An 8D report comes back for a short list of repeatable reasons, and most of them have nothing to do with how hard the team worked. The reviewer on the customer side is not grading effort. They are checking whether each discipline can be audited — whether the containment protects them today, whether the root cause was verified rather than asserted, and whether the corrective action would survive an auditor asking to see the evidence next year.
One reason dominates: the escape root cause is missing. The report explains why the defect was made and never answers why it reached the customer. After that come containment that does not contain, retraining offered as a permanent action, and a D6 that says “implemented” without data. Formatting rejections — wrong form, blank mandatory fields, contradictory dates — cost a full cycle and are the cheapest to prevent. The checklist below is the reviewer’s view of your report, 24 checks in the order they read them.
Download the 8D pre-submission checklist
Twenty-four checks across D0–D8 plus the format traps, each written as what gets rejected and what passes instead, with a status column and a place to record where the evidence sits. The second sheet is the wording table from further down this page. No signup, no email.
Pre-submission checklist — Excel (.xlsx)
Status drop-down per check, Fix highlighted in red, panes frozen so the discipline stays visible. Second sheet: wording that gets a report sent back.
Pre-submission checklist — Word (.docx)
Landscape table for a review meeting, or for pasting into your own supplier quality procedure.
Pre-submission checklist — PDF (print)
Three printable pages — the version that goes on the table when two people read the report before it is sent.
Writing the report rather than checking it? The 8D report generator keeps the two D4 chains separate on screen and flags the answers that reviewers reject, and the 8D report template gives you the same structure as a file.
What a rejection actually costs
The obvious cost is the rework. The expensive ones are quieter.
- The clock usually does not restart. Customers commonly count elapsed days from the original complaint, not from the rejection, so a returned report eats the time you had left rather than buying more.
- Containment stays in force longer. Sorting, 100% inspection or a firewall at the customer’s plant continues until D6 is accepted — and those are the actions with a daily cost attached.
- The scorecard moves. Most supplier rating schemes score problem-solving quality and responsiveness separately from the defect rate itself, so a report returned twice can cost more rating points than the defect that triggered it.
- Escalation becomes easier to justify. Repeated rejection is one of the reasons customers cite when moving a supplier to tighter controls such as controlled shipping, which is paid for by the supplier.
Where this page names a requirement, it names the source: Ford’s Global 8D defines the escape point in D4, AIAG’s CQI-20 covers effective problem solving as a discipline, and IATF 16949:2016 puts problem solving in clause 10.2.3 and error-proofing in 10.2.4. Everything else — response times, the number of monitoring days, which form to use — comes from your customer’s specific requirements, not from a standard. When the two disagree, the customer wins.
Nine reasons reports come back
1. The escape root cause is missing
This is the one that dominates. The report explains why the defect was created and stops, leaving the reviewer’s second question unanswered: your process made a bad part and then shipped it — which control should have caught it, and why did it not? Ford’s Global 8D names this the escape point and makes it part of D4 for exactly this reason: fixing occurrence alone means the next defect of a different kind still reaches the customer.
A near-miss version gets rejected too. “The inspector missed it” names a person, not a control. Keep going: was the feature visible at that station, was there a sample plan that allowed it through, was the gauge capable of measuring the characteristic, was the check defined at all?
2. The cause terminates at a person
“Operator error” and “human error” are events to be explained, not explanations. If a person could make the error, the reviewer reads that as a process that permits it — an assembly that goes together two ways, a work instruction that leaves the sequence open, a missing poka-yoke, a workload that makes the shortcut rational. IATF 16949 pushing error-proofing into its own clause (10.2.4) is the same logic written into a standard: the expected answer is a control, not a name.
3. Containment that does not contain
“Operators informed” and “team briefed to pay attention” are instructions. A containment is a barrier the customer can verify: stock on hold with quantities, sortation with results, a gauge added at the last operation, a firewall inspection at the receiving plant. Two more things get checked here and are routinely missing — the scope (material at the customer, in transit, in your warehouse, WIP and finished goods, each with a disposition) and the effectiveness (how you checked it works, on what sample, with what result).
4. D2 restates the complaint instead of quantifying it
“Customer found defective parts” is the complaint, not the problem description. D2 is where what, where, when and how many become numbers measured against a target, and where the population is bounded by lot, date code or serial range. Two tools do most of this work: 5W2H for completeness and Is / Is Not for the boundary. Reviewers notice the boundary specifically, because a problem with nothing excluded leaves every candidate cause equally plausible — which is why D4 arrives unconvincing.
