Sampling levels used in amazon fba inspection: part 2
Choosing a sampling level answers the easy question; running a sampling program answers the hard ones. When does a supplier earn a smaller sample, and when should failures trigger a larger one? How do mixed-SKU lots get fair coverage? What happens to the sample after a failure, and where does laboratory testing fit beside visual sampling? This second part turns one-off sampling decisions into a program that tightens and relaxes on evidence.
Key takeaways
- Apply switching rules explicitly: tighten sampling after failures and relax it only after a run of consecutive clean lots at normal inspection.
- Stratify every mixed lot by SKU, pallet, and production date so clustered defects cannot hide inside blended averages.
- Re-inspect failed lots at the same level or stricter, and verify corrections on product rather than accepting process paperwork alone.
- Confirm material and safety properties through accredited laboratories whose competence is recognized internationally, not by on-site opinion.[2]
- Keep compliance verification independent of AQL results, because technical measures bind shipments under WTO TBT disciplines regardless of sampling verdicts.[1]
A single sampling plan is a snapshot; a sampling program is a policy that learns. Suppliers improve and regress, product mixes shift, and seasonal pressure changes defect patterns, yet many sellers inspect every lot under identical rules for years. That rigidity wastes money on proven suppliers and exposes the business on troubled ones, often simultaneously within the same mixed shipment.
Switching rules solve this by linking sample strictness to demonstrated history. Clean runs earn lighter inspection, failures trigger heavier inspection, and the criteria for each transition are written down before emotions run high. Combined with stratified draws for complex lots and clear re-inspection policy, the program spends inspection effort where the risk actually sits.
The sections below lay out each mechanism with the practical detail needed to write it into supplier agreements: switching logic, stratification method, re-inspection discipline, the laboratory track, and the compliance boundary that sampling never crosses.
Switching rules that make history matter
Switching discipline has three states. Normal inspection applies to proven situations and standard risk. Tightened inspection follows failures: larger samples, stricter acceptance, expanded packaging and labeling checks, applied automatically to the next lots until the supplier demonstrates recovery. Reduced inspection rewards sustained excellence with smaller samples and lower cost, granted only after a defined run of consecutive clean lots and revocable at the first sign of regression.
The transitions must be contractual, not conversational. Write the trigger counts into the supplier agreement: how many failures tighten, how many clean lots at tightened level restore normal, and how many clean lots at normal earn reduced. Without numbers, factories negotiate every transition under delivery pressure, and buyers grant exceptions that the next failure makes regrettable.
A FBA compliance inspection booked under switching rules names its state explicitly in the scope so the factory knows which regime applies before the inspector arrives. Surprise tightening breeds resentment and disputes; announced tightening, applied evenly per the written rule, breeds improvement. Publish each supplier's current state quarterly alongside the scores that determine it.
Stratify lots that sampling tables assume uniform
Standard sampling tables assume a homogeneous lot, yet real FBA shipments are layered: multiple SKUs, several production dates, pallets packed at different times by different crews. Treating such a lot as uniform lets a concentrated defect pocket dilute into a passing average while still damaging every customer who receives from that pocket. Stratification corrects this by dividing the lot into meaningful groups and sampling each one.
Allocate sample units deliberately across strata rather than proportionally alone. Small but review-sensitive strata deserve minimum coverage regardless of their share: a fifty-unit variant in a five-thousand-unit lot still needs enough examined units to catch a systematic error like wrong inserts or misprinted labels. Document the allocation per stratum in the scope so coverage is verifiable rather than assumed.
Record results per stratum in the report, not just as lot totals. Stratum-level findings diagnose origins: defects concentrated in one production date point at a material batch or a shift change, while defects spread evenly point at design or process. This diagnostic value often exceeds the pass-or-fail verdict in long-term worth.
