What rejection and rework really cost
Rejection is the most visible loss in a machine shop and the most under-counted. The scrapped part is only the surface. Underneath sit the machine time already spent, the operator hours, the tooling wear, the inspection labour that caught it, the containment and sorting, the disruption to schedule, and — when a defect escapes — the customer’s confidence. A plant that tracks only a monthly reject total is measuring the tip and ignoring the iceberg.
Rework is subtler still, because it feels like a save. A part that would have been scrapped is recovered — so rework looks free. It is not. Rework consumes capacity you could have used to make good parts, it risks introducing new defects, and it hides the real defect rate because a reworked part often gets counted as “good” in the end. The goal is not to rework better; it is to need less rework, by stopping the defect at its source. This guide lays out the fix chain that gets you there.
The fix chain that actually works
Reducing rejection is not a slogan campaign; it is a repeatable sequence. Each link depends on the one before it, which is why plants that jump straight to “inspect more” or “retrain the operator” rarely move the number. The chain that works is: code → prioritise → solve → prevent → verify.
Skip the first link and the rest collapse: you cannot prioritise defects you never coded, you cannot solve a problem you cannot name, and you cannot verify a drop you cannot measure. So the reduction always starts with data discipline, not with a kaizen event.
Start with defect-code discipline
The foundation of every reduction is a controlled defect-code catalogue and the habit of tagging every rejection from it. “Bore oversize”, “taper”, “burr”, “surface finish”, “wrong OD” — a finite, agreed list. Without it, your rejection data is a heap of free text that cannot be counted, and every improvement conversation degenerates into anecdote and blame.
Two refinements make the codes far more powerful. First, map each defect code to the work centre / operation where it tends to arise, so a defect immediately points at a process rather than a vague “quality problem”. Second, keep the same code language across incoming, in-process and final rejection, so the whole plant speaks one defect vocabulary. In Fast Quality the defect masters — the defect-code catalogue and the defect-vs-work-centre mapping — are shared across every rejection entry, which is what makes the Pareto in the next step trustworthy. For the full non-conformance workflow, see NCR and non-conformance management.
Let the Pareto choose the fight
With coded data, the rejection Pareto tells you where to spend your limited improvement effort. The Pareto principle is remarkably durable in machining: a handful of defect codes typically drive the majority of your rejection cost. Ranking defects by cost — not just count, because an expensive part rejected rarely can outweigh a cheap part rejected often — points you at the fight worth having.
Resist the urge to fix everything at once. A plant that opens fifteen corrective actions closes none of them well. The discipline is to take the top one or two defects on the cost Pareto, drive them hard to root cause, verify the drop, and only then move to the next. Slice the Pareto by defect, by process and by part (see quality reports and KPIs) to confirm you are attacking a real concentration rather than noise. In Fast Quality, the process rejection MIS produces exactly these Pareto views off the coded rejection data.
8D the top defects — root cause, not re-inspection
Here is where most reduction efforts quietly fail. A top defect is identified, and the “corrective action” is: re-inspect the batch, counsel the operator, add a 100 % check. None of those is a root cause — they are containment dressed up as a fix, and the defect returns the moment attention moves on.
The tool that forces a genuine fix is the 8D. Its discipline is the point: D3 contains the immediate problem, but D4 demands the actual root cause — why did the process produce this defect? — and D5 a permanent corrective action that removes that cause. A worn tool with no life-tracking, a fixture that shifts, a program offset never verified, a material lot out of spec: these are root causes. “Operator error” almost never is; it is usually a process that permitted the error. Best practice is to open an 8D on the top Pareto defects and hold the team to a real D4. In Fast Quality, a major or recurring rejection escalates into an 8D whose CAPA is driven back into the FMEA and control plan.
Attacking every defect and moving none of them?
See a rejection Pareto pick the fight, an 8D drive a real root cause, and a control-plan change close the loop — in 30 minutes, on your own top defects.
Tighten the control plan
A root cause found and fixed on the line will drift back unless the control plan is updated to prevent recurrence. This is the step that converts a one-time fix into a permanent one. If the 8D found a worn tool, the control plan gains a tool-life check at a defined frequency. If it found an unverified offset, the control plan adds a first-off verification. If it found a slipping fixture, the reaction plan changes.
Crucially, the control-plan change must flow through change management, not a quiet edit. The special characteristic the defect touched is re-examined, the FMEA occurrence score is updated to reflect the fix, the control plan is amended under approval, and — for automotive parts — a PPAP re-submission is triggered where the change is significant. This is what keeps your risk analysis honest: the FMEA should reflect the risk you just proved was real and then reduced. In Fast Quality, this loop runs through documents and change management, amending the control plan and FMEA under a controlled change.
