Quality control
How to reduce rejection rates in manufacturing: inspect at the operation, not at the end
Most factories find quality failures at the wrong stage. Inspection at each operation catches the defect where it starts, stops it compounding, and tells you which operation to fix.
· 7 min read · Updated
In most factories, quality control happens at the end. A unit completes every operation, reaches the final inspection bay, and only then is a failure found. By that point the defective unit has consumed material, machine time, operator hours and packing. All of it is wasted.
Why rejection is found too late
End-of-line inspection exists because it is convenient. One checkpoint, one team, one register. But it creates a blind spot: the operation where the problem started is invisible. Was the defect made at welding, at assembly, at finishing? Nobody knows, and without that, the root cause cannot be fixed.
- Defects compound across operations. A dimensional error at the first operation becomes a fitment failure at assembly.
- Rework is heavier, because the unit has already been fully assembled.
- Scrap cost is at its maximum at the end of the line. All the value added is lost.
- The root cause is hours old by the time the defect is found.
- Operators get no feedback. They do not know their output is failing downstream.
What end-of-line inspection costs
Take a sub-assembly line running 200 units a shift. At a 5 percent final rejection rate, that is 10 units a shift going to rework or scrap. If each unit carries ₹2,000 of cost by the time it reaches final inspection, that is ₹20,000 a shift, before the rework labour and the line capacity spent on it.
It also creates a habit. If every line expects some rejection at the final bay, rejection becomes a budget line rather than a problem to solve.
Inspection at every operation that can make the defect
In-process inspection means a plan sits on the operation. Dimensional checks at machining, visual at plating, torque at assembly, function at test. QC records the measured values before the batch moves to the next operation.
When a check fails, the batch is held. The supervisor sees it. The batch is sent to rework at the same operation, before it compounds, or rejected. It does not move forward and create failures downstream.
- Operation-specific plans: each operation has its own characteristics, not a generic form.
- A name on every record: the operator who ran the operation and the QC person who checked it.
- Pass or fail with a reason: failure modes from a short list, for trend analysis.
- Rework as a work order: a failed check creates a rework operation with the reason on it.
- History on the batch: every batch carries its inspections, dispositions and rework.
Rework and scrap, without losing the trace
Not every failure is scrap. Many defects caught at the second operation are corrected in fifteen minutes and rejoin the flow. The system has to support that without re-entry and without losing the record.
In FleekERP, a failed inspection at any operation creates a rework operation on the same work order, with the failure reason on it. Once rework is done, the batch is re-inspected before it moves. The batch record shows the original inspection, the rework, and the re-inspection. That is the trail an OEM auditor asks for.
Measure rejection by operation, not overall
The biggest change inspection at the operation brings is in how quality is measured. Instead of one plant-wide rejection rate, which says nothing about where to fix, you get a rate per operation. Assembly has 4.1 percent. Testing has 0.3 percent. Assembly is the problem operation. Go and look.
The same data splits by shift and by operator. If assembly rejection rises every night shift, that is a supervision or training problem. If it follows one operator, that is a coaching conversation. End-of-line data can never point that precisely.
A rejection map after thirty days
With a month of inspection data by operation, you can lay operations against failure modes and see where the counts concentrate. Those cells are where the corrective action goes. Most factories never have this, not because they do not want it, but because end-of-line inspection does not produce the data.
The factories that bring rejection down are not the ones with the best final inspectors. They are the ones that catch the failure at the operation that made it, and stop it compounding.
Questions this article answers
What is in-process inspection?
An inspection plan that sits on the operation instead of at the end of the line: dimensional checks at machining, visual at plating, torque at assembly, function at test. QC records the measured values before the batch moves on, and a failed check holds the batch until it is reworked or rejected.
Why does end-of-line inspection make rejection so expensive?
Because by the final bay the defective unit has used all its material, machine time and labour. Defects compound across operations, rework is heavier on an assembled unit, and the root cause is hours old. On a line running 200 units a shift, 5 percent rejection at ₹2,000 a unit is ₹20,000 a shift before rework labour.
How should a factory measure its rejection rate?
By operation, not as one plant-wide figure. A rate per operation shows where to fix, for example assembly at 4.1 percent against testing at 0.3. Split the same data by shift and by operator, and after thirty days lay operations against failure modes to see where rejections concentrate.
How is rework recorded without losing traceability?
In FleekERP a failed inspection creates a rework operation on the same work order, with the failure reason on it. The batch is re-inspected before it moves, and its record shows the original inspection, the rework and the re-inspection: the trail an OEM auditor asks for.