Factory audit versus product inspection: part 7.
Product inspection answers whether this shipment is shippable. A factory audit answers whether the next ten shipments will be. Buyers who confuse the two keep paying for inspections that pass lots from systems that cannot repeat them, then act surprised when the following order fails on a defect nobody booked. Part 7 of this series draws the boundary precisely: what each tool proves, what each tool costs, and the escalation ladder that connects them. The international framework for conformity assessment supports the distinction, expecting product checks and system assessments to follow transparent, appropriate procedures rather than substituting for one another.
Key takeawaysUse product inspection to verdict finished lots and factory audits to verdict the management system that made them. · Escalate to a systems investigation when the same defect class survives two rounds of promised corrections. · Verify corrective actions with dated shop-floor evidence, not with corrective action forms alone. · Keep sampling arithmetic inside inspection reports, since audit findings do not carry accept numbers. [1] · Write the escalation triggers into the supplier agreement before trouble starts, while everyone is still reasonable. [2]
Every experienced buyer has lived through the loop: an order fails inspection, the factory promises corrections, the re-inspection passes, and the next order fails on the same defect. The loop persists because the tool being applied answers the wrong question. Inspection verifies that a particular lot meets its criteria; it says nothing about whether the process that produced the lot is capable of repeating the result.
A factory audit investigates the machinery behind the lots: incoming material control, process discipline, measurement competence, nonconforming product handling, and whether corrective actions change anything on the floor. Its evidence is systemic rather than statistical, built from records, interviews, and observation across departments rather than from a drawn sample.
This instalment maps the decision a buyer faces after any failed shipment. Re-inspect the lot, demand rework with evidence, qualify a second source, or commission an independent factory audit service to determine whether the supplier's system can hold the requirement at all. Each exit has a price and a proof standard, and choosing well depends on reading the pattern rather than the incident.
What each tool actually proves
| Question | Product inspection answers | Factory audit answers |
|---|---|---|
| Is this lot shippable | Yes, with counted defects against accept numbers | No, audits do not verdict lots |
| Will the next lot be good | No, one lot predicts nothing about the next | Partly, capable systems repeat results |
| Where did the defect come from | Sometimes, when the defect pattern is obvious | Yes, by tracing controls and records |
| Did the correction work | Only for the re-inspected lot | Yes, by verifying systemic implementation |
| Should we keep this supplier | One data point among many | A structured input to the decision |
| How fast is the answer | Days, tied to the shipment schedule | Weeks including reporting and follow-up |
The escalation ladder after a failed lot
- Contain the lotSort, rework, or reject the failed shipment on its own sampling arithmetic; do not let lot containment wait for systems analysis.
- Demand evidence, not promisesRequire dated photos, measurements, and records proving the correction, then re-inspect the reworked scope before release.
- Read the patternCheck whether the same defect class appeared in earlier reports; a repeat converts an incident into a system hypothesis.
- Commission the auditWhen repeats survive two correction cycles, book an independent factory audit service scoped to the failing processes, with the defect history attached to the brief.
- Decide commerciallyUse the audit findings to choose between supervised development, volume reduction, dual sourcing, or exit, with dates attached.
Why sampling logic stays on the inspection side
Inspection verdicts rest on acceptance sampling: a defined lot, a drawn sample, counted defects, and accept numbers fixed before the visit [1]. That machinery gives the verdict its defensibility in commercial disputes. Audit findings work differently: they describe the presence or absence of controls, the completeness of records, and the effectiveness of past corrections, graded by severity rather than counted against a table.
The practical consequence is that audit reports should never be read as pass or fail certificates for product. An audit can find a capable system that still produced one bad lot through a special cause, just as it can find a weak system behind a lot that happened to pass. The two tools complement each other precisely because they measure different things.
International expectations for conformity procedures reinforce the separation: product and system assessments each deserve procedures suited to their object, applied transparently and consistently [2]. A buyer who demands sampling statistics from an audit, or systems assurance from a lot inspection, is asking each tool a question it was never designed to answer.
Scoping the audit from the defect history
The cheapest way to waste an audit is to brief it generically. An audit scoped as check everything lightly will confirm that paperwork exists and change nothing. An audit scoped from the defect trend, naming the failing processes, the suspect controls, and the two correction cycles that did not hold, arrives with hypotheses to test and leaves with findings that bite.
Attach the last three inspection reports to the audit brief with the repeated defect classes highlighted. Good auditors use that history to plan their floor time: which workstations to observe, which records to pull, which operators to interview. The history also sets the follow-up test, because the next inspection after the audit should book the previously failing criteria explicitly and verify that the counts moved.
Price the whole ladder in advance with the supplier while relations are calm. Agree that repeated failures trigger deeper investigation at defined thresholds, and write who pays for re-inspection and audit time into the supply agreement. Clauses negotiated after a failure feel punitive; clauses negotiated at onboarding feel professional.
Inspect the lot, audit the pattern. One failing shipment is an incident; the same failure twice is a hypothesis about the system.
Sources
- Wikipedia: Acceptance sampling —— Overview of acceptance sampling including AQL concepts used in product inspection.
- WTO: Agreement on Technical Barriers to Trade —— WTO portal for the TBT Agreement covering technical regulations, standards and conformity assessment.
Frequently asked questions
My supplier failed one inspection. Do I need an audit?
Usually not. Contain the lot, require evidenced correction, and re-inspect. Commission an audit when the same defect class returns after corrections, when records look assembled rather than lived-in, or when the commercial stakes of the next failure exceed the audit cost. One failure is data; a pattern is a diagnosis waiting to happen.
Can an audit replace pre-shipment inspection?
No. An audit verdicts the system, not the lot, and even strong systems produce occasional bad lots through special causes. Keep lot-level inspection for shipment decisions and use audits to reduce how often those decisions go wrong. Buyers who drop inspection after a good audit rediscover sampling statistics the expensive way.
How do I verify corrective actions without camping at the factory?
Ask for dated evidence tied to the process: revised work instructions with version numbers, retraining records with attendance, measurement logs from the corrected operation. Then book the next inspection with the previously failing criterion named explicitly. If the counts moved and stayed moved across two orders, the correction is probably real.
What should the audit brief contain?
The defect trend across recent orders, the correction attempts and their evidence, the processes you suspect, and the commercial decision the audit must inform. A brief that names hypotheses gets a focused audit; a brief that says evaluate the factory gets a generic report. Attach the inspection history, not just the latest failure.
See your blog's Pillar Health score in 90 seconds.
Free audit. No credit card. Connects to HubSpot, Webflow, Ghost, WordPress and Notion.