Follow-up actions after quality management system audit: part 3
A quality management system audit only creates value when its findings change how the factory works day to day. Too many follow-ups stall at paperwork: a plan is written, documents are sent, and nobody checks whether the process actually behaves differently under normal production pressure. The discipline that matters is closure with evidence, where each finding is graded by risk, assigned to an owner, fixed at its cause, and then verified on the shop floor. This article sets out that routine from grading through effectiveness checks that confirm the fix holds.
Key takeaways
- Grade every finding by risk to product and customer before agreeing any deadline, so critical system gaps draw resources first.
- Require corrective action plans that name the cause, the process change, and the evidence, following the corrective-action logic described in quality management standards. [2]
- Verify fixes where the work happens through records, observation, and repeat checks rather than accepting photographs alone.
- Pair system follow-up with sampling-based product checks, because lot verdicts show whether process changes actually reduced defects. [1]
- Close findings only after effectiveness evidence covers a full production cycle, and keep the history for the next audit.
The most common failure after a quality management system audit is not disagreement about the findings. It is drift. The report arrives, everyone agrees action is needed, and then daily production pressure pushes the corrective work aside. Three months later the same gaps reappear at the next audit, and both sides have paid for an exercise that changed nothing.
A useful mental model comes from the quality management tradition itself. Standards in the ISO 9000 family treat corrective action as a cycle: identify the nonconformity, find its cause, act on the cause, and check that the action worked. [2] Each step needs an owner and evidence. Skipping the last step, the effectiveness check, is why so many audit findings recur despite plausible-looking plans.
This article works through that cycle in practical terms. It covers grading findings by risk, writing plans an auditor can actually verify, choosing closure evidence for each finding type, running effectiveness checks, and using product-level results to confirm that system fixes changed outcomes.
Grade findings before promising dates
Not every finding deserves the same urgency, and treating them equally guarantees that the important ones wait behind the easy ones. Start by sorting findings into three bands: system gaps that can put nonconforming product in a buyer's hands, weaknesses that raise the probability of defects without guaranteeing them, and documentation gaps that matter for traceability but do not directly affect product. The first band gets the shortest deadlines and the most senior owner on the factory side.
For each finding, write one sentence that connects it to product risk in plain language. A missing incoming-material check for a safety-related component is a first-band issue; an unsigned training record for an experienced operator performing a stable process is a third-band issue. This sentence does more work than any severity code, because it tells the factory manager where to spend the next two weeks and gives the buyer a defensible reason for the timetable.
Then negotiate dates against the grading, not against optimism. Documentation fixes can close within weeks because they need only office work. Process changes need a full production cycle to prove themselves, since a fix demonstrated on a trial run may not survive shift changes, rush orders, or substitute operators. Buyers who accept the same two-week deadline for every finding end up closing paperwork while the real risks stay open.
Write corrective action plans an auditor can verify
A verifiable plan has four parts: the cause as the factory understands it, the change to the process, the person responsible, and the evidence that will prove the change happened. Cause analysis should go one level deeper than the finding itself. If the audit found uncontrolled rework, the cause is not that operators reworked product; it is that no defined rework route, authorization, or re-inspection step existed, so rework happened informally.
The process change should be specific enough that a stranger could check it. Revised work instruction number, new checkpoint on the line, additional gauge at a stated station, retraining of named roles with attendance records: these are checkable. Promises to strengthen awareness or enhance attention are not, and experienced auditors treat them as placeholders. Many buyers schedule a UTS supplier QMS audit follow-up visit once the plan is submitted, so the same audit logic tests whether the promised change is visible on the floor.
