Why rejection costs twice
Rejection is the most under-managed cost in most discrete plants because it hides. A part that fails at operation 30 has already absorbed the material, the machine time and the labour of operations 10 and 20. Scrap it and you lose all of that. Replace it and you buy the material again. Rework it and you spend fresh operations, re-inspection and schedule slack recovering it. Every one of those paths costs real money, and none of them shows up as a line item called "rejection."
That is why a plant that records only a single end-of-line reject total cannot improve. The total tells you that you have a problem; it tells you nothing about where, why, or which fix would move the number most. Reducing rejection is therefore first a data problem and only then a process problem — and the data has to be captured at the moment and place the reject occurs.
Step 1 — Capture rejection at source
The foundation is booking good and reject separately at each operation as the work order runs its route, rather than counting only what fails final inspection. Every operation records three buckets: good WIP that moves forward, reject WIP that failed there, and process scrap removed from the route. Because the reject is booked at the operation, it carries the identity of the step and the work center that produced it.
On top of operation-level WIP, line rejection captures parts rejected on the production line as distinct from incoming-material rejection — at the level of the manufactured part, at child-part level, and booked against the specific work order. That child-part granularity matters in assembly: a reject caused by one component is not the same problem as a reject caused by the assembly operation, and lumping them together hides the real cause. See FG, Rejection & Rework.
Step 2 — Pareto the biggest cause
Once rejection is captured with its operation, work center and defect, the improvement method is almost mechanical: rank the causes and attack the largest first. A rejection MIS built on source data lets you sort losses two ways — by defect type and by work center — and the two views together usually point straight at the constraint.
The discipline is to resist spreading effort evenly. Most plants find that a small number of defect-and-station combinations drive the majority of reject cost. Fixing the top one or two moves the number more than a dozen scattered actions, and because the data is granular you can prove the improvement afterward rather than argue about it.
| Approach | What it tells you | Can you act on it? |
|---|---|---|
| Single end-of-line reject % | That quality is a problem | No — no location |
| Reject by operation | Which step loses yield | Partly |
| Reject by work center + defect | Which cause at which station | Yes — ranked |
| Reject at child-part level | Component vs assembly cause | Yes — precise |
Step 3 — Run a controlled rework loop
Capturing and ranking rejection reduces how much you produce; a controlled rework loop reduces how much of what you rejected you actually lose. The difference between rework as chaos and rework as a managed process is whether it is defined and measured.
The crucial detail is step 4: rework has its own yield. If salvage operations themselves reject a large share of what they touch, the rework may be costing more than the material it saves. Only a measured rework loop can tell you that, which is why rework should be routed and booked exactly like primary production rather than done off the books.
Rejection tracked as one number nobody can act on?
See reject captured per operation and work center, Pareto-ranked, with a controlled rework route that measures salvage — on your own parts, in 30 minutes.
Step 4 — Make the scrap decision deliberate
Not everything can or should be reworked. The goal is not zero scrap; it is that every scrap decision is a decision rather than a default. When a rejected quantity cannot be salvaged economically, it is removed from the route as process scrap, and any replacement need can raise a purchase requisition against the work order so the shortfall is visible to procurement rather than absorbed silently.
A deliberate scrap decision also protects the rework loop from abuse. If borderline parts are shoved into rework to avoid recording scrap, rework yield collapses and the loop stops recovering value. Recording scrap honestly — and comparing scrap cost against rework cost — keeps both paths pointed at the cheaper outcome for each case.
Rejection, rework and scrap — the terms
These three words are used loosely on most floors, and the looseness hides cost. Kept distinct, they describe a clean chain.
- Rejection — the event: a part failed at an operation or on the line, captured with its work center and defect.
- Rework — the salvage outcome: a defined rework route repairs the part and returns it to finished goods or the main flow.
- Scrap — the write-off outcome: the part is removed from the route; replacement may raise a purchase requisition.
How Fast Production Software does it
Fast Production Software implements this whole chain as linked documents on one engine, so rejection, rework and scrap are captured, ranked and recovered within the same work-order flow — not on a separate quality spreadsheet that never reconciles to production.
For the full execution chain this sits inside — from BOM release through material issue to finished-goods transfer — see the pillar guide on production management software.
Frequently asked questions
How do you reduce rejection in production?
Reduce rejection by first capturing it where it happens — good and reject WIP at each operation, plus line rejection at part and child-part level booked against the work order — and mapping every defect to the work center that produced it. That data lets you Pareto losses by defect type and by station, so you fix the single biggest recurring cause first instead of reacting to a total. You cannot reduce what you only measure as one end-of-line number.
What is the difference between rejection, rework and scrap?
Rejection is the event: a part failed at an operation or on the line. Rework is one outcome: the rejected part can be salvaged through a defined set of repair operations and returned to finished goods or the main flow. Scrap is the other outcome: the part cannot be saved and is removed from the route entirely, and replacement material may raise a purchase requisition. A disciplined system records the rejection, then the rework-or-scrap decision, then the result.
Why does rejection cost a plant twice?
Rejection costs first in the scrapped material and the operations already spent on it, and again in the rework, re-inspection and schedule delay it triggers. If the rejected part is quietly written off and replaced, the plant pays for the same material twice — once for the reject and once for its replacement. Capturing rejection against its operation and work order, and salvaging through a controlled rework route where possible, is how that double cost is contained.
What is a rework route?
A rework route is a small process sheet of its own that lays out the salvage operations for a rejected quantity. Rather than informally reworking parts and hoping they pass, the rework route is defined, executed and tracked with its own good and reject booking, and the salvaged parts either transfer to finished goods or return into the main work-order flow. Because rework itself has a yield, measuring it tells you when salvage is worth the effort and when scrap is the cheaper decision.
How does capturing rejection per operation help?
When reject is booked at the operation that produced it, and each defect is tied to a work center, the plant can see exactly where yield is lost — not just that it is lost. That converts a vague sense that quality is a problem into a ranked list of specific causes at specific stations, which is the only starting point for a rejection-reduction programme that actually moves the number.