Key takeaways
- Keep eligibility gates separate from weighted performance.
- Freeze KPI definitions before comparing suppliers.
- Every red metric needs an owner, evidence link, due date and closure test.
Prequalification and performance are two different tools
Prequalification asks whether a supplier is eligible for a particular program. It checks legal identity, scope, product capability, required audits, material traceability and commercial fit before an order is awarded. Ongoing evaluation asks how the supplier performed against approved orders, specifications and dates. Mixing the two produces a misleading score.
Create a hard-stop gate for requirements that cannot be averaged away. An expired mandatory document or unapproved material substitution should trigger review even when delivery and service scores are high. SAP’s supplier-evaluation documentation similarly treats evaluation as structured criteria rather than an informal impression.
Write the metric dictionary before the scorecard
For each KPI, record its formula, numerator, denominator, data source, period, owner, exclusions and escalation rule. “Quality: 92” is not auditable. “Rejected inspected units divided by inspected units for final inspections closed in the quarter” can be checked. If rework units, sample failures and customer returns belong in different datasets, keep them separate.
The metric dictionary also prevents a supplier from being penalized for buyer-caused holds. Delivery calculations should distinguish the contractual due date, approved revised due date and actual receipt or shipment milestone. Freeze the milestone in the purchase order rather than choosing whichever date supports the desired narrative.
Quality KPIs need denominators and defect ownership
For this buyer-defined scorecard, defect rate = rejected inspected units ÷ inspected units × 100. This is an explicit convention, not a universal standard. It is usable only when the inspection stage, sampling plan and rejection rule are stated. Track critical incidents separately because a rare safety or compliance failure should not disappear inside an average. Add corrective-action recurrence to show whether the cause was removed.
For belts, useful evidence may include incoming material records, inline findings, final AQL results, measurement checks, plating or colorfastness reports, approved samples and customer claims. Do not collapse a laboratory test failure and a visual workmanship defect into the same unlabeled count.
Delivery KPIs must identify the promised event
On-time delivery = purchase orders meeting the agreed milestone ÷ purchase orders due in the period × 100. This follows the denominator logic in the JSI procurement-indicators guide. State whether the event is ex-factory, port handover, vessel departure or receipt. Freight delays after the supplier-controlled milestone should be visible, but not silently assigned to factory performance.
Pair the percentage with late-order count, days late and reason ownership. A supplier with one severe delay can look identical to one with several one-day misses if only an average is shown. Keep buyer approval delays, material holds and supplier production misses as separate reason codes.
Service KPIs should measure closure, not politeness
Response speed is easy to count and easy to game. Stronger service measures include clarification quality, first-complete document submission, sample revision traceability and corrective-action closure. Closure lead time is the calendar difference between the accepted issue-open date and accepted close date; define pauses and the acceptance owner.
Add evidence links instead of narrative praise. A good service score can point to a complete change log, correct packing-list revision or root-cause package. A poor score should cite the missing or repeated item. This turns the review into a working improvement list.
Treat compliance as evidence status plus risk
List each required record, scope, issuing body, expiry date, facility name and verification owner. Mark it valid, missing, expired, out of scope or under review. Do not award partial points for a document that does not cover the factory or material in question.
Social-audit, chemical, origin and material claims belong to different evidence chains. The scorecard should link to the applicable record, not infer one claim from another. A Sedex membership, for example, is not the same thing as a specific audit result; keep labels precise.
Stop a weighted score from hiding a hard failure
Weights reflect business priorities, but they do not remove decision gates. A supplier might achieve a strong composite score while carrying an unresolved critical defect or missing mandatory evidence. Put hard-stop conditions above the calculation and show sub-scores beside the total.
| Layer | Example evidence | Decision use |
|---|---|---|
| Eligibility | Required document is valid and in scope | Pass, hold or reject |
| Performance | Defined quality and delivery KPIs | Trend and compare |
| Incident | Critical defect or unauthorized change | Escalate regardless of total |
| Improvement | Corrective action with due date | Track closure |
Use the symbolic calculation Composite = Σ(normalized KPI result × approved weight). Then apply the override: if any mandatory gate is fail, the sourcing decision is HOLD or ESCALATE regardless of Composite. This makes the arithmetic reproducible without inventing target scores. Record who approved each weight and when it changed.
Use one cadence for data and another for decisions
Capture transactional evidence as orders, inspections and actions close. Review operational exceptions at a program-set interval, then conduct a broader stage or periodic review. The CRS procurement guide supports the wider monitoring, feedback and action-plan discipline; the exact cadence remains a program decision. An annual review can reset weights and sourcing strategy, but it should not be the first time a serious problem is surfaced.
The review chair should confirm data completeness, challenge exclusions and record the action. Suppliers need the same metric definitions and evidence window used by buyers. That transparency makes improvement possible and reduces arguments over arithmetic.
Copy this scorecard structure
| Gate / KPI | Definition or formula | Evidence source | Result / weight | Gate status | Owner / due / closure |
|---|---|---|---|---|---|
| Mandatory facility or material evidence | Valid, in scope and current | Linked record | N/A | Pass / hold / fail | Compliance owner |
| Quality: buyer-defined defect rate | Rejected inspected units ÷ inspected units × 100 | Final inspection dataset | ___ / ___ | Trend; critical incident overrides | Quality owner |
| Delivery: OTD | Orders meeting agreed milestone ÷ orders due × 100 | PO and milestone record | ___ / ___ | Trend | Supply owner |
| Service: action closure | Accepted closed actions ÷ actions due | Corrective-action log | ___ / ___ | Reopen if effectiveness fails | Named action owner |
| Compliance status | Valid / missing / expired / out of scope | Verification register | Not weighted if mandatory | Pass / hold / fail | Compliance owner |
| Composite | Σ(normalized KPI × approved weight) | Locked calculation sheet | ___ | Overridden by any mandatory fail | Review chair |
Do not publish invented target percentages. Set targets from the program’s contract, risk class and historical baseline, then version-control the change. HongDing’s documented workflow can contribute tech pack or BOM control, approved samples, material and hardware records, QC evidence, packaging evidence and coordinated change records; it does not supply performance results until a real program creates them.
Turn the scorecard into a sourcing action
The final meeting should produce one of four outcomes: continue, continue with corrective actions, reduce exposure, or pause qualification. A mandatory gate failure routes to hold or escalation. A missed action closes only after the evidence owner accepts completion and an effectiveness check passes; recurrence reopens it. Record the rationale, approver and next evidence date. A score with no action owner is reporting, not supplier management.
HongDing’s quality-control pathway can anchor a program-specific evidence discussion. Use the related guides on manufacturer evaluation, quote comparison and AQL inspection for the upstream and inspection details.
Verified source notes
Frequently asked questions
What KPIs belong in a belt supplier scorecard?
Use defined quality, delivery, service and compliance evidence. Add cost or innovation only when the data definition and decision use are clear.
Should compliance receive a weight?
Mandatory compliance is better handled as an eligibility or hard-stop gate. A weighted score can otherwise hide an expired or out-of-scope requirement.
How often should suppliers be reviewed?
Capture evidence as transactions close, review exceptions on a short operating cadence, and run broader stage, quarterly or annual decisions based on program risk.
Can I use the same targets for every supplier?
Not automatically. Targets should come from the contract, product risk, agreed milestone and approved baseline, not an invented universal benchmark.
