Sample approval after quality issues: records and retesting

Write measurable acceptance criteria before the sample

Identify supplier, SKU, drawing/specification revision, material claim, sample type and intended production process. State whether the sample is stock, a prototype or made with the intended bulk method. Agree the characteristic, unit, target, tolerance, check point, method, conditions and decision rule. “Good quality” and “same as photo” are not measurable criteria. Ask for sample, remake and courier prices separately.

ISO distinguishes inspection and testing. A visual check or a seller's certificate name does not establish every function, material or regulatory requirement. The worksheet below is an editorial buying tool, not an AQL plan, accredited inspection or laboratory approval. Instrument suitability and uncertainty must be considered for real acceptance; a result near a limit may need a better method rather than a quick pass.

Use severity labels only within an agreed scope

• Critical: a suspected hazard or a missing mandatory safety/compliance check. Stop the affected decision and seek the appropriate qualified assessment; a photo does not settle it.
• Major: a discrepancy that prevents an agreed fit, use or material requirement, such as a divider that cannot fit its slot. Hold the affected sample approval.
• Cosmetic: an appearance issue without an identified fit/function effect in the stated check. It can still fail an agreed appearance criterion. Any exception needs a written scope and buyer decision.

These are buyer-defined triage labels for discussion, not universal defect classifications or acceptable batch percentages. They do not authorize an inspector, certify safety or establish a sampling plan.

Sample issue disposition

Sample issue disposition

Issue	Evidence	Action
Wrong agreed variant	Black required / blue received, assumed	Hold and request matching sample
Undefined cosmetic criterion	No agreed reference or limit	Define criteria before approval
Function failed	Method, conditions and recorded result	Correct and repeat affected checks

View full diagram

Copyable discrepancy and correction record

Supplier / SKU / sample ID / sample type: [ ]
Specification and packaging revisions / received date: [ ]
Issue ID and agreed requirement with unit/tolerance: [ ]
Actual readings / check point / method / conditions: [ ]
Photo/video IDs: overview, close-up, label and measuring setup [ ]
Assigned severity within agreed scope / reason: [ ]
Decision: hold / request clarification / rework / decline [ ]
Supplier proposed cause and correction / owner / readiness date: [ ]
Cost responsibility: remake [ ], test [ ], courier [ ], duties [ ]
Replacement ID / new revision / changes and unchanged criteria: [ ]
Retest readings and evidence / unresolved requirements: [ ]
Buyer decision and exact approval scope / date / exceptions: [ ]
Bulk checks, packaging and nonconforming-goods handling still required: [ ]

Keep the failed sample and packaging linked to its issue record. Photograph the whole sample, its ID and the measurement setup as well as the defect. Use comparable lighting and position; a close-up without scale or identity is weak evidence. Photographs support the observation, not every technical or legal conclusion.

Completed hypothetical failed-sample case

Every reading, response, cost and decision here is invented. This is not a real inspection or JTLGO case. An unpowered desk organizer, sample S1, is compared with drawing R1. Buyer-defined criteria are length 200.0 ± 2.0 mm (198.0–202.0), divider-slot width 10.0 ± 0.2 mm (9.8–10.2), reference divider insertion without forced bending, and no front-face scratch longer than 1.0 mm. Dimensions use the same recorded points and a suitable measuring method; appearance uses the same white desk lamp at 30 cm. These conditions and tolerances are invented for this product, not an industry standard.

For a symmetric size tolerance, deviation Δ = measured value − target; the arithmetic criterion is |Δ| ≤ tolerance. S1 length mean = (204.0 + 203.8 + 204.2) ÷ 3 = 204.0 mm; Δ = 204.0 − 200.0 = +4.0 mm. Since 4.0 > 2.0, it fails. The slot's |9.5 − 10.0| = 0.5 mm also exceeds 0.2. Repeated readings describe this measurement, not three independent samples or the batch defect rate. Each S1 length reading is outside the allowed interval, so the mean is not being used to hide an individual failure.

Record D1 length with photos P1 overview/ID and P2 measuring setup; D2 slot/failed insertion with P3 and video V1; D3 scratch with P4 showing scale and front-face position. Label D1/D2 major within this buyer's fit scope and D3 cosmetic but nonconforming. Decision: HOLD S1 and bulk approval. No real safety or material test has been completed by this fictional exercise.

Rework, cost responsibility and S2 retest

In the invented reply, the supplier proposes correcting the manufacturing setup, supplying S2 under drawing R2, and removing the appearance defect. R2 documents the correction; it does not widen R1 tolerances. The imagined agreement assigns USD 20 remake to the supplier, USD 18 courier and USD 12 defined recheck to the buyer: additional total = 20 + 18 + 12 = USD 50, buyer cash = USD 30. Actual duty or other unquoted fees remain TBC. The proposed schedule is four working days for remake, three for courier and one for recheck: 4 + 3 + 1 = eight working days in this fictional plan, not a delivery guarantee.

Trace the approval version

Trace the approval version

Record	Release condition
S1 + specification A	Discrepancy recorded; not approved
S2 + confirmed revision	Changed details and checks reviewed
Bulk specification + packing	Separate inspection and remedy agreement

View full diagram

S2 length mean = (200.3 + 200.4 + 200.5) ÷ 3 = 200.4 mm; |200.4 − 200.0| = 0.4 ≤ 2.0. Slot width 9.9 gives |9.9 − 10.0| = 0.1 ≤ 0.2; the reference divider inserts and the front-face mark is 0.5 ≤ 1.0 mm. In this invented scenario the buyer approves S2 only for these stated size, fit and appearance checks. Bulk release remains HOLD until the material record, packaging and separately agreed batch checks are complete. If any required result is missing, a disputed measurement remains unresolved or a criterion still fails, keep approval on hold; do not quietly change the tolerance to accept it.

Approve an identified version, not “all future production”

Retain S1/R1, the discrepancy evidence, supplier correction and cost agreement, then link S2/R2, retest readings, explicit scope and any agreed exception. Sample approval is not product certification or a guarantee for the batch. Agree bulk inspection, production-change control and nonconforming-goods handling separately. This tutorial establishes no automatic free remake, refund or cancellation right; use the actual order terms and applicable platform process.

Send the specification and discrepancy record through JTLGO sourcing, with the inspection or laboratory scope stated explicitly. If you have not yet agreed a specification, start with the RFQ template; for a material or category-specific check, use the relevant product checklist rather than borrowing these desk-organizer tolerances.

Worksheets and evidence status

Download the text worksheet · Download the editable tables (CSV)

Real redacted order materials and final human review are still pending. Examples and worksheets explain the method; they are not completed customer cases.