Reading inspection reports without guesswork: part 3
Every inspection verdict is an arithmetic statement wearing a quality costume. Behind the words pass and fail sit counted defects, sorted into classes, compared against accept and reject numbers that were fixed before the visit. Readers who understand that machinery can audit any report in minutes: re-sort the defects, replay the comparison, and check whether the functional evidence supports the paper verdict. This instalment explains the three defect classes, the accept-reject comparison, and the role of functional testing that factory acceptance practice has refined over decades.
Key takeaways
- Recompute the verdict from the printed defect counts and accept numbers instead of trusting the verdict line alone. [1]
- Read critical defects as a separate veto track: one confirmed critical can fail a lot regardless of the major and minor totals.
- Check that functional and safety tests were actually run on the day, not merely referenced from older certificates. [2]
- Demand per-class defect lists with locations and photos so rework instructions are unambiguous.
- Track repeated defect classes across reports, because a defect that returns every order is a process signal, not bad luck.
Parts 1 and 2 of this series covered report structure and scope coverage. Part 3 goes one level deeper into the verdict engine itself: how defects are classified, how counts turn into accept or reject, and where functional testing fits. These mechanics are identical across most consumer goods inspections, so learning them once pays off on every future report.
The classification scheme will feel familiar to anyone who has bought industrial equipment. Critical defects threaten safety or function, major defects impair use or salability, and minor defects are cosmetic blemishes within tolerance. What differs in consumer goods is the sampling frame: instead of testing one machine exhaustively, the inspector classifies defects across a drawn sample and applies statistical accept numbers.
Functional testing deserves special attention because paper checks cannot catch it. A product can be dimensionally perfect, beautifully packed, and completely non-functional, which is why acceptance practice in capital goods insists on running the equipment before signing. The same instinct belongs in every consumer goods inspection.
The three defect classes and what each one decides
Critical defects create safety hazards or total functional failure: exposed mains wiring, sharp metal edges on a toy, a charger that overheats. Acceptance logic treats them as a veto track with effectively zero tolerance, because no sampling arithmetic can make a safety hazard shippable. When a report lists a critical, skip straight to the containment section: how many pieces are affected, whether the whole lot is suspect, and what corrective evidence is required.
Major defects impair the use, performance, or salability of the product without endangering anyone: a motor that stalls, a zipper that separates, a display with dead pixels beyond tolerance. These carry the numeric accept limits that decide most verdicts, typically calibrated so that a lot at the agreed quality level passes most of the time while a worse lot fails. The report should name the limit applied so the arithmetic is checkable.
Minor defects are cosmetic departures inside agreed tolerance: slight color variation, faint molding lines, barely visible scratches. They matter commercially in aggregate, which is why they are counted rather than ignored, but their accept numbers are generous. A lot failed on minors alone usually signals either a tolerance set too tight for the price point or a factory drifting out of process control.
Replay the accept-reject arithmetic
| Element | Where to find it | How to verify |
|---|---|---|
| Lot size and sample size | Scope header and sampling section | Sample matches the plan for that lot size and inspection level |
| Accept and reject numbers | Sampling section, per defect class | Figures match the published tables for the stated plan |
| Defect counts by class | Findings pages with defect list | Counts reconcile with photos and per-carton notes |
| Critical defect handling | Dedicated critical section | Any confirmed critical triggers the veto procedure |
| Final verdict | Summary page | Verdict follows mechanically from counts versus limits |
| Retest or resample notes | Appendix or remarks | Any second draw is documented with cause and new counts |
Functional testing is where paper verdicts get honest
Run it, do not reference it
The most common weakness in consumer goods reports is a functional section that cites earlier certificates instead of recording tests performed on the day. Factory acceptance practice for capital equipment takes the opposite stance: the machine is run, measured, and stressed in front of witnesses before anyone signs [2]. Import buyers should demand the same discipline at sample scale: named functions tested, on how many pieces, with what result.
On-the-day testing catches the failures that visual inspection cannot: firmware that crashes after ten minutes, batteries that sag under load, seals that leak under pressure. None of these show up in a dimensional check, and all of them generate the customer returns that destroy margins.
Sampling logic applies to tests too
Functional tests are usually run on a subsample of the drawn sample, and that subsample needs the same statistical honesty as the visual check. Testing three pieces out of a two-hundred-piece draw proves little; the report should state the functional subsample size and any failures found within it. Acceptance sampling theory is explicit that conclusions only travel as far as the sampling plan carries them [1].
Where full functional testing of every piece is impractical, define the subsample by risk: test the functions customers use first, the functions with past failures, and the functions tied to safety. Record the selection logic so the next order can repeat or improve it rather than reinventing it.
Turning one report into a trend
- Extract per-class countsLog critical, major, and minor counts per SKU from this report into a running sheet with the order date.
- Name the repeating classesHighlight any defect class that appears in two consecutive reports; repetition is the signal that randomness has left the building.
- Trace to the process stepAsk which operation produces the repeated defect and whether its controls, tooling, or incoming materials changed.
- Escalate deliberatelyWhen repeats survive two rounds of rework promises, move from product inspection to systems investigation with a factory audit service that verifies whether corrective actions exist on paper or on the floor.
- Tighten the next scopePromote the repeated defect class into explicitly booked criteria with photos of the boundary between pass and fail.
Never let a verdict travel without its arithmetic. Counts, classes, and accept numbers fit on one page and settle most disputes before they start.
Sources
- Wikipedia: Acceptance sampling —— Overview of acceptance sampling including AQL concepts used in product inspection.
- Wikipedia: Factory acceptance test —— Overview of factory acceptance testing performed before shipment or installation.
Frequently asked questions
Can a shipment pass with critical defects noted?
Only if the suspected criticals were investigated and downgraded with documented evidence, for example when an exposed wire turned out to be a removable transit tie. A confirmed critical defect should fail the lot or at minimum trigger contained sorting with re-verification. Any report that lists confirmed criticals alongside a pass verdict needs a written explanation you can defend to your customer.
The factory says the defect is within tolerance. How do I judge?
Ask for the tolerance in writing with its source: the approved sample, the product specification, or the cited standard. Then compare the retained boundary sample against the failed pieces. If no written tolerance exists, the dispute is really about an undefined requirement, and the fix is to define it with photo-illustrated limits before the next production run.
How many pieces should functional testing cover?
Enough that a systematic functional failure would almost certainly be caught, which usually means a stated subsample of the drawn sample weighted toward high-risk functions. The exact number depends on the failure history and the cost of a field failure. Whatever number you choose, print it in the scope so the report can confirm it was executed.
When do repeated minor defects justify failing a shipment?
When their count breaches the stated accept number for minors, or when the pattern shows the process is out of control even though this particular lot scraped through. Failing on minors is commercially painful, so reserve it for cases where the trend sheet proves the factory is drifting. Otherwise convert the pattern into tightened criteria for the next order.
Position sharper before your next launch.
We cap at 18 new clients per quarter to protect senior attention. A Strategy Intensive is the front door — 60 minutes, working session, no pitch.