One Failed Unit. How Far Should Containment Go?
Containment should follow shared exposure evidence across component lots, machines, programmes, time windows, batches and shipments—not fear or convenience.
The short answer
Quality containment should cover the smallest population that evidence can defend—not the smallest convenient group or the entire production history by default. Electronics manufacturers define scope by shared exposure: component lot, product revision, process, machine, programme, fixture, operator, time window, work order, serial range and shipment.
Connected genealogy turns one failed unit into a controlled investigation rather than a factory-wide guess.
One failure changes the room
A final-test unit fails unexpectedly. Production wants to continue because hundreds passed. Quality wants to hold everything. Sales asks whether today’s shipment is safe.
All three teams need the same answer: which other units could share the cause?
Containment is a risk decision made under uncertainty. Speed matters, but unsupported confidence and unsupported fear both create cost.
Protect first, refine second
Apply an immediate conservative hold when severity or uncertainty demands it. Secure work in progress, finished goods, warehouse stock and relevant shipments before debating the perfect boundary.
Record the trigger, initial scope, time, owner and physical locations. Make digital status prevent issue or shipment, and ensure physical identification matches.
Then refine the population as evidence improves. The initial hold and later release decisions should remain in history.
Build a hypothesis around shared exposure
Start with the failure mode and likely creation points. A component failure suggests manufacturer part and lot. A solder defect may suggest line, programme, material, stencil, shift or time. A test escape may suggest programme and fixture.
Do not choose scope from the work order simply because it is convenient. The true exposure may cross orders or affect only part of one order.
List hypotheses and the evidence that would include or exclude units.
Use product genealogy
Genealogy should connect actual material consumption and process events to batch or serial identity at the required granularity.
For a suspect component lot, find every receiving lot, split, transfer, issue, return, work order, serial and shipment. For a machine event, identify units processed between last known good and confirmed recovery.
Missing genealogy widens containment because uncertainty becomes part of the risk.
Define time windows carefully
The failure discovery time is not necessarily the creation time. Use equipment logs, material issue, programme change, maintenance, verification and process history to establish last known good and first confirmed good boundaries.
Avoid arbitrary shift or calendar boundaries unless the process evidence supports them. A reel or fixture can cross shifts; a programme can change mid-order.
Record why the selected window is defensible.
Reconcile quantity and location
Containment must account for planned, completed, scrapped, reworked, held, released and shipped quantities.
Search every relevant physical state: line, buffer, inspection, rework, warehouse, subcontractor, transit and customer location. Avoid double counting split or transferred units.
An unexplained quantity difference should remain an exception, not be rounded away to close the exercise.
Decide inspection strategy
Additional inspection is valuable only if the method can detect the suspected condition with appropriate confidence. A visual check cannot contain an intermittent electrical failure merely because it is easy to deploy.
Define method, equipment, sampling or full-screening rule, acceptance, result capture and reaction to another failure. Control inspectors’ instructions and training.
If detection capability is uncertain, maintain the hold while engineering evaluates alternatives.
Manage disposition by identity
Affected units may be reworked, retested, used under concession, returned, scrapped or released after evidence review. Link disposition to exact serial, batch or quantity and authorised decision.
Do not release a location or work order generically if mixed status exists. The system should prevent held and released quantities from becoming indistinguishable.
Preserve rework and retest history when product returns to normal flow.
Extend containment beyond the site
Check subcontractors, distribution, service stock and customer shipments. External communication should identify product, quantity, serial or lot range, required action and response evidence.
Share enough information for effective containment while protecting proprietary details. Record acknowledgement and reconciliation.
NIST traceability principles emphasise linked provenance across organisational boundaries; the same logic makes external containment practical.
Update the customer promise
Containment can change releasable quantity and shipment timing. Customer teams need current evidence, recovery scenarios and authorised communication.
Avoid declaring the entire order delayed if safe partial quantity exists. Equally, do not protect a date by releasing units whose exposure remains unresolved.
Connect disposition to order status so the promise reflects quality reality.
Move from containment to cause
Containment protects the customer; it does not correct the process. Begin root-cause work in parallel without weakening the immediate hold.
Separate occurrence cause, escape cause and systemic cause. Corrective action should prevent creation, improve detection and strengthen the control that allowed uncertainty.
Verify effectiveness through later production evidence, not meeting closure alone.
Measure precision and response
Track time to initial hold, time to reconciled population, contained quantity, confirmed affected quantity, inspection effort, escapes and release time.
A very broad ratio of contained to affected may indicate weak genealogy or detection. It may still be appropriate for severe risk, so interpret the measure with context.
Review recurring scope gaps by material, process and system relationship.
Connect containment to the quality system
The Product Passed Every Gate. Why Did the Customer Still Find the Failure? describes the full quality thread. Containment tests whether that thread can support action under pressure.
Equipment impact is explored in The Gauge Was Calibrated. Was It Valid for This Run?, while supplier evidence can narrow material exposure.
Run a tabletop exercise
Select a plausible high-impact failure. Give the team one serial number and ask them to identify all potentially exposed product within thirty minutes.
Trace material, process, equipment, test, rework, stock and shipment. Record every missing link and manual inference.
Repeat after improvements and compare speed, scope and confidence.
Control suspect material during investigation
Suspect components may exist as unopened reels, partial reels, staged kits, floor stock and material already consumed. Apply status at each level without losing quantity reconciliation.
Returns from the line should preserve lot identity. Do not merge suspect and cleared balances while investigation is open. If sampling releases part of a lot, record the exact quantity and method.
Use decision checkpoints
Containment evolves as evidence arrives. Set review points for initial protection, population reconciliation, inspection outcome, cause confirmation and final release.
At each checkpoint, record what changed and why. This prevents yesterday’s conservative assumption from being mistaken for today’s confirmed scope, or an early optimistic hypothesis from weakening protection.
Plan manpower and flow
Screening thousands of units can overwhelm inspection and create new handling damage. Estimate equipment, people, space and time before promising recovery.
Prioritise by customer need and risk while preserving identity. Temporary inspection areas need controlled input, result, accepted output, rejected output and remaining balance.
Learn from near escapes
Review cases where genealogy narrowly prevented broad recall, and cases where missing data forced excessive containment. Both reveal the business value of traceability.
Update material capture, equipment-use records, programme deployment or shipment linkage based on the specific gap. The containment report should improve the operating system, not disappear after release.
Frequently asked questions
Should one failure always stop production?
No. Severity, recurrence, detectability and shared exposure determine response. High uncertainty may justify a temporary stop while scope is established.
Is the work order the normal containment boundary?
It is a useful commercial boundary, but technical exposure can cross work orders or affect only a subset. Follow the suspected cause and genealogy.
When can held product be released?
After authorised review shows it is outside exposure, passes a capable containment method or receives approved disposition.
Must customers be notified immediately?
Follow contractual, regulatory and risk requirements. Communication should be timely, authorised and based on the clearest available scope.
Does better traceability reduce quality risk?
It does not prevent every defect, but it reduces uncertainty, speeds containment and supports more precise response and learning.
Let evidence draw the boundary
Containment protects customers and buys time for understanding. Its quality depends on the evidence used to define scope.
Connect material, process, equipment and shipment history. Then the factory can act quickly, narrow responsibly and release confidently.
The team should also record what evidence was unavailable at the first decision and when it became available. This timeline distinguishes a reasonable conservative hold from avoidable delay. It shows which integrations, scans or ownership changes would have improved the response. Over time, containment should become faster and more precise because each event strengthens the next investigation.