Follow-up converts evaluation findings into verified corrections: prioritize the list by system scope, assign every item an owner and a date, verify each fix against the original criteria, and lock the gains into documents and training. Without this discipline, findings drift and recur; with it, each evaluation leaves the supplier measurably stronger. This fourth part of the follow-up series gives buyers the full sequence from finding list to sustained improvement.
Key takeaways
- Prioritize shared-system findings first, then family-specific items ordered by shipment risk.
- Structure every correction as finding, action, owner, date, evidence, and verification.
- Verify corrections against original criteria with evidence gathered where fixing is still possible.
- Lock gains into documents, training, and routine checks so they survive staff changes.
- Run follow-up on closed-loop quality principles that link correction to verified effectiveness. [2]
Findings without follow-up are just observations. The value of a factory evaluation is realized in the weeks after the visit, when each finding is converted into an owned correction with a date, evidence, and verification. This fourth part of the follow-up series covers that conversion end to end: prioritizing the list, structuring corrections, verifying them properly, and locking the gains into the management system so they survive staff turnover.
The failure mode is familiar to every sourcing professional. The report lists fourteen findings, the factory replies with general assurances, months pass, and the next evaluation finds eleven of the fourteen still open. Nobody owned them, nobody dated them, and nobody defined what evidence would close them. Follow-up discipline exists to prevent exactly this drift.
Two bodies of established practice guide the work. Quality management foundations describe the closed loop from policy through operation to verified correction, and the industrial habit of proving conformance at the factory before acceptance models how verification should be staged. Foundational references on quality management lay out the former. [2] Public references on factory acceptance testing describe the latter. [1] Together they turn follow-up from nagging into method.
Prioritize by system scope, not by page order
Sort the finding list into three tiers. Tier one holds shared-system gaps: missing incoming control, absent document control, no management review, untrained key posts. These poison every line and must be fixed first, because family-level corrections built on a broken system will not hold. Tier two holds family-specific findings ordered by the shipment value and risk of the affected products. Tier three holds observations worth monitoring but not worth delaying shipments over.
Assign each tier-one and tier-two item a single named owner on the factory side and a calendar date, and publish the matrix to both sides. Shared ownership means no ownership, and dateless commitments mean eventual commitments. The matrix is short enough to review on a weekly call: what closed, what slipped, what evidence arrived. Buyers who run this review visibly get faster closure than buyers who file the report and wait.
Buyers coordinating independent follow-up often anchor the schedule to the original UTS supplier verification UTS supplier verification visit, because the finding numbers, scope boundaries, and evidence expectations carry over directly. Continuity between the evaluation and its follow-up keeps both sides arguing about evidence rather than re-litigating what was found.
Structure every correction as a chain
Each finding becomes a six-link chain: the finding number and text, the agreed corrective action, the named owner, the due date, the evidence that the action happened, and the verification result. Publish the chain in a shared tracker so status is visible without chasing emails. Any chain missing its last two links is activity, not closure, and the tracker should label it honestly as pending.
The quality management logic behind chaining is the closed loop: corrections must address causes, and effectiveness must be checked, or the same finding recurs next cycle. Foundational references on quality management describe this expectation that systems learn from their failures. [2] Factories that internalize the loop start writing better root-cause notes unprompted, because they know the verification step will test whether the cause was really addressed.
Judge evidence against the finding, not against effort. A retraining program with attendance sheets answers a competence finding; the same attendance sheets do not answer a missing-gauge finding. Require the factory to state, per item, which evidence demonstrates which correction, and reject submissions where the link is asserted rather than shown. Precision here saves a wasted verification visit later.
Correction tracker with worked example
| Finding | Correction | Evidence required |
|---|---|---|
| F-03 incoming control absent | Write procedure, assign inspectors, start logging | Approved procedure plus four weeks of receiving logs |
| F-07 gauge uncalibrated | Quarantine gauge, calibrate, label all gauges | Certificates plus dated photos of labels |
| F-09 training gaps | Retrain three operators, check competence | Attendance sheets plus interview pass on re-visit |
| F-11 rework uncontrolled | Number rework orders, record dispositions | Sequential rework log with dispositions |
| F-12 packing errors | Add barcode scan step, log scans | Scan logs matched to shipment cartons |
| F-14 review minutes missing | Hold review, record decisions and dates | Signed minutes with dated actions |
Verify the way acceptance is verified
- Test against original criteriaRe-examine the corrected records, gauges, or lots against the same criteria and limits the original finding cited.
- Verify on the floorCheck the fix where it operates: the line, the gauge rack, the packing station, not just the conference room.
- Confirm routine operationTrace the correction into routine: new log entries after the fix date, new labels on all gauges, new scans on shipments.
- Interview the peopleInterview the retrained or reassigned staff to confirm competence, not just attendance.
- Close the loop in writingRecord the verification result against the finding number so the lot or system history reads as one story.
Make verification meaningful with acceptance logic
Good verification borrows the logic of acceptance testing: demonstrate, with evidence gathered where correction is still possible, that the corrected item now conforms. Industrial practice proves equipment conformance at the factory before acceptance rather than after delivery, as described in public references on factory acceptance testing. [1] Re-checks of corrected lots, recalibrated gauges, or rewritten procedures follow the same principle: prove the fix before relying on it.
Stage verification promptly. Evidence decays: operators forget retraining, temporary labels fall off, logs started with enthusiasm thin out. A verification visit four weeks after correction tests whether the fix took root; a verification visit four months later tests archaeology. Agree the verification date when the correction is agreed, not after the factory declares itself ready.
Treat failed verifications as information, not betrayal. A correction that did not hold reveals that the root cause was misdiagnosed, and the second attempt can target the real cause. What must not happen is silent closure: every finding ends either verified-closed with evidence or escalated with a new plan. The tracker admits no third state.
Follow-up is where evaluations earn their fee. Findings are cheap to write and expensive to fix, and the buyer's discipline in the weeks after the visit decides which side of that equation the supplier lands on.
Sources
- Wikipedia: Factory acceptance test —— Overview of factory acceptance testing performed before shipment or installation.
- Wikipedia: ISO 9000 —— Overview of the ISO 9000 family of quality management standards.
Frequently asked questions
How should buyers prioritize a long finding list?
Rank by scope: shared-system gaps first, then family-specific findings ordered by shipment risk, then observations that need only monitoring. Assign each a named owner and a date, and sequence verification so systemic fixes are confirmed before their dependent lines are re-checked. Everything else can wait its turn.
Why do auditors insist on training records behind new procedures?
Because a procedure nobody was trained on will not be followed, and training nobody checks will evaporate within weeks. The chain from document to competence to verified practice is what closes the loop, and quality management foundations describe exactly this closed-loop expectation. Follow-up visits should test the last link, not just admire the first.
What does good verification of a correction look like?
Confirm the corrected goods or records against the same criteria the original finding used, with evidence gathered where correction is still possible. Acceptance testing of equipment at the factory follows this same prove-it-before-accepting logic, and re-checks of corrected lots or systems borrow the principle directly.
How do surveillance visits confirm that fixes held?
They test whether yesterday's fixes survived: re-pull a sample of previously closed findings, interview the retrained staff, and observe whether the new practice runs without prompting. Fixes that held prove the system learns; fixes that evaporated reopen the original finding with a stronger recommendation.