Rework or scrap — the honest economics
When a part is non-conforming but recoverable, someone must decide: rework or scrap. The honest answer depends on numbers, not instinct. Rework is worth it only when the cost to recover the part is comfortably less than its value, and the rework does not risk the part’s integrity or a hidden defect.
| Consideration | Lean toward rework | Lean toward scrap |
|---|---|---|
| Part value | High-value, many operations already invested | Low-value, early in the route |
| Recovery cost | Small, defined rework operation | Rework cost approaches or exceeds new-part cost |
| Risk to integrity | Cosmetic or dimensional, safely correctable | Structural, or a special/safety characteristic |
| Capacity | Spare capacity to absorb the rework | Rework would starve good-part production |
| Traceability | Rework route is controlled and recorded | Uncontrolled rework — scrap is safer |
The non-negotiable rule: rework must be a controlled, recorded route, not an off-book fix at the bench. A proper rework process sheets the operations, tracks the parts through the rework route, and transfers the corrected stock back into production only after re-inspection — so a reworked part is verified, not merely assumed good. In Fast Quality the rework route flows from a rework process sheet through rework status to a transfer of the corrected stock back to finished goods, with a rework report on volume and yield. Rework yield is itself a KPI: if a defect’s rework volume is climbing, that defect belongs at the top of your next Pareto, not in a growing rework queue.
One Pareto, one 8D, a measurable drop
A machine shop runs at an uncomfortable reject PPM. For years the response was “inspect harder”, which caught more defects but changed nothing. The team switches to the fix chain: every rejection is coded, and after a month the Pareto is unambiguous — “bore undersize” on one turned part is 40 % of the reject cost, concentrated at one operation. An 8D contains the suspect stock, and D4 traces the cause to a boring tool run with no life-tracking. The permanent action is a tool-change frequency; the control plan gains a tool-life check; the PFMEA occurrence is re-scored and the change recorded. Over the next two months the PPM trend confirms the defect has all but disappeared, and the rework queue for that part empties. The number moved because the root cause was killed — not because inspection got tougher.
How Fast Quality drives the reduction
Fast Quality Software, built by Improsys in Pune on the shared Fast Suite platform, gives you every link of the fix chain in one system:
Because the rework workflow is shared with Fast Production, corrected stock returns to the same work orders it left, and the whole reduction is traceable. For the closed loop this reduction sits inside, see the pillar guide on what quality management software is.
Frequently asked questions
How do you actually reduce a machine-shop reject rate?
Follow a fix chain, not a slogan: code every rejection with a controlled defect code, use a rejection Pareto to rank defects by cost and pick the top one or two, open an 8D that drives a real root cause (not a re-inspection), tighten the control plan so the defect cannot recur, and verify with the PPM trend that the number actually fell. Each link depends on the one before — you cannot prioritise defects you never coded or verify a drop you cannot measure — which is why reduction starts with data discipline rather than a kaizen event.
Why doesn’t more inspection reduce rejection?
Because inspection finds defects; it never prevents them. Adding a 100 % check or re-inspecting a batch is containment — it stops a known defect escaping, but it does nothing about the process that keeps producing the defect. The reject rate only falls when the root cause is removed, which is the job of an 8D and a control-plan change, not of tougher inspection. The plants with the lowest reject rates are not the ones inspecting most; they are the ones whose corrective actions actually closed the root cause.
When should you rework a part and when should you scrap it?
Rework only when the cost to recover the part is comfortably less than its value and the rework does not risk the part’s integrity. Lean toward rework for high-value parts late in the route with a small, controlled recovery operation; lean toward scrap for low-value parts, when recovery cost approaches new-part cost, or when the defect touches a structural or safety/special characteristic. The non-negotiable rule is that rework must be a controlled, recorded route with re-inspection before the corrected stock returns to production — never an off-book fix at the bench.
What is the hidden cost of rework?
Rework feels free because it recovers a part that would otherwise be scrapped, but it consumes capacity you could have used to make good parts, it risks introducing new defects, and it hides the real defect rate when reworked parts get counted as good. Rework yield is itself a KPI: if a defect’s rework volume is climbing, that defect belongs at the top of your next Pareto, not in a growing rework queue. The goal is not to rework better — it is to need less rework by stopping the defect at its source.
Why must a control-plan change go through change management?
Because a quiet edit leaves your risk analysis lying. When an 8D finds and fixes a root cause, the control plan must be updated to prevent recurrence, and that change should flow through change management so the affected special characteristic is re-examined, the FMEA occurrence score is updated to reflect the fix, the amendment is approved, and — for automotive parts — a PPAP re-submission is triggered where the change is significant. This keeps the FMEA honest about the risk you just proved was real and then reduced, and leaves an auditable trail from defect to root cause to controlled change.