Supplier relationships are usually reviewed by feel. Someone remembers a late batch, someone else remembers a good price, and the annual decision gets made on whichever memory is loudest. A scorecard replaces that with three or four numbers computed the same way every month, which is enough to see decline early and to have a specific conversation rather than a general complaint.
The point is not bureaucracy. Vendors respond to measurement because measurement is what they are held to by their own customers. A supplier who knows you track on-time-in-full will sequence your orders differently from one who knows you do not.
Pick Four Metrics, Not Fourteen
Scorecards fail by including everything. Four metrics, each with a clear definition and a data source, beat a dashboard nobody maintains:
- On time in full (OTIF) – the share of orders arriving by the confirmed date with the confirmed quantity. Combining both conditions matters: a partial shipment on time is still a problem.
- Defect rate – units rejected at inspection divided by units received, with the rejection reason recorded.
- Lead time variability – the standard deviation of actual lead times against the quoted lead time. Consistency often matters more than speed; a supplier who always takes 18 days is easier to plan around than one who takes 10 or 30.
- Responsiveness – time to first meaningful reply on a query, measured in working hours.
Scorecards also make commercial terms negotiable in a way that impressions never do. When a vendor asks for a larger order commitment, a documented record of delivery and defect performance is the evidence behind either agreeing or declining, and it turns a price conversation into a value conversation.
Weight them to your model. A catalogue where stockouts are the main risk should weight OTIF heavily. A catalogue where returns drive cost should weight defect rate. There is no correct weighting, only one that matches where your money is lost.
The Data You Must Capture Per Order
A scorecard is only as good as the order record. For every purchase order, capture: order date, confirmed ship date, actual ship date, quantity ordered, quantity received, inspection result, and any exception note. That is six fields, and four of them are already in the system if the receiving process records them at all.
The failure mode is recording at the wrong moment. If quantity received is logged when the carton is opened for picking rather than at goods-in, short shipments surface weeks late and the scorecard reflects discovery rather than performance.
Set Thresholds and Consequences in Advance
A score without a consequence is a decoration. Define three bands – acceptable, watch, and action – and decide what each triggers before the first review. Action might mean a smaller trial order, a second source for that SKU, or a call with the factory owner rather than the sales contact.
Publish the definitions to the supplier. Most factories are used to scorecards from larger customers and will engage with the numbers directly. Sharing the method also removes the argument about whether a late shipment counted, because the rule was agreed before the order.
Technical Background
Supplier measurement draws on key performance indicator practice, where the discipline is choosing few measures with unambiguous definitions. The delivery metric itself comes from supply chain management, where on-time-in-full is the standard composite of schedule and quantity. The sourcing context for applying it is described under procurement.
Using the Scorecard in a Review
Open with the trend, not the absolute number. A supplier at 92% OTIF and improving is a different conversation from one at 94% and sliding. Ask what changed on their side: a raw material substitution, a new shift, a machine down, a logistics partner switched. Most declines have a specific cause that a specific fix addresses.
Then reconnect to commercial terms. A supplier with strong scores is worth a larger share of volume, and a supplier with weak scores should not be rewarded with a bigger order just because the unit price is lowest. Ranking on price alone is how a catalogue acquires its most expensive problems.
Scorecards sit downstream of verifying a factory before the first order and next to the technical guide to reducing defect rates. If a review ends in a decision to move volume, the migration checklist sets out how to do it without breaking live orders, and peak season readiness explains why the review should happen well before Q4.
Perguntas Frequentes
How many metrics should a supplier scorecard have?
Three to five. Each needs an unambiguous definition and a data source, and more metrics usually means weaker data quality.
What is on time in full?
The share of orders delivered by the confirmed date with the confirmed quantity. A partial shipment on the due date does not count as complete.
Why measure lead time variability?
Because planning needs predictability. A supplier with a stable 18-day lead time is easier to buffer than one ranging from 10 to 30 days.
Should the scorecard be shared with the supplier?
Yes. Sharing the definitions removes disputes about whether a shipment counted and signals that the numbers are used.
How often should suppliers be reviewed?
Monthly for active vendors, with a quarterly commercial review where allocation and pricing are discussed alongside the scores.
What if a small supplier has too little data to score?
Score on the metrics you can compute and expand as volume grows, or run a deliberate trial order to generate a comparable data point.
Does a good score justify paying more per unit?
Often yes. The cost of a stockout or a defect batch usually exceeds a small unit price difference, which is why allocation should follow the scorecard.