Evidence needs a date and a form agreed in advance. State which records will be shown, over which production period, and who will present them. For a new incoming-material check, that might mean three consecutive lots of receiving records with rejection entries where applicable, plus observation of the check during the visit. Agreeing this upfront prevents the familiar endgame where the factory sends a handful of photographs and considers the matter closed.
| Finding type | Typical corrective action | Closure evidence |
|---|---|---|
| Missing or outdated procedure | Revise and approve the document, train affected roles | Approved revision, distribution record, training attendance |
| Skipped in-process check | Restore the checkpoint with defined frequency and records | Completed check records over one production cycle plus observation |
| Uncontrolled rework or repair | Defined rework route with authorization and re-inspection | Rework log entries with re-inspection results |
| Uncalibrated or missing gauges | Calibration schedule with recall and verification | Calibration records and gauge status labels seen on site |
| Untrained or unqualified operators | Qualification program with supervised practice | Qualification records and roster showing only qualified staff assigned |
| Weak traceability or identification | Lot identification rules from receiving to packing | Traceability exercise on a recent shipment, forward and backward |
- Check persistence, not just existenceReturn after one production cycle and check whether the changed step is still performed the agreed way, by the agreed people, with complete records. A fix that lasts two days is a demonstration, not a correction.
- Sample records across time and shiftsFor record-based fixes, sample across dates, shifts, and operators. Consistent entries across a month prove more than a perfect week prepared for the visit.
- Observe and interview on the floorWatch the process run and interview operators about what changed and why. People who lived through a real change describe it concretely; people handed a new form describe the form.
- Trace a real lot through the fixPick one recent lot or shipment and trace it through the corrected process, from the affected input to the finished record. The trace shows whether the fix connects to real product.
- Compare product data before and afterWhere the original finding predicted defects, pull the inspection data for lots produced after the fix and compare it with the baseline from before. Numbers that do not move mean the cause analysis missed.
Let product results confirm the system fix
System audits and product inspections answer different questions, and follow-up is stronger when both are used. The audit asks whether the process is controlled; the inspection asks whether the lot is acceptable. A factory can hold a tidy corrective-action meeting yet keep shipping the same defect mix, and only lot-level data will reveal it. Sampling theory explains why a well-drawn sample supports a verdict on the whole lot, which is what makes this cross-check legitimate rather than anecdotal. [1]
In practice, tighten product inspection for the lots produced while corrective actions are being implemented, and track defect classes that relate to the audit findings. If the audit found skipped function testing, watch functional defect rates specifically. If it found weak packing control, watch transit-damage and count-accuracy results. When the targeted defect trend falls across several consecutive lots while the process records show the fix in use, the buyer has converging evidence from two independent angles.
Keep this combined history where the next audit can use it. The opening meeting of the following audit goes faster and lands harder when the buyer can show which findings closed, what evidence closed them, and how product data moved afterward. Suppliers also take the cycle more seriously once they see that audit promises are checked against shipment results rather than filed away. That linkage is what turns a single audit from an event into a management routine.
Sources
- Wikipedia: Acceptance sampling —— Overview of acceptance sampling including AQL concepts used in product inspection.
- Wikipedia: ISO 9000 —— Overview of the ISO 9000 family of quality management standards.
Frequently asked questions
How much time should a supplier get to respond to audit findings?
Give the factory a short deadline for the plan, typically two to three weeks, and a longer window for effectiveness evidence, typically one full production cycle. Plans can be written quickly; proof that the fix works under normal production takes longer. Agree both dates in writing before the auditors leave or within days after the report arrives.
Does submitting a corrective action plan count as closing a finding?
No. A closed finding means the cause was removed and the fix was verified, not that a document was sent. Reworded procedures, training slides, or photographs of a single good lot are progress reports, not closure. Keep the finding open until records and observation confirm the new way of working is actually in use.
What if the supplier ignores the corrective actions?
Escalate in steps: first a joint review of the specific gap, then tighter inspection of outgoing lots, then commercial consequences such as reduced order share. Most suppliers respond to the second step because extra inspection is visible and costly. Document each step so the history supports the next sourcing decision.
Is a desk review of evidence enough, or is a revisit needed?
A desk review of documents costs less and works for documentation findings such as missing signatures or outdated procedures. A revisit is needed when the finding concerns shop-floor behavior such as skipped checks, uncontrolled rework, or untrained operators. When in doubt, verify on site, because behavior findings are exactly the ones paperwork cannot prove fixed.