Supplier scores look fine but nobody exits? Performance must feed share allocation

The radar chart looks good, but I don't dare to cut the orders.

At the quarterly supplier meeting, the projector showed four or five dimensions forming a beautiful polygon. Privately, the purchasing manager said: the ones truly brave enough to increase orders are still those three 'familiar' suppliers; the one with the most problems, because 'the line change cost is high' and 'the boss knows them,' still holds the main share. The SQE opens the overdue 8D list, and the planner opens the details of consistently late OTIF—three tables talking about the same supplier, yet the decision stops at 'let's keep observing.'

The list of qualified suppliers is getting longer and fluctuating continuously, which usually is not because of 'insufficient indicators,' but becauseThe assessment results cannot change the range and share available for placing ordersIf performance is only ranked without making decisions, the system is nothing more than an electronic version of the year-end ceremony.

First ask which decision the performance is supposed to serve

Before implementing the scoring online, first clearly write down the actions that performance needs to drive:

If the organization only wants a 'score for the supplier to see,' there is no need for a complex model; if it wants to control directory inflation, the score must be able to reflect the cooperation status.

Sub-items and red lines: the total score is a summary, not a cover-up

It is recommended to retain at least four subcategories and allow adjustment of category weights:

• Quality: Incoming material nonconformance rate, major defects, 8D repetition and overdue;
• Delivery: OTIF or line-level delivery compliance, delivery deviation (consistent with the scope of delivery topic);
• CollaborationOrder confirmation timeliness rate, secondary confirmation of changes, portal response;
• ComplianceCertificates valid, agreement and factory inspection conclusion, risk labels.

Red line examples (after being categorized and written into the plan attachment): Major quality/safety incidents → supply suspension assessment; license expiration → freeze new orders; repeated overdue similar 8D → enforce quantity limits; OTIF continuously below threshold → delivery rectification and set quota limit.

Business points and relationship points can exist, butCannot erase the red lineThe dashboard by default displays sub-items and overdue queues, with the total score only serving as a summary. Procurement carries the share, SQE carries the 8D, and planning carries the material shortages — the regular meeting should address the same supplier, not each going through their own sheet.

Rectification must approach procurement visibility from the quality side.

Typical symptoms of the disconnect between performance and quality: 8D goes unused in the SQE folder, and the procurement quarterly rating is still given as 'Good.' We need to integrate the event flow into the evaluation:

1. Incoming materials are non-conforming, deviations, 8D with deadlines and responsible parties;
2. Automatic upgrade upon overdue, procurement and SQE see the same disposal view;
3. Closing requires validity verification (small batch stability, reduction in repeated defects, etc.) and prohibits 'closing immediately after filling out the form'.
4. Overdue and repeated openings are counted in process performance and trigger limit rules.

Access control addresses 'whether one can enter,' while performance addresses 'the share and retention after entering.' Observation period indicators should be of the same source as the above dimensions to avoid having one set for newcomers and another set for veterans. Attention can be given to deduction or payment configuration, butCannot use deductions to replace quality rectification——The money has been deducted, but the root cause remains; next season the radar chart will still be misleading.

Eliminating is difficult, often because there is no backup.

When the rules are strict but the strategic materials come from a single source, no one dares to execute an exit. Before implementing hard constraints online, at least do three things:

• Alternative sourcing or safety stock strategy for strategic materials;
• The limited period truly restricts new orders and shares, rather than just being written in the meeting minutes;
• Re-entry after withdrawal should go through the formal process to prevent silent registration.

Re-evaluation for recovery also requires evidence: rectification closure, small batch stability, and meeting observation period indicators. A 'recovery' without evidence is equivalent to telling the site that rules are negotiable.

The alternative list is not about adding a long string of names 'for backup but never placing orders.' At the very least, there should be a usable sourcing record: samples or factory audit conclusions are still valid, production capacity and delivery time can be met, and price master data can be activated. Otherwise, the suspension assessment will pass in the meeting but get stuck during execution, and next time no one will bring up withdrawal again. For monopolies or customer-designated suppliers, write 'cannot withdraw' into the category strategy, and instead use limited quantities, on-site presence, and stricter inspections, rather than pretending that general elimination rules apply.

