Last reviewed: June 21, 2026
A scrap reason code is a short label assigned to each scrapped part to identify the cause. Eight to twelve codes is the right range for most plants, with twelve as the practical ceiling. The starter list below covers nine of the most common scrap causes in discrete manufacturing and can be implemented this week with a paper log, a briefing at shift start, and the free printable reference card linked below.
Key takeaways:
Eight to twelve codes is the right range, with twelve as the practical ceiling. Below eight, categories are too broad to Pareto into actionable priorities. Beyond twelve, floor compliance starts to degrade because operators cannot reliably remember which code applies.
"Other" should account for no more than 10% of scrap events over any 30-day period. If it runs higher, the code list is missing a specific category.
Handling damage is the most underreported category in most plants because it happens between operations and has no natural owner. Without a dedicated code, it disappears into "Other."
The codes only work if operators trust that filling them in accurately does not put them at personal risk. The rollout briefing matters as much as the list itself.
WHAT MAKES A SCRAP REASON CODE ACTUALLY USABLE?
A usable reason code has three properties. It is specific enough to point to a countermeasure, broad enough that an operator can classify a part in under 10 seconds, and consistently defined so two different operators on two different shifts assign the same code to the same defect.
Most plants that fail at reason codes fail on the third property. "Process variation" means something different to a machining operator and a stamping operator unless the definition is written down and reviewed at the briefing. The table below includes definitions for exactly this reason.
The codes also need to be physically accessible at the point of scrap. A list posted at the machine, printed on a tag, or entered on a tablet at the cell gets used. A list that requires walking to the supervisor's office does not.
THE STARTER LIST OF 8 TO 12 CODES
This list covers nine of the most common scrap causes in discrete manufacturing. Add one or two process-specific codes if your operation has a known repeating cause that does not fit any of these categories. Download the printable reference card to post at each scrap point.
| CODE | NAME | DEFINITION | EXAMPLE |
|---|---|---|---|
| 01 | Process Variation | Part rejected due to process drift or instability during a run | Dimension creeping toward the reject limit across 50 consecutive parts |
| 02 | Tooling Failure | Part defective due to broken, worn, or improperly seated tooling | Punch cracked mid-run; burr on all parts from that die position |
| 03 | Setup Error | Part defective due to incorrect machine setup, offset, or fixturing before the run | Wrong offset entered; first 8 parts out of tolerance before caught at first-off |
| 04 | Incoming Material | Defect caused by dimensional, chemical, or surface variation in the input material | New coil lot thicker than spec; blanks exceed upper tolerance |
| 05 | Handling Damage | Part damaged after production, during movement between operations or to storage | Scratch introduced in transit tote; dent from conveyor transfer |
| 06 | First-Off Reject | Part rejected during first-off inspection before the run is approved to continue | Setup required two attempts; first part scrapped per procedure |
| 07 | Rework Converted | Part originally logged as rework but downgraded to scrap after rework failed or exceeded cost threshold | Weld repair cracked on reinspection; part scrapped |
| 08 | Operator Error | Catch-all for defects attributable to operator action not covered by other codes; retire as specific patterns emerge | Part loaded in wrong orientation; marked surface scratched during handling |
| 09 | Other | The cause cannot be confidently classified into any existing code, whether unknown or not yet categorized | Novel defect from new tooling not yet categorized; cause under investigation |
HOW MANY CODES SHOULD YOUR PLANT HAVE?
Eight to twelve codes is the right range for most discrete manufacturing plants, with twelve as the practical ceiling. In plants we have worked with, lists below eight are too broad to produce actionable Paretos. A scrap report showing 60% of events attributed to "Process" does not tell you which process, which machine, or what to change. Beyond twelve codes, compliance starts to degrade: operators cannot reliably remember which code applies, they default to "Other," and the data loses its value.
Start with nine. Add one or two codes for process-specific causes that appear repeatedly in the first 30 days of tracking. Resist adding more than two at a time. Adding codes after seeing the data is a sign the list is working, not a sign of failure.
"Other" deserves its own rule: if it exceeds 10% of total scrap events over any 30-day period, the list is missing a specific category. In plants we have worked with, this review typically happens once after the first month and once more after month three. After that, "Other" stabilizes below 5%.
HOW TO ROLL OUT CODES WITHOUT FLOOR PUSHBACK
Rollout resistance almost always comes from one source: operators who believe the codes will be used to assign personal blame. A briefing that addresses this directly is worth more than any system configuration.
The rollout briefing should cover four things, in this order:
1. What the data is for. The codes feed a Pareto that identifies which causes account for most scrap. The Pareto tells the team where to put improvement resources. The data is about the process, not the person.
2. What happens to "Operator Error." It is a temporary catch-all that will eventually be broken into specific process categories as patterns emerge. It is not a permanent personal attribution.
3. What to do when uncertain. If the cause cannot be confidently classified into any existing code, whether unknown or not yet categorized, "Other" is the right answer. Guessing is worse than "Other." The goal is accurate data, not 100% categorization.
4. How to log it. Walk through the physical process: where the log is, how to find the code, how long it should take. If it takes more than 30 seconds, simplify the method.
Print the reference card and post it at every scrap point before the briefing. It lists all nine codes with their definitions in large print. Having it on the wall before the conversation means operators can read along rather than trying to memorize.
WHAT TO DO WHEN "OTHER" IS TOO HIGH
Pull all "Other" events from the log and have a supervisor or quality technician classify them offline. Look for the single most common cause where the operator could not confidently pick an existing code. Add one code for it, update the reference cards, and run another 30 days before adding anything else.
Handling damage is the category most commonly discovered this way. In plants we have worked with, it starts in "Other" because operators do not see themselves as causing the damage (the part was fine when they last touched it) and because no single operation owns the event. Giving it a dedicated code and posting the definition removes the ambiguity.
For context on where your scrap rate stands relative to benchmarks, see OEE, Scrap Rate, and FPY Benchmarks by Process Type. For the full reduction framework beyond tracking, see How to Reduce Manufacturing Scrap Rate.
FREQUENTLY ASKED QUESTIONS
HOW MANY SCRAP REASON CODES SHOULD A PLANT HAVE?
Eight to twelve codes is the right range for most discrete manufacturing plants, with twelve as the practical ceiling. Below eight, categories are too broad to drive a specific corrective action. Beyond twelve, compliance starts to degrade because operators cannot reliably remember which code applies, and Other usage increases.
WHAT IS A SCRAP REASON CODE?
A scrap reason code is a short label assigned to each scrapped part at the time it is pulled from the line. It identifies the cause of the defect. When collected consistently, the codes produce a Pareto chart that shows which causes account for most scrap by count or cost, allowing the team to prioritize improvement work.
SHOULD HANDLING DAMAGE GET ITS OWN REASON CODE?
Yes. Handling damage is one of the most underreported scrap categories because it happens between operations and has no natural owner. Without a dedicated code, it defaults to Other or goes unrecorded entirely. A dedicated code with a clear definition makes the cause visible and assignable.
HOW DO I GET OPERATORS TO FILL IN REASON CODES ACCURATELY?
Address the personal-blame concern directly at the rollout briefing. Make clear the data is used to improve the process, not to evaluate individuals. Provide a physical reference card at every scrap point and make Other the right answer when the cause cannot be confidently classified, whether unknown or not yet categorized. Inaccurate data from reluctant compliance is worse than incomplete data from honest uncertainty.