What OEE is
OEE means Overall Equipment Effectiveness. It combines availability, performance, and quality into one measure of how effectively a machine produced during planned production time. The standard formula is OEE = availability × performance × quality.
Inputs you can write down by hand
For a manual estimate, record the same machine and the same shift throughout. Keep the notes close to the time they describe. Recording stops and counts as they happen avoids relying on memory at the end of the shift.
- Planned time: How long the machine was scheduled to produce. Exclude time when production was not planned, such as a scheduled break, if that is how your shop defines planned production time.
- Run time: How long the machine was actually producing. Start with planned time and subtract recorded stops that count as downtime under your shop's rules.
- Total finished count: How many parts the machine finished during the period. Count parts consistently; do not mix completed parts with work in progress.
- Daily target: The number of parts expected from that station for the day. This shows whether output is on track, at risk, or behind. It is useful context, but it is not a substitute for OEE inputs.
- Ideal cycle time: The shortest standard production time for one part on that machine. You need this to calculate performance. Use a value your team has defined for that part and machine.
- Good count: How many finished parts passed your quality check the first time, without rework. Exclude rejected parts and parts requiring rework, using the inspection method your shop already uses.
Write down what counts as a stop and what counts as a good part before comparing shifts. If those rules change, the numbers stop being comparable. For make-to-order work, note which part or job was running. When the product changes, check whether its ideal cycle time and target differ.
Availability and performance from manual numbers
Availability compares actual run time with planned production time:
Availability = run time ÷ planned time
If the machine was planned to run for the shift but stopped for a recorded period, subtract that downtime to estimate run time. Then divide run time by planned time. Keep planned time and run time in the same units, such as minutes.
Performance compares actual output with the output expected at the ideal cycle time:
Performance = (total count × ideal cycle time) ÷ run time
Use matching time units for ideal cycle time and run time. For example, if the cycle time is written in seconds, convert run time to seconds before calculating. If performance exceeds 100%, check the ideal cycle time, count, and recorded run time before accepting the result.
These figures depend on the notes you enter. A missed stop changes run time and availability. A wrong ideal cycle time changes performance. Treat the result as an estimate when records are incomplete or cycle times vary between jobs.
| Input or result | Manual OEE | FactoryVision camera |
|---|---|---|
| Run time and stops | Recorded by a person; missing stops affect the result. | FactoryVision tracks downtime with a camera on each machine. |
| Finished count | Counted and entered by a person. | FactoryVision uses one camera on each machine to count finished parts. |
| Daily target context | Enter the target separately and compare it with the manual count. | FactoryVision compares each station's count with its daily target and projects the end-of-shift count. |
| Quality | Enter a first-pass good count from a separate inspection. | A camera cannot see defects inside material, so quality still needs suitable inspection data. |
| OEE result | Calculated from the values you recorded. | Still depends on valid planned time, cycle time, and quality data; a camera count alone does not provide every OEE input. |
FactoryVision counts finished parts and tracks downtime for small make-to-order factories, using one camera on each machine. It runs in a Turkish door and furniture factory. It fits discrete parts you can count, such as doors, furniture, and metal parts. Its best fit is factories with 10 to 100 machines. Continuous processes such as extrusion are out of scope. FactoryVision is not an ERP, MES or SCADA system. It does not publish accuracy numbers until they are measured on fixed cameras.
Quality input is manual: the camera cannot see defects inside material
Quality is the share of total output that is good the first time, without rework:
Quality = good count ÷ total count
For this calculation, "good" needs a consistent definition from your inspection process. Record the number of finished parts that passed that process the first time, then divide by the total finished count. If you do not have a good count, mark quality as unknown rather than treating every part as good.
A camera that counts finished parts cannot determine whether a defect is hidden inside the material. A camera also cannot measure finer than about 0.1 mm with sensible optics. These are scope limits of camera counting, not measured accuracy figures for a particular machine or installation.
Use inspection records, operator checks, or another method your shop trusts to enter good count. If inspection happens later, record when the count was checked and which shift or job it belongs to. This keeps the quality count tied to the production period being assessed.
Worked example for one machine, one shift
Illustrative, not measured. Suppose a machine had 8 hours of planned production time. The operator recorded 1 hour of downtime, leaving 7 hours of run time. The machine finished 360 parts. The ideal cycle time was 1 minute per part. The daily target was 400 parts. Inspection found 342 first-pass good parts.
Availability is 7 hours divided by 8 planned hours, or 87.5%.
Performance is 360 parts × 1 minute per part, divided by 420 minutes of run time. That gives about 85.7%.
Quality is 342 good parts divided by 360 total parts, or 95%.
Using the unrounded ratios, OEE is (7 ÷ 8) × (360 ÷ 420) × (342 ÷ 360) = 71.25%, or about 71.3%. This result comes only from the hypothetical inputs above. It is not a FactoryVision measurement, customer result, or benchmark.
The daily target comparison is separate: 360 finished parts against a target of 400 puts the count 40 parts below target at the time represented by these example figures. That comparison says where output stands against the target. OEE describes availability, performance, and quality together. One does not replace the other.
Limits and the next step
A hand-filled OEE estimate is only as useful as its definitions and records. When a shift includes product changes, setup time, short stops, rework, or jobs with different cycle times, record those differences. Treating unlike parts or time periods as interchangeable makes the final percentage harder to interpret.
Use a consistent rule for planned time, downtime, ideal cycle time, total count, and good count. Keep the result tied to one machine, one shift, and the job being produced. If an input is missing, show it as missing. Do not fill gaps with guesses and present the output as measured OEE.
FactoryVision focuses on visible finished-part counts by machine. It compares each station's count with its daily target, projects end-of-shift output, and its station light shows whether the station is on track, at risk, or behind. This gives teams a view of output progress. The camera cannot see defects inside material, and a camera count alone does not supply every value in the standard OEE formula.
The daily target calculator is not built yet. Use a target your team has set and keep it distinct from the OEE calculation.
Buying starts with an introductory call. After the call, the FactoryVision team gets to know your company so the setup fits your machines and processes. Book the introductory call at https://calendly.com/esat-factoryvision/intro.
Frequently asked questions
Which inputs do I need without any sensors?
Record planned time, run time, total finished count, ideal cycle time, and good count. Keep the daily target separately for comparing output with the day's goal.
Can I get a usable OEE number from finished counts alone?
No. Finished count alone does not give planned time, run time, ideal cycle time, or good count. You can compare it with a daily target, but that comparison is not OEE.
How do I fill quality when the camera cannot see defects?
Use a first-pass good count from your existing inspection process. Divide good count by total finished count. Without inspection data, leave quality unknown instead of assuming every part passed.
How does the result connect to a daily target and the station light?
OEE combines availability, performance, and quality. A daily target is a separate output goal. The FactoryVision station light shows whether a station is on track, at risk, or behind against its daily target.
When is an OEE number misleading for a small shop?
It misleads when shifts use different definitions, when downtime goes unrecorded, when ideal cycle times do not match the job, or when quality counts are missing. If a make-to-order shift runs different parts, label the jobs and review the inputs before comparing results.
Does a camera count tell me whether each part is good?
No. A camera cannot see defects inside material. Use a separate inspection method for good count.
Does FactoryVision publish camera accuracy numbers?
FactoryVision does not publish accuracy numbers until they are measured on fixed cameras. No accuracy data appears on this page.
Is the daily target calculator available now?
No. The daily target calculator is not built yet.
Book an introductory call
Buying starts with an introductory call. After the call, the FactoryVision team gets to know your company so the setup fits your machines and processes. Book the introductory call at https://calendly.com/esat-factoryvision/intro.