How Should Buyers Screen Customer Stock After a Complaint From a Dispensing Supplier Recovery Lot?
Direct answer: Screen customer stock only with a documented method shown capable of detecting the reported defect at the required sensitivity. First define the affected serial or lot population, quarantine it, agree the defect standard with the customer, qualify inspectors and equipment, and maintain unit-level pass/fail/rework traceability. Use 100% screening for detectable critical or major defects, but do not claim that visual inspection clears hidden voids, incomplete cure, poor adhesion or thermal-interface failure. Those risks need destructive, functional or validated non-destructive evidence.
Who this is for: customer quality, supplier quality, process engineers and operations teams containing a complaint linked to dispensing or potting recovery production. Buyer readiness: L5 field containment.
Screening is containment, not corrective action
Customer-site screening can separate known good and suspect units quickly. It can protect a production line while root-cause work continues. It does not prove the supplier process is corrected, and it does not close CAPA.
The first decision is whether screening is technically possible. If the defect is visible, repeatable and defined by a boundary sample, screening may work. If the failure sits inside cured potting material and no validated non-destructive method exists, inspecting every exterior surface still leaves the hidden risk untouched.
| Question | Evidence required | Stop condition |
|---|---|---|
| Can the defect be detected? | Known signature, method sensitivity and repeatability | Failure can pass the proposed test |
| Can the population be bounded? | Serial, lot, shipment and process genealogy | Affected units cannot be separated |
| Can results remain traceable? | Unit or container status and inspection record | Passed and uninspected stock can mix |
| Does the customer approve? | Written scope, criteria, access and disposition | No authority to alter customer stock |
Step 1: define the affected customer population
Map the complaint serial to recovery supplier lot, finished production order, machine, recipe, material, shift and shipment. Expand the boundary to all units sharing plausible causal inputs. Then reconcile customer quantities: received, unopened, line-side, work in process, assembled, tested, shipped onward and installed.
- Identify every customer site and intermediate warehouse.
- Separate mixed pallets and split shipments.
- Include replacement and reworked serials.
- Track units already consumed into higher assemblies.
- Record what cannot be located rather than treating it as zero exposure.
Start broad when genealogy is uncertain. Narrowing a hold requires evidence; expanding later after an escape is more disruptive.
Step 2: choose a method that matches the defect
| Defect | Possible screen | Evidence of capability | Important limit |
|---|---|---|---|
| Missing or misplaced bead | Visual/image inspection with defined view and lighting | Known defect set, borderline examples and repeatability | Cannot prove buried bead performance |
| Overflow/contamination | Visual or dimensional fixture | Acceptance standard and inspector agreement | Residue may be hidden under assembly |
| Seal discontinuity | Validated leak or pressure-decay test | Known leak standard and gauge validation | Test conditions must represent requirement |
| Incomplete cure | Approved hardness/functional test or representative specimen | Correlation to cured property | Surface tack alone may mislead |
| Internal voids | X-ray/ultrasound or other validated NDT where suitable | Detection limit using known references | Geometry/material may hide small critical voids |
| Poor adhesion | Validated proof or destructive sample plan | Correlation to required bond strength/failure mode | 100% destructive screening is usually impractical |
ISO 28594 notes that its accept-zero sampling procedures are not intended where destructive testing or product screening is infeasible. That matters here. When no reliable screen exists, use broader containment, representative destructive testing, replacement or an authorized field-action decision.
Step 3: validate the screen before touching customer inventory
Run a short qualification with known good, known bad and borderline units. Blind the status when possible. Record false accepts, false rejects, inspection time and disagreements between inspectors. Challenge normal variation in lighting, fixture, orientation and operator.
| Validation item | Buyer question | Required output |
|---|---|---|
| Detection sensitivity | What smallest or weakest defect is reliably detected? | Documented detection boundary |
| False acceptance | Can defective units pass? | Risk estimate and added safeguard |
| False rejection | Will good stock be unnecessarily blocked? | Escalation/review rule |
| Repeatability | Does one inspector get the same result twice? | Repeated trial record |
| Reproducibility | Do different inspectors agree? | Agreement evidence and training action |
| Throughput | Can the method keep pace without fatigue shortcuts? | Realistic hourly capacity and staffing |
Step 4: control the screening station
Use physically separated incoming-hold, inspection, pass, fail and review areas. Give each unit one-way status movement. Where labels can fall off or packaging changes, link status electronically to serial or container ID.
