Vendor Scorecard: Metrics, Weighting, and a Template That Works

A vendor scorecard is a short, weighted set of measurements that rates how a supplier actually performed: did they deliver on time, at the agreed price, at the agreed quality, and were they workable when something went wrong. Its purpose is narrow and useful. It replaces "I think they have been fine" with a number you can put in front of a supplier at a business review, and it gives you a defensible basis for renewing, renegotiating, or replacing them.

Most scorecards fail for the same two reasons: they measure too many things, and they measure things nobody has data for. A scorecard with 22 criteria that gets filled in from memory once a year is theater. Four metrics you can pull from your own systems, reviewed quarterly, will change supplier behavior.

The four metrics that carry almost all the signal

Weight them to your actual risk. A contract manufacturer supplying a regulated product should carry quality at 40% or more. A distributor of commodity consumables where any defect is trivially replaceable should carry on-time delivery and price accuracy heavier. The weights are the part of the scorecard that encodes your strategy, so do not copy someone else's.

Two other criteria are worth adding only if they matter to your business and you have real evidence for them: compliance (current insurance certificates, required certifications, diversity status) and total cost (not just unit price, but freight, expedite fees, and the cost of the rework their defects cause). Do not add a "partnership" or "innovation" score. It always becomes a popularity contest.

Scoring: keep the scale short

Use a 1 to 5 scale with published thresholds so two different buyers scoring the same supplier land on the same number. For on-time delivery, for example:

• 5: 98% or better
• 4: 95% to 97.9%
• 3: 90% to 94.9%
• 2: 80% to 89.9%
• 1: below 80%

Then the weighted score is straightforward. A supplier scoring 4 on OTD (30%), 5 on price accuracy (25%), 3 on quality (25%), and 4 on responsiveness (20%) gets (4 × 0.30) + (5 × 0.25) + (3 × 0.25) + (4 × 0.20) = 1.2 + 1.25 + 0.75 + 0.8 = 4.0 out of 5.

Set your action bands in advance and write them into the supplier agreement, so nothing is a surprise: 4.0 and above is a preferred supplier, 3.0 to 3.9 triggers a corrective action plan with a named owner and a date, below 3.0 starts a sourcing review. The bands are what give the scorecard teeth. Without them, you have a report.

Where the data actually comes from

This is the step that quietly kills most scorecard programs, so it is worth being blunt about it. Every metric above is a comparison between two documents:

• On-time delivery compares the promised date on the purchase order against the receipt date on the goods receipt.
• Price accuracy compares the unit price on the purchase order against the unit price on the invoice. That gap is exactly what purchase price variance measures.
• Quality compares units received against units rejected, which lives in your receiving records.
• Responsiveness depends on when the supplier sent the purchase order acknowledgment relative to when you issued the PO.

If your POs, receipts, and invoices are already structured records in an ERP, these calculations are queries and you can automate the whole scorecard. If a meaningful share of your purchase orders and confirmations arrive as supplier PDFs that somebody retypes, you have a data problem long before you have a scorecard problem. You cannot measure on-time delivery against a promised date that only exists inside a PDF nobody parsed.

That is the practical fix to make first. Getting the PO header and every line into structured form, automatically, is what makes the scorecard computable rather than anecdotal. Our purchase order line item extraction pulls the promised dates, SKUs, quantities, and unit prices out of any supplier layout, and bulk purchase order upload clears a backlog of them in one pass. The invoice side of the price-accuracy comparison needs the same treatment, which is a job for AI invoice data extraction, since a price variance you cannot compute is a price variance you keep paying.

Running the review so it changes something

Score quarterly for strategic suppliers and annually for everyone else. Send the scorecard to the supplier before the meeting, not during it. The point of the review is not to present a grade; it is to agree on what happens next. A supplier who sees a 2 on quality three days early arrives with an explanation and a plan. A supplier who sees it on a slide arrives defensive.

Three rules keep the program honest:

1. Only score what you can evidence. If you cannot show the supplier the underlying lines, drop the metric.
2. Score every supplier in a tier the same way. Applying stricter thresholds to a vendor you already dislike is how scorecards lose credibility internally.
3. Own the input side. If your team confirmed a delivery date verbally and never put it on the PO, a late delivery is not entirely the supplier's fault, and they will say so.

What should be on a vendor scorecard?

A vendor scorecard should carry four to six weighted metrics you can evidence from your own records: on-time delivery, price accuracy against the purchase order, quality or defect rate, and responsiveness, plus compliance and total cost where they are material. Each metric needs a published scoring scale and a weight, and the overall score needs action bands that say what happens at each level.

How often should you review supplier performance?

Quarterly for strategic and high-spend suppliers, annually for the rest. Quarterly is frequent enough that a corrective action plan can be set and then actually checked at the next review, and infrequent enough that the data covers a meaningful sample of deliveries. Monthly reviews usually produce noise rather than signal, because one late shipment swings a small denominator.

What is a good vendor scorecard score?

On a 1 to 5 weighted scale, 4.0 and above generally marks a supplier performing to expectation and worth preferring for new awards. From 3.0 to 3.9 the supplier is workable but has a specific weakness that needs a corrective action plan with an owner and a date. Below 3.0, the honest reading is that you should be dual sourcing or resourcing while you work the issue.

Vendor scorecard vs supplier evaluation: what is the difference?

A supplier evaluation happens before you award business: it assesses capability, financial stability, capacity, and compliance to decide whether to use a vendor at all. A vendor scorecard happens after: it measures how the supplier actually performed against the orders you placed. Evaluation is a prediction; the scorecard is the result. Both matter, and the scorecard is what tells you whether the evaluation was right.

Start smaller than you think

If you have no scorecard today, do not build the perfect one. Take your top ten suppliers by spend, measure on-time delivery and price accuracy only, and run one review cycle. Those two metrics come straight off documents you already receive, they are hard to argue with, and they will surface the problem suppliers immediately. Add quality and responsiveness in the second cycle once the data pipeline is proven.

Related reading: the vendor onboarding process, procurement spend analysis, maverick spend, and consolidating supplier spend.