AboshOEE can rise without throughput rising for seven reasons: you improved a non-constraint, the constraint moved, the extra output became inventory, the OEE definition changed, the product mix shifted, quality problems moved downstream, or demand never needed the recovered capacity. Check the constraint first, then the metric definition, then downstream flow. OEE measures loss at a resource. Throughput is decided by the system.
Seven causes with a check for each
A simple example where OEE rises and output does not
The metrics to pair with OEE so it cannot mislead
Record matched windows, product mix, definitions, inventory and shipments.
CSVFictional filler and capper comparisonReproduce the 65% to 72% change and the unchanged 100 shipped units. Static CSV values.
An OEE increase is not automatically a business improvement
A higher OEE can reflect more good production at one machine, a different product mix or a changed calculation. First establish which of those happened. Then follow the units through the rest of the line.
The diagram shows a filler with spare capacity ahead of a capper that limits output. The worked record below checks that explanation with a simple inventory balance.
Reconcile the units before choosing another improvement
Fictional example: compare two one-hour windows on the same product. The filler’s ideal rate is 200 units/hour. Its planned time, cycle standard and stop threshold stay fixed. Good filler output rises from 130 to 144 units, so its OEE rises from 65% to 72%. Both figures use good units at that machine, not shipments.
The capper processes 100 units in each hour and remains the limiting step. A packer capable of 120 units/hour does not add another throughput constraint in this simplified example; it receives only the capper’s 100 units. There are no rejects, rework loops or finished-goods stock changes. Demand is 110 units per hour; shipments stay at 100.
The queue before the capper starts at 20 units in each comparison window. It ends at 50 in the first and 64 in the second: opening queue + filler good output - capper output. The extra 14 filler units become 14 additional queued units. These are matched teaching windows, not consecutive shifts or observed plant results.
Vorne’s FAQ separates resource OEE from shipment counts and recommends measuring losses at the constraint. Check that recommendation against your current product mix and flow. In a real investigation, also record downstream rejects and opening/closing finished goods. Shipments alone cannot tell you how much the line made during that hour.
| Measure | Window A | Window B |
|---|---|---|
| Product / planned window | Same SKU / 60 min | Same SKU / 60 min |
| Filler ideal rate | 200 units/h | 200 units/h |
| Good filler units / filler OEE | 130 / 65% | 144 / 72% |
| Capper output / limiting rate | 100 / 100 units/h | 100 / 100 units/h |
| Opening queue before capper | 20 | 20 |
| Closing queue before capper | 20 + 130 - 100 = 50 | 20 + 144 - 100 = 64 |
| Downstream rejects / finished goods change | 0 / 0 | 0 / 0 |
| Demand / shipped units | 110 / 100 | 110 / 100 |
Seven causes, and how to check each
Work through the list in order. The first three are about where the improvement landed, the next three about how the numbers were produced, and the last about whether the capacity was needed.
| # | Cause | How to check |
|---|---|---|
| 1 | You improved a non-constraint | Compare the resource’s capacity with the constraint’s. Recovering 100 minutes at a resource with spare capacity creates local capacity only. |
| 2 | The constraint moved | Recheck the bottleneck after every improvement. See dynamic bottlenecks. |
| 3 | Output became inventory | Look for WIP growing in front of the next operation. |
| 4 | The OEE boundary changed | Audit planned time, ideal cycle time, stop coding and product segmentation before and after. |
| 5 | Mix changed | Recalculate OEE per product. Higher-OEE products may dominate the period while total output value stays flat. |
| 6 | Quality moved downstream | Check downstream rework, inspection and customer returns for the same period. |
| 7 | Demand did not need the capacity | Compare recovered capacity with demand at the constraint. |
Check cause 4 for a changed reporting boundary
Teams can improve a score without improving the process by excluding planned time, loosening the ideal cycle time, recoding stops or segmenting products differently. Retain the old and new definitions so the team can distinguish a reporting change from a change in production.
Audit the metric definition before celebrating. The OEE formula guide lists the definitions to lock down.
When the improvement is still worth having
If the plant already has enough capacity, the economic benefit of a higher OEE may be fewer shifts, resilience, more maintenance time or shorter lead time, not more throughput. Those are valid reasons, but they should be stated, because otherwise a team reports a throughput gain that never arrived.
Use the theory of constraints to decide where recovered time turns into shipments and where it does not.
What to measure alongside OEE
Pair OEE with measures that show whether the system moved. A resource-level score should never be read without them.
- Constraint throughput
Good units per hour at the resource that limits output.
- Lead time and WIP
Are queues shrinking or growing?
- On-time in-full
Did the customer see a change?
- First-pass yield
Is quality holding across the process?
- Demand and schedule attainment
Was the recovered capacity needed and used?
A related example from Abolfazl
Abolfazl connects release decisions, queues and launch lead time.
In English, on founder Abolfazl’s personal blog. This is related reading, not an independent technical source.
Practical checklist
- Identify the constraint before choosing an OEE project.
- Recheck the constraint after every significant improvement.
- Watch WIP in front of the next operation.
- Freeze the OEE definition before measuring improvement.
- Recalculate per product when the mix changes.
- Check downstream rework and returns for the same period.
- Compare recovered capacity with demand.
- Report constraint throughput, lead time and on-time delivery beside OEE.
FAQ
FAQ
Sources and further reading
Authoritative references used to research and verify this guide.