5. Retraining is offered as the permanent corrective action
Training depends on memory and cannot be verified in a year’s time. It is a supporting action, not a fix. The same applies to “increase inspection” and “added to the audit plan”: they are containment wearing the wrong label. A permanent action is expected to sit as high in the hierarchy as is feasible — eliminate the failure mode, substitute the method, error-proof it, engineer the control — with training named separately as the thing that supports it.
6. D6 says “implemented” and shows no data
Implementation is a date; validation is evidence. What the reviewer is looking for is before-and-after data over a stated period, with the sample size, showing the defect at target — and containment removed after that, not on the day the action went in. A D6 that closes at the same moment the action is implemented tells the reviewer that nobody has yet shown the fix works.
7. D7 is empty or repeats D5
D7 asks what changed so that this cannot come back somewhere else. That means named documents with revision numbers and dates — PFMEA, control plan, work instruction, inspection standard — plus a read-across to similar parts, lines, shifts and plants. Writing “FMEA updated” without a revision number is the same as writing nothing, because there is no document for an auditor to open.
8. The evidence is referenced but not attached
Layout inspection reports, capability studies, photographs of the containment, the sortation log: if the analysis leans on them, they belong in the file. Attachments get rejected on their own terms too — a photograph with no date, no scale and no caption proves nothing, and a folder of files called IMG_0423 leaves the reviewer to guess which discipline each one supports.
9. The form, the fields and the dates
The cheapest rejections of all. The customer requires their own template or portal and gets yours; mandatory fields are blank rather than marked N/A with a reason; the timeline contradicts itself, with containment removed before the D6 data exists or a D3 date earlier than D0. None of these require reading the analysis, which is exactly why they are caught first.
The wording that gets a report sent back
None of these phrases are forbidden. Each one is flagged because of what it leaves unanswered — so the third column is the point, not the vocabulary.
| Wording in your report | What the reviewer concludes | Write this instead |
|---|---|---|
| Operator error / human error | The analysis stopped at a person, so the system that allowed it is untouched. | What made the error possible, and what let it through undetected. |
| Operator re-trained | The weakest action available, and it depends on recall. | A change to the process or control that works without memory; training stays as support. |
| 100% inspection (as the permanent action) | Inspection is containment — offering it as the fix says no cause was removed. | Keep it as containment, name the permanent action separately, remove the inspection with D6 data. |
| No problem found / could not reproduce | The investigation stopped, and the customer still holds the part. | What was tested, with what method and sample, and what the result excludes. |
| Will be monitored | No acceptance criterion, so nobody can say later whether it worked. | The measure, the target, the period, and who confirms it. |
| Process improved / procedure updated | Nothing auditable — there is no document to check against. | Document name, revision number, release date. |
| Supplier issue | Responsibility moved, escape cause still missing. | Why your incoming control passed it, and what changed in that control. |
| As per standard procedure | If the defect happened while the procedure was followed, the procedure is part of the cause. | What in the procedure permitted the outcome, and what was changed. |
A rejected D4, rewritten
Illustrative, not a real case. A Tier-2 supplier ships a moulded housing; the customer finds a short-shot on the mounting boss at their assembly line and raises a complaint. The first submission was returned in two days.
Three rejections in three lines: the cause ends at a person, the permanent action is training, and nothing explains how a short-shot left the plant.
The rewritten version is not longer because more words were added. It is longer because the second chain existed all along and was never written down — and it is the chain that produced the control plan change, which is what the customer was actually asking for.
What to do when it comes back
- Read the rejection literally, and map each comment to a discipline before changing a single line. A comment written against D5 often belongs to D4: the action looks weak because the cause it addresses was never verified.
- Ask for the acceptance criterion when a comment is vague. “Root cause not sufficiently justified” is not actionable. Asking what evidence would satisfy them is a normal question, and it is cheaper than a second rejection.
- Fix the analysis where the comment points at the analysis. Rewording the same content is the most common cause of a second rejection — the reviewer recognises it immediately, and the second review is less patient than the first.
- Check whether containment is still in force and still documented. An open 8D with lapsed containment turns one finding into two.
- Resubmit with a change log — one line per discipline saying what changed and where the new evidence sits. It shortens the review, and it makes the new content visible to a reviewer who is skimming a document they have already read once.
- Keep the argument out of the report. If you disagree — and sometimes you should, for instance when the defect is genuinely outside your control — that conversation belongs in the covering message. A report that argues with its reader reads as a report that is not going to change.
