OEE Is Not a Weekly Report: Making Overall Equipment Effectiveness Actionable
OEE multiplies availability, performance, and quality into one number. Calculated weekly it is a scorecard. Calculated live it is a control system. Here is the difference.
The short answer
Overall Equipment Effectiveness (OEE) is the product of availability, performance, and quality, expressed as a percentage of theoretical maximum output. A weekly OEE number describes what already happened, while OEE calculated continuously from machine and MES data lets a shift supervisor correct a running line before the shift is lost.
Key takeaways
- OEE = Availability x Performance x Quality. A component view is mandatory, because three very different plants can produce the same 65 percent.
- World class OEE is commonly cited at 85 percent. Most discrete manufacturers run between 45 and 65 percent and do not know their true baseline.
- The value is not the number. It is the loss attribution underneath it, which tells you which of the six big losses to attack first.
- OEE computed at the end of the week cannot change the week. Computed every few minutes, it changes the shift.
- The integration work is joining the historian or PLC data to MES production orders and quality results, so downtime has a reason code attached.
Nearly every plant we walk into reports OEE. Far fewer can tell you, without leaving the room, which of the six big losses cost them the most last month, or what their OEE was on line three at 2pm yesterday. The metric is present. The control loop is missing.
The formula, and why the components matter more
Availability = Run Time / Planned Production Time
Performance = (Ideal Cycle Time x Total Count) / Run Time
Quality = Good Count / Total Count
OEE = Availability x Performance x QualityThe headline number hides more than it reveals. Consider three lines that all report 65 percent OEE.
| Line | Availability | Performance | Quality | OEE | Actual problem |
|---|---|---|---|---|---|
| Line A | 68% | 97% | 98% | 65% | Breakdowns and changeovers. A maintenance and SMED problem. |
| Line B | 95% | 70% | 98% | 65% | Running slow. Minor stops, speed loss, or a mis-set ideal cycle time. |
| Line C | 94% | 95% | 73% | 65% | Scrap and rework. A quality and process control problem. |
Three different capital plans, three different teams, three different budgets. A single blended OEE number would send all three plants to the same unhelpful conversation about needing to do better.
The six big losses, mapped to the components
| Component | Loss | What it looks like on the floor |
|---|---|---|
| Availability | Equipment failure | Unplanned breakdown, line stopped, maintenance called |
| Availability | Setup and adjustment | Changeover time between product runs |
| Performance | Idling and minor stops | Jams, misfeeds, sensor faults under a few minutes each |
| Performance | Reduced speed | Line running below rated cycle time, often deliberately |
| Quality | Process defects | Scrap and rework during stable production |
| Quality | Startup rejects | Yield loss during warmup and changeover stabilization |
Why the weekly number cannot help you
A shift supervisor makes decisions on a horizon of minutes. Reallocate an operator, call maintenance, change the run order, adjust a setpoint. A number that arrives on Monday describing last week cannot influence any of those decisions. It can only be used to explain them afterward.
| Cadence | Who uses it | Decision it supports |
|---|---|---|
| Live, every 1 to 5 minutes | Shift supervisor, line lead | Intervene on the running shift |
| Daily | Plant manager | Reprioritize maintenance, adjust schedule |
| Monthly and quarterly | Operations director, CFO | Capital allocation, capacity planning, headcount |
All three are legitimate. Most plants have only the third and have quietly concluded that OEE is a reporting metric rather than an operating one.
What the integration actually involves
Real-time OEE is a data joining problem more than an analytics problem. Four sources have to line up on a common time axis and a common asset identifier.
- 1Machine state and counts from the PLC, SCADA, or a historian such as OSIsoft PI, Wonderware, or Ignition. This gives run time, stop events, and total count at high frequency.
- 2Production orders from the MES or ERP such as SAP S/4HANA, Plex, Epicor, or IQMS. This tells you what was supposed to be running, its ideal cycle time, and planned production time.
- 3Quality results from the quality system or inline inspection. This gives good count versus total count, and separates startup rejects from steady-state defects.
- 4Downtime reason codes, whether captured by operator entry on an HMI or inferred from machine state patterns. Without reason codes, OEE tells you that you lost time but not why.
A realistic sequence
- 1Start with one line, not the plant. A single well-instrumented line that people trust beats plant-wide coverage that nobody believes.
- 2Automate downtime capture before you refine the OEE math. Reason-coded stop events are worth more than a precise number with no attribution.
- 3Validate against the manual number the plant already produces, and explain every discrepancy. There will be discrepancies, and they are usually the manual method being wrong.
- 4Put a live display on the floor, not just in the office. OEE visible at the line changes operator behavior within days.
- 5Expand line by line, reusing the model. Line two should take a fraction of the effort line one took.
The plants that get real value from OEE are not the ones with the most sophisticated dashboards. They are the ones where a supervisor changed something at 10:15 because of what the screen said at 10:10.
Questions we get on this topic
How do you calculate OEE?
OEE is availability multiplied by performance multiplied by quality. Availability is run time divided by planned production time. Performance is ideal cycle time multiplied by total count, divided by run time. Quality is good count divided by total count. Each component should always be reported alongside the combined number.
What is a good OEE score?
85 percent is the figure commonly cited as world class for discrete manufacturing, with 60 percent often described as typical. In practice most plants land between 45 and 65 percent once measurement is automated, and the first accurate baseline is frequently lower than the manually reported number that preceded it.
What are the six big losses?
Equipment failure and setup or adjustment time reduce availability. Idling with minor stops and reduced running speed reduce performance. Process defects and startup rejects reduce quality. Mapping every lost minute to one of these six categories is what makes OEE actionable rather than merely descriptive.
Do we need an MES to track OEE in real time?
No. You need machine state and count data, which can come from a PLC, SCADA system, or historian, plus production order context which can come from an ERP if there is no MES. An MES makes the join easier because it already holds order context at the line level, but its absence is not a blocker.
Why is our automated OEE lower than our manual OEE?
Almost always because automated capture finds minor stops that operators never logged, and because it uses honest ideal cycle times rather than optimistic nameplate values. The lower number is the accurate one, and the gap between the two is usually the most valuable finding of the whole implementation.
Founder and CEO of VisualFlow Analytics. Former data analyst at Pratt & Whitney Canada, computer science and mathematics at McGill University. Leads technical delivery and client strategy across engineering, construction, and industrial data programs.