The regular meeting only records four types of outcomes

It is recommended to compress the joint supplier/quality meeting into a decision-making meeting: after drilling down on overdue rectifications and OTIF sub-items on the Kanban, only output—maintain, limit, suspend supply evaluation, start alternatives. Decisions should be entered into the system status, and the first item next week should check whether it has been executed to an orderable scope. Colors and rankings can be kept, but if the weekly meeting is still debating 'which table is accurate,' it indicates that the events are not yet from a common source.

Limited execution requires spot checks: Whether new orders from limited suppliers during the limited period are restricted or need additional approval; whether the quota cap is written back to sourcing and order strategies. If orders can still be placed normally during the suspension assessment period, it indicates that the state machine is stopped at the reporting layer. After exiting and archiving, any re-admission must go through the admission or re-evaluation process again, and the reason for exit should be retained to prevent 'rebirth under a different code'.

Supply management solutions typically place access control, master data, licenses and risk tags, performance evaluation, and corrective elimination within the same lifecycle; quality collaboration displays incoming material nonconformities and 8D reports, with procurement and SQE sharing timelines. In JEDP-SRM, observation periods, red lines, and evidence are configurable and linked to shares/corrective actions—the value lies in "there is movement in the directory," not in radar chart animations. Process data should be drillable down to order lines, inspection batches, and correction sheets, with meeting discussions focused on evidence rather than impressions.

Weight, Cycle, and the 'Looks Score' Trap

Different categories should have different weights: safety components should emphasize quality red lines, long-lead-time components should emphasize delivery and coordination, and compliance-sensitive components should emphasize licenses and certifications. The whole company uses a single weighting table, but breaking it down by item dilutes it into 'everything is important equals nothing is important.' In terms of cycle, monthly reviews focus on the process (overdue instances, breach counts), quarterly reviews focus on share and retention decisions; if only done at year-end, corrective actions are always half a step behind.

Be wary of three types of 'good-looking scores': business scores that are too high, masking quality; extremely small sample sizes yet giving full marks; supplier self-assessments and internal data each calculated separately but then averaged. Scoring rules should be announced after being finalized, and any changes should follow a controlled process; otherwise, suppliers may think the rules are negotiable, reducing their motivation to make improvements.

Event-driven scoring is harder to falsify than manual end-of-year scoring: material nonconformities, overdue 8Ds, OTIF failures, and expired certifications are recorded in the process ledger as they occur. Manual score adjustments can have a channel, but they must include the reason, the approver, and a validity period; if score adjustments become the default, the itemized score board will revert to being decorative. When the observation period for a new supplier ends, use the same set of scores to decide 'confirmation / observation extension / exit,' and do not invent a separate, more lenient scoring system.

In which situations should you not force elimination first

• Single source of supply with no alternative plan: first prepare a backup, then discuss a hard exit;
• Quality and delivery incidents are still in the form: handle the incidents first, then calculate the score automatically;
• Treating performance merely as a punitive tool, without providing guidance or a review path: suppliers will conceal problems;
• Traders and manufacturers use the same weight: distortion in individual items;
• The group's shared directory does not review the scope of subsidiaries: exits are placed in Company A and Company B orders still.

First, freeze the category weights and red-line items, run a trial with one or two categories to see if the 'limits are really enforceable,' and then expand to automatic scoring and the full catalog. Over the past year, there have been zero exits and zero limits in the catalog, yet quality fluctuations remain, indicating that performance is still at the display level—executing a real-limited case first is more convincing than launching ten dimensions at once. Guidance and punishment must go hand in hand: if you punish without helping, suppliers will hide problems; if you only help without limiting, the catalog will continue to expand.

What to do next

Export two lists this week: Top overdue rectifications, and suppliers with poor OTIF sub-scores but medium to high overall scores. Bring them to the joint meeting of Purchasing and SQE, and for each name, only record one action (maintain / limit / evaluate for suspension / initiate backup), and check whether the system's orderable range is consistent within three days.

• SRM Supply Solution: Performance and Rectification Elimination
• Supplier Access and Risk Label
• Incoming Material Quality and Supplier Closed Loop
• How is OTIF measured at the line level for performance?