A defect’s cost is set by where you catch it
Follow one defect downstream
A defect is born at a workstation. From there it can be caught in four different places — and each one costs more than the last. Watch the same defect travel.
It starts at a workstation.
Somewhere on the line, a bolt goes in cross-threaded. Right now it’s invisible. The question that decides everything: where does someone finally notice?
A In-process.
The operator catches their own mistake and fixes it before the part moves on. Nobody downstream ever sees it. Cost: 1×. This is the cheapest defect there is.
B Internally found.
It slips past the source and a later process finds it. Still inside the company — but now it’s rework, scheduling, scrap. Cost: ~10×.
C Externally found.
It reaches final inspection or the customer. Containment, sorting, an unhappy phone call. (A bad delivery from your own supplier, caught at your receiving dock, is the same class.) Cost: ~100×.
D Market defect.
It escapes everything and fails in the field. Warranty, recall, reputation. The same bolt — now the most expensive object in the building. Cost: ~1000×.
*The 1–10–100–1000 figures are the “cost of quality” rule of thumb — illustrative, not exact.If a defect’s cost is set by where it’s caught, then quality work has one job — move the catch point upstream, toward Ⓐ, and let nothing flow to the next process. But to manage that, you first have to see the whole chain a defect can travel through.
Material flows one way.
Suppliers feed your plant; your plant feeds the customer; the customer feeds the market. Quality has to be assured across every link, not one department at a time.
Four places to be caught.
Inside your plant (A at the station, B a later process), at the customer or final inspection (C), or out in the market (D). The dots mark where each class is found.
Every defect sends a signal upstream.
A claim or rejection has to travel back to whoever is responsible, fast, so the cause gets fixed at the source. The dashed arrows are that quality-information flow.
Quality is cross-sectional.
No single department can see the whole chain — so quality assurance owns the end-to-end view, from first supplier to end user. That’s the job.
Classification sounds bureaucratic until you do it. The rule is only ever about one thing: the place the defect was first detected — not who caused it, not how bad it is. Drag each defect to where it was caught and see its class.
The Defect Sorter
Zoom from the supply chain into a single line. Every station is a chance to catch a defect at the source. When it works, the station lights green and the part moves on clean. When it doesn’t, the miss rides downstream — and changes class.
Four stations and a gate.
A part moves left to right through the line, then a final-inspection gate before it ships. Each station can catch a defect — if the operator is set up to.
A The station that made it, finds it.
Station 2 makes the defect and catches it on the spot. Green light. The part moves on clean — no one downstream is affected.
B A later station catches the miss.
This time station 2 misses. The part rides on until station 4 spots it. Internal rework — better than shipping it, worse than catching it at the source.
C Final inspection saves the day.
It rode the whole line undetected. Only the inspection gate stops it from shipping. Expensive to contain — and a sign the line itself isn’t catching enough.
D Nothing caught it.
Past every station, past inspection, out the door. Now it’s the customer’s problem — and yours, at 1000×.
One defect is an anecdote. The four classes, counted every month, are a dashboard — they tell you whether you’re catching more at the source, whether outflow is falling, and whether your fixes actually held.
Defects per month, by class.
Stack the four classes month over month. The total height is your total escape problem; the colors tell you where it’s being caught.
Draw the line you committed to.
A target isn’t a hope — it’s a line on the chart you don’t move, even on a bad month. The gap between the bars and the line is the work.
Mark what changed.
A model change, a new operator, a die swap — flag every change point on the timeline. Most spikes line up with one. The flags turn “why did it jump?” into a short list.
Red and amber shrink first.
The market (D) and external (C) bands are what you attack first — every one is a defect that escaped. Watch them thin out.
Illustrative data.This is the single picture of a quality transformation: the defect mix shifts from red and amber — caught late, or not at all — toward green. Fewer total defects, and the ones that remain are caught at the source.
Caught late — if at all.
At the start, most defects show up downstream: external and market. The total is high and the cost is hidden in warranty and rework.
The source starts catching.
As stations get standards, training and the daily habit of fixing today’s defect today, the green band grows and the total falls.
Mostly green, and smaller.
Now the bulk of what little remains is caught at the source. Fewer defects overall — and the expensive ones have nearly vanished.
Catch everything at the source.
The destination: a short green bar. Every defect found by the person who made it, nothing flowing downstream, market escapes at zero.
Classification isn’t paperwork. It’s a map of where to fight.
Name where every defect is caught, count the four classes every month, and the path is obvious: push the catch point upstream until the operator who makes a defect is the one who finds it. That’s what Kaizumi trains teams to do.
See how Kaizumi builds quality habits →Sources & method
Frequently asked
- What is defect classification?
- Classifying a defect by the place it is first detected, not by who caused it or how serious it looks. The four common classes are: in-process (caught by the operator who made it), internally found (caught at a later internal process), externally found (caught at final inspection or by the customer), and market (it escapes everything and fails in the field).
- Why does it matter where a defect is caught?
- Because the cost is set by where it is caught. The same defect is a quick fix at the source, a rework headache one process later, a containment scramble at final inspection, or a recall in the field — a rough escalation often summarized as the 1-10-100-1000 “cost of quality” rule of thumb. Classifying defects makes that escalation visible so you can push detection upstream.
- What does “your next process is your customer” mean?
- It means any outflow of a defect to the next station counts as an escape, not just outflow to the end market. Treating the subsequent process as a customer is what drives a plant toward zero outflow anywhere, which is the real goal of quality assurance.
- What is the difference between an in-process and an internally-found defect?
- An in-process defect is caught by the worker who made it (or others on the same shop floor) before the part moves on — the cheapest defect there is. An internally-found defect slipped past the source and was caught at a later internal process; it never left the company, but it now means rework, scheduling, and scrap.
- How do you use defect classification to improve quality?
- Count the four classes every month. The total is your escape problem; the colors tell you where defects are being caught. A maturing plant shows the mix shifting upstream — the external and market classes shrink first, and the in-process class grows — until most defects are caught by the person who made them and market escapes approach zero.
Related
Founder of Kaizumi, an AI-powered Lean training platform. More about Matthew →
Updated June 27, 2026 · The defect classes and quality-information flow draw on Sadao Nomura's The Toyota Way of Dantotsu Radical Quality Improvement (Lean Global Network). The cost multipliers are the common “cost of quality” rule of thumb, and the charts and examples are illustrative.