The 24 checks, in the order a reviewer reads them
Before you press send
- D0 — the symptom is in the customer’s own words, with their reference and the date received
- D0 — the emergency response is recorded with date, time and owner
- D1 — three to seven cross-functional members, including the process owner and a champion with authority
- D2 — what, where, when and how many are numbers, measured against target
- D2 — the boundary is defined: what was equally exposed and stayed unaffected
- D2 — the suspect population is bounded by lot, date code or serial range
- D3 — containment is a barrier the customer can verify, not an instruction
- D3 — scope covers customer stock, transit, warehouse, WIP and finished goods, each with a disposition
- D3 — effectiveness is stated with method, sample size, result and start date
- D4 — the occurrence cause is verified, not asserted
- D4 — the escape cause is present, and it names a control rather than a person
- D4 — the chain reaches the system, not the individual
- D4 — both chains have named evidence, attached
- D5 — every root cause has at least one action, with owner and date
- D5 — the action is stronger than training; training is listed as support
- D5 — the verification method and acceptance criterion were written before the data was collected
- D6 — before-and-after data over a stated period, with sample size
- D6 — containment removed after the data, with the decision recorded
- D7 — PFMEA, control plan and work instructions named with revision numbers and dates
- D7 — read-across to similar parts, lines, shifts and plants, including a negative finding
- D8 — champion sign-off, closure date, and a lesson another team could use
- Form — the customer’s own template and every mandatory field completed or marked N/A with a reason
- Form — dates are internally consistent and match the portal
- Form — attachments numbered, captioned, dated and referenced from the discipline they support
The same 24 lines are in the downloadable file with a status column, the rejected-when and accepted-when wording for each, and a place to note where the evidence lives. A check is passed when the evidence exists — not when the box is ticked.
Frequently asked questions
Why do customers reject 8D reports?
Almost always because a discipline is filled in but not evidenced. The reviewer is checking whether each claim can be audited: whether containment protects them now, whether the root cause was verified rather than asserted, whether the corrective action is stronger than retraining, and whether effectiveness was shown with data. A report can be complete, well written and still rejected if none of that can be traced to a document, a measurement or a date.
What is the most common reason an 8D is rejected?
A missing escape root cause. Most reports explain why the defect was created and stop there, leaving the second question unanswered: why did the control system let it reach the customer? Ford’s Global 8D names this the escape point and makes it part of D4. When the escape cause is present but reads “the inspector missed it”, reviewers treat it the same way as missing, because it names a person rather than the control that failed.
Is “operator error” ever an acceptable root cause in an 8D?
Not as the final answer. An operator error is an event to be explained, not an explanation. The reviewer’s next question is what made the error possible and what allowed it to pass undetected: an assembly that can be done two ways, a work instruction that permits it, a missing poka-yoke, an inspection that cannot see the feature. A cause that terminates at a person leaves nothing to correct except the person, which is why it comes back.
Can a customer reject an 8D report for formatting alone?
Yes, and it happens more often than suppliers expect. If the customer requires their own form or portal fields, a submission on your layout can be rejected without the content being read at all. The same applies to blank mandatory fields, dates that contradict each other, and attachments with no caption or date. These rejections cost a full review cycle and are the cheapest ones to prevent.
What should I do when my 8D comes back rejected?
Read the rejection literally and map every comment to the discipline it belongs to before changing anything. Ask for the acceptance criterion if a comment is vague. Fix the analysis where the comment points at the analysis, rather than rewording the same content. Resubmit with a change log that says what changed in each discipline, and never argue with the reviewer inside the report itself.
How long do I have to resubmit a rejected 8D?
That is set by your customer’s requirements or portal, not by the method. The practical point is that many customers keep counting from the date of the original complaint, not from the rejection, so a resubmission cycle usually eats your original due date rather than restarting it. Check whether your containment is still in force and still documented while the report is open.
Does a rejected 8D affect the supplier scorecard?
Commonly, yes. Responsiveness and problem-solving quality are scored separately from defect rate in most supplier rating schemes, so a report that is returned twice can cost more rating points than the original defect. Repeated rejection is also one of the triggers customers cite when escalating to tighter controls, such as controlled shipping.
What is an escape point in 8D?
The earliest point in the process where the defect could have been detected and was not. It belongs to D4 alongside the cause of occurrence, and it has its own why-chain and its own evidence. Identifying it produces a second corrective action in D5: the occurrence cause is fixed so the defect stops being made, and the escape point is fixed so a future defect of that type is caught before the customer sees it.
Related resources
- 8D report template — the document itself, in Excel, Word and PDF
- 8D problem solving: complete guide — all eight disciplines, with the D0–D8 progress tracker
- 8D examples — five complete filled-in reports, each with both D4 chains
- Is / Is Not analysis — the D2 boundary that makes D4 convincing
- 5W2H template — the completeness half of D2, with an evidence column
- Control plan — the D7 document reviewers open first
- CAPA — the same discipline where the record is regulated
- RCA meeting agenda — the session where a rejected D4 gets redone properly
- 8D report generator — builds the report with both chains kept apart