Too many rejections
Scrap, rework and customer returns are rising. Find the operation and defect that create most of the loss before asking people to work harder.
What it looks like
- Rejected pieces sit beside a machine without a count or defect label.
- The quality report gives one daily rejection percentage for the whole factory.
- Parts are reworked quietly and then counted as good production.
- The same defect appears after every tool change or new shift.
- The customer finds a dimension error that internal inspection should have caught.
Why it happens
Defects are counted too late. The team records rejected pieces after the batch. It cannot reliably connect them to a machine, shift, tool or material lot.
Rework is hidden. A part that needed two extra hours is treated like a good part made right the first time. The true cost disappears.
Instructions and drawings are unclear. Operators use different revisions or learn a special tolerance verbally. Each shift works from a different understanding.
First-piece checks are skipped. A wrong setting stays on the machine for hundreds of pieces before someone measures the output.
Quality is treated as one department's job. Inspection catches defects after value has been added. The operation that created them receives no useful feedback.
An overlooked cause is measurement itself. A gauge with the wrong setting or overdue calibration can wrongly accept or reject parts.
What to check first
- Take yesterday's rejected parts or rejection slips. Group defects by type, machine and operation. Keep scrap and rework separate. If labels are missing, write 'unknown' instead of guessing.
- Check the current job instruction at the affected machine. Compare its drawing revision, tolerance and set-up notes with the approved master. Flag any handwritten changes without approval.
- Check the first-piece and gauge records. Find the actual first-piece approval for the affected batch. Check whether the measuring instrument was suitable and within its required check period.
These checks identify patterns. They do not prove the technical root cause without a controlled investigation.
How to fix it
- Start a defect-by-operation count this week. Record produced quantity, scrap quantity, rework quantity and defect type. Keep each shift visible.
- Choose the highest-impact defect. Base this on rejected quantity, material value and rework effort. Avoid running ten corrective projects at once.
- Contain current bad material. Label suspect parts and keep them separate until quality decides their status. Prevent accidental dispatch.
- Observe the exact operation. Check settings, tools, incoming material, handling and inspection method. Ask operators what changes when the defect appears.
- Test one corrective action. Assign an owner, start date and verification method. Check the next several batches before declaring it effective.
- Update the work instruction. Train the affected shift and remove outdated copies. Record what changed so the defect does not return after staff rotation.
Use one line per defect group:
| Shift | Operation | Produced | Scrap | Rework | Defect | Action owner |
|---|---|---|---|---|---|---|
| Morning, example | Boring | 500 | 18 | 12 | Hole oversize | Quality lead |
Always record rework separately. Otherwise, the first-pass result appears better than the actual process.
A worked example
For example, an auto-component unit produces 500 bushes in one shift. Eighteen are scrapped and twelve need rework. Each scrap part contains ₹140 of material and processing cost. Scrap cost for the example is 18 × ₹140 = ₹2,520. Rework has an additional cost that must be counted separately.
The team finds most oversize holes began after a tool change. It checks the tool setting and first-piece measurement before adjusting anything. A different pattern might point to worn tooling or measurement error. The record tells the team where to investigate.
How you know it is fixed
- For four weeks, compare scrap and rework per 100 pieces at the same operation.
- Check how many batches have a recorded first-piece approval before full production.
- Confirm that the selected defect stays lower across different shifts and batches, not only one trial run.
When it is more serious
Stop dispatch and seek technical guidance when a defect affects safety, customer-approved specifications or regulated goods. Follow the customer's notification rules. If measuring instruments cannot be trusted, quarantine affected work until its status is clear. Repeat customer rejections need documented containment and corrective action, not only replacement stock.
This works if someone does it every day. In most businesses, that someone is you.
Questions people ask
Should rework be counted as rejection?
Track it separately from scrap and first-pass good pieces. Both scrap and rework show work that failed the first time.
What should we investigate first?
Start with the defect that has the greatest effect on cost, customer risk or production. Confirm its pattern by operation.
Does more final inspection reduce the root cause?
It may catch more bad parts. You still need to change the operation that creates the defect.
Can one good batch prove the fix worked?
Usually not. Check several batches, shifts and relevant operating conditions before closing the action.