Switching and stratification quick reference
| Situation | Sampling response | Return condition |
|---|---|---|
| New supplier or new SKU | Normal level with full packaging and labeling checks | Three consecutive clean lots before any relaxation |
| One failed lot | Tightened level on next lots with expanded scope | Two consecutive clean lots at tightened level restore normal |
| Repeated failures | Tightened level plus per-stratum reporting and root-cause evidence | Corrective action verified on product, then clean-run counting restarts |
| Sustained clean history | Reduced sampling considered for mature SKUs only | Any failure or major change resets immediately to normal |
| Mixed or seasonal-risk lots | Stratified draws with minimum coverage per stratum | Program-level rule, reviewed each booking |
Re-inspect failures without negotiation theater
- Freeze the failed portionRequire physical segregation of affected units with confirmed counts before discussing remedies. Lots that remain co-mingled during negotiation have a habit of partially shipping themselves.
- Agree corrections in writingSpecify rework method, replacement quantities, responsible party, deadlines, and who pays for re-inspection. Verbal correction promises are the most common source of second failures.
- Re-sample at equal or stricter levelNever relax the plan to help a corrected lot pass. Re-inspection at the same level or tighter proves the fix; a smaller sample merely hides whether the fix worked.
- Verify on product with emphasisFocus re-inspection effort on the defect that failed, while retaining enough general coverage to catch damage introduced by rework itself. Rework handling creates fresh defects often enough to check for them.
- Close the file with evidenceFile the original report, the correction agreement, and the re-inspection verdict together. Lots without a closed file should not sail, because open failures compound with transit risks.
Laboratory testing and compliance boundaries
Visual sampling estimates defect rates; laboratory testing proves material properties. When a dispute concerns coating adhesion, colorfastness, restricted substances, or packaging strength, on-site opinion cannot settle it regardless of inspector experience. Retention samples go to an accredited facility whose technical competence is recognized through international mutual recognition arrangements, and the resulting numbers decide.[2] Budget testing triggers into the program for high-risk categories rather than discovering the need mid-dispute.
Compliance verification runs on its own track beside both sampling and testing. Technical regulations at destination can hold or refuse shipments that pass every AQL check, and these measures operate within international disciplines requiring transparency and notification.[1] The sampling program therefore ends at the factory gate verdict, while a separate compliance confirmation, markings, documents, and applicable test evidence, governs whether the lot should sail.
Sources
- WTO: Agreement on Technical Barriers to Trade —— WTO portal for the TBT Agreement covering technical regulations, standards and conformity assessment.
- ILAC: International Laboratory Accreditation Cooperation —— ILAC homepage describing the international cooperation on laboratory and inspection body accreditation.
Frequently asked questions
When should we switch to tightened inspection?
Immediately after a lot failure, and keep it tightened until the supplier produces the contractually defined run of consecutive clean lots under the stricter regime. Tightening means larger samples, stricter acceptance numbers, and expanded packaging and labeling checks. Write the trigger and the release condition into the supplier agreement so the transition is automatic rather than negotiated under delivery pressure.
Is reduced sampling ever safe for FBA goods?
Only for mature SKUs from proven suppliers with a sustained clean record, stable materials, and no recent design or packaging changes. Even then, keep compliance and variant checks intact and revoke reduced status at the first failure or significant change. Reduced sampling saves money on surveillance; it must never be used to make a troubled supplier's lots pass more easily.
How do we sample a lot with many small-quantity SKUs?
Stratify by SKU with guaranteed minimum coverage per variant, including ones that proportional allocation would nearly skip. Small SKUs carry full review risk per unit and often share systematic errors like wrong inserts across their entire quantity. Record and report results per SKU so a failing small variant cannot hide inside a passing lot average.
What is the difference between re-inspection and re-testing?
Re-inspection repeats the sampling examination on corrected or replacement goods to verify the fix worked, using the same or stricter plan. Re-testing sends retention or failed samples to a laboratory for property measurement when the dispute concerns materials rather than visible defects. Failed lots often need both: testing to establish the material facts and re-inspection to confirm the corrected lot meets the visual and functional criteria.
Take the next step in your practice.
Browse ritual supplies, spell kits and divination tools curated by our in-house coven — backed by a 90-day Right Craft guarantee.
Shop the Collection Book a Consultation