ERP integration: Native integrations with AFAS and Microsoft Dynamics, and connectors for SAP and JDE, mean work order data, asset history, and parts inventory stay synchronized across your systems without manual reconciliation.
- Purpose-built for asset-heavy industrial operations: Across 13 manufacturing customers and more than 177,000 work orders, teams using Gomocha have reduced unplanned equipment downtime by up to 41%.
If quality losses on your production assets are costing more than your OEE dashboard is showing, an Efficiency Assessment is the right starting point. We will map your current field service workflows against your OEE targets and identify where maintenance execution gaps are hiding. You can also use our Efficiency Calculator to quantify what those gaps are costing you today.
Explore how we support industrial manufacturing operations specifically, or learn more about us and the platform behind the results.
Frequently Asked Questions
Can a machine have a high OEE score but still have a poor good count percentage?
Yes, and this is one of the most misleading scenarios in OEE reporting. A machine can post strong Availability and Performance numbers — meaning it runs often and runs fast — while quietly producing a high volume of rejects. Because OEE multiplies all three components together, a 99% availability and 98% performance score can still result in a mediocre overall OEE if quality sits at 85%. Always review each OEE component independently before drawing conclusions about overall equipment health.
How should reworked units be handled when reporting good count in OEE?
Reworked units should never be counted as good units in your OEE quality calculation, even if they are corrected and eventually shipped to the customer. OEE measures right-first-time output exclusively, so any unit that required secondary processing or failed an initial inspection is classified as a defect for reporting purposes. If your operation tracks rework separately, that data is valuable for understanding true cost of quality, but it belongs in a rework analysis report — not in the OEE good count figure.
What is the most common mistake manufacturers make when trying to improve their good count percentage?
The most common mistake is treating quality losses as a process engineering problem while overlooking the maintenance execution root cause. Worn tooling, drifting calibration, and skipped or deferred PM tasks are responsible for a significant share of chronic good count losses — but they rarely trigger a visible downtime event, so they go unattributed. Before redesigning a process, audit your PM compliance rates, calibration records, and first-time fix history on the assets with the lowest quality scores.
How often should good count data be reviewed to catch quality trends early?
Good count data should be reviewed at the shift level at minimum, and in real time where monitoring infrastructure supports it. Reviewing quality data only at the end of a production run or on a weekly basis allows defect-generating conditions to persist for far longer than necessary, compounding material waste and rework costs. Shift-level visibility allows maintenance and operations teams to correlate quality dips with specific events — such as a recent changeover, a tool replacement, or a partial PM — and respond before the problem scales.
Is there a meaningful difference between tracking good count at the machine level versus the line level?
Yes, and the difference matters significantly for root cause analysis. Line-level good count data tells you that a quality problem exists somewhere in the production flow, but machine-level data tells you exactly where it originates. In multi-step production lines, a defect introduced at station two may not be detected until station five, making line-level reporting misleading about the source. Tracking good count at the individual asset level — and correlating it with maintenance history for each machine — gives your service and engineering teams the precision they need to act on the right asset.
How do startup rejects factor into OEE benchmarking, and should they be reported separately?
Startup rejects are included in the OEE quality calculation and do reduce your good count percentage, but reporting them separately from production rejects is a best practice that most OEE frameworks recommend. Separating the two categories allows you to distinguish between expected process stabilization losses — which can be reduced through better changeover procedures — and mid-run quality failures, which typically signal an equipment or input material issue. Many manufacturers track startup reject rates as a standalone KPI alongside OEE to benchmark changeover efficiency independently.
What is a realistic timeline for seeing good count improvements after tightening PM and field service processes?
Most asset-heavy manufacturing operations begin to see measurable good count improvement within one to three PM cycles after implementing tighter maintenance execution processes, which typically means two to six months depending on PM frequency. The initial gains tend to come from catching and correcting calibration drift and wear conditions that had been accumulating undetected. Sustained improvement — reaching and holding a target good count percentage — generally requires three to six months of consistent data collection, process adjustment, and technician adherence to updated PM and calibration protocols.