- Verify identity before inspection.
- Inspect using the controlled instruction and calibrated equipment.
- Record inspector, time, result and defect code.
- Move passes into a sealed, clearly identified released area.
- Move failures into locked quarantine; photograph representative evidence.
- Send ambiguous results to engineering review, never the pass bin.
- Reconcile starting quantity with pass, fail, review and missing quantities.
Supervisors should perform layered audits during the job. Inspector fatigue is a real risk when hundreds of visually similar units are screened under time pressure.
Application scenario matrix
| Application | Complaint | Potential field screen | When replacement is safer |
|---|---|---|---|
| EV battery potting | Void, cure, insulation or thermal concern | Only validated NDT/electrical method suitable for geometry | Hidden risk cannot be reliably detected |
| PCB dispensing | Missing dot, overflow or contamination | Controlled image/visual inspection and functional test | Deposit is inaccessible after assembly |
| LED driver potting | Underfill, connector contamination or cure issue | Fill/visual plus defined electrical or cure evidence | Internal cure/void status remains unknown |
| Automotive sensor sealing | Leak or discontinuous bead | Validated leak/seal test | Field screen cannot reproduce required sealing condition |
| Power electronics TIM | Thermal rise or poor coverage | Validated thermal/functional test if sensitive enough | Bondline cannot be assessed without disassembly |
Step 5: define disposition before results arrive
| Result | Disposition | Required record |
|---|---|---|
| Pass | Release only for agreed use/status | Unit/container ID, result and authorization |
| Fail | Quarantine, return, replace, repair or scrap per agreement | Defect code, evidence and quantity |
| Ambiguous | Hold for engineering/customer review | Observed condition and review decision |
| New defect | Stop screening and reassess method/scope | New symptom and affected window |
| Repeated original defect | Escalate stop-ship/CAPA/revalidation | Trend by serial, lot and location |
Do not rework customer product without approved instructions. Successful rework must be reinspected and remain linked to its original identity.
How much stock should be screened?
For a bounded, detectable defect, screen 100% of the potentially affected customer population. Statistical sampling can help estimate prevalence or audit the screening process, but a passing sample does not clear all units when the containment decision requires every suspect unit to be separated.
ISO 2859-1:2026 provides lot-by-lot attribute sampling and switching rules. Use it where appropriate to the contract and inspection purpose. Do not confuse acceptance sampling with sorting. One decides about a lot under stated risk; the other attempts to classify each unit.
Screening report and exit rules
Report total quantity, inspected quantity, passes, failures, ambiguous units, missing units, defect locations, hourly rate, auditors and deviations from instruction. Provide the customer with a reconciliation, not only a pass percentage.
- All defined customer locations and quantities are reconciled.
- No uninspected affected stock remains available for use.
- Passed stock is positively identified.
- Failed stock has an authorized disposition.
- Results are analysed by lot, serial and process window.
- Supplier containment expands if the original defect is found.
- Corrective action and revalidation remain active beyond sorting.
- The customer accepts closure of the screening activity.
Evidence sources
ISO 28594:2017 covers accept-zero sampling and notes limitations where destructive testing or screening is infeasible. The ISO 2859-1:2026 overview describes lot-by-lot attribute sampling and switching rules. ISO 10002:2018 provides complaints-handling guidance. Apply purchased standards, customer requirements and applicable law.
Frequently asked questions
Does 100% visual inspection guarantee defect-free stock?
No. It only addresses defects the visual method can reliably detect, and inspectors can still make errors.
Can sampling replace customer stock screening?
Not when every potentially affected unit must be classified. Sampling may support lot decisions or audits but does not inspect every unit.
Who pays for customer-site screening?
Commercial responsibility follows the contract and agreed cause; record labor, travel, downtime and disposition costs accurately.
Can temporary workers perform screening?
Only after qualification to the controlled method with supervision and audit checks.
What happens if a new defect appears?
Stop, quarantine recent output, reassess scope and method, and escalate the complaint investigation.
What should buyers send OBO Precision?
Send defect photos, serial/lot scope, material data, recipe, customer test method and available failed samples.
When suspect units may already be outside direct control, use this guide to decide whether customer notification or field containment is required after a dispensing complaint.
Closure follow-up: use documented field containment exit criteria after a dispensing complaint before removing customer screening or heightened controls.
Design a screen that can actually find the defect
OBO Precision can review the process, complaint evidence, material and product geometry, then recommend practical verification and containment options.
