How to Build a Vendor Scorecard for Electronics Suppliers

A supplier scorecard earns its keep when it changes a decision: who gets the next award, which supplier gets a development plan, where the quality team audits next, and when a sole source needs a second source. A score out of 100 that looks healthy while hiding repeated late deliveries or an unauthorised substitution does the opposite. It gives everyone a reason not to look.

This guide is about measuring suppliers you already buy from. Vetting a new one before the first order is a separate job, covered in the pre-PO due diligence guide.

Group suppliers before you compare them

An authorised distributor has no control over a semiconductor fab's yield. A transformer winder controls its whole process. Put them in one league table and the numbers mislead. Score within comparable groups:

• original component manufacturers;
• authorised distributors;
• independent distributors and stockists;
• EMS and contract manufacturers;
• bare-board PCB fabricators;
• cable, harness, magnetics and other build-to-print suppliers;
• programming, test and logistics service providers.

Then split each group by criticality. Strategic, sole-source and safety-related suppliers get the full scorecard monthly. Occasional low-risk vendors can be reviewed by exception. NIST's manufacturing extension programme lists supplier segmentation alongside supplier metrics and scorecards as core supply-chain practices, the latter for measuring performance and deciding which suppliers earn continuing or increased business.

Pick dimensions and weights that match the risk

Quality, cost and delivery are the long-standing core of supplier performance measurement, and ASQ's supplier-quality material adds responsiveness and the total cost of doing business with a supplier. A practical electronics scorecard uses six dimensions. The weights below are an illustration for a build-to-print supplier, not a standard:

Move the weights by segment. Safety-critical custom parts might carry 40 per cent on quality. For a catalogue distributor, delivery, stock-data accuracy and traceability matter more than process capability, which the distributor does not control. Agree the weights with quality, planning, engineering and finance before the first period, and then leave them alone for the year.

Write the quality formulas down exactly

Most scorecard disputes are really disputes about definitions. Each metric needs a written numerator, denominator, data source and period.

Defect PPM = rejected quantity ÷ received quantity × 1,000,000.

If 120,000 parts were received in the quarter and 54 were rejected and confirmed as supplier defects, PPM = 54 ÷ 120,000 × 1,000,000 = 450. State where rejection counts: incoming inspection, the line, or only after the supplier agrees. Keep administrative returns, such as wrong quantity on a label, out of the defect count and track them under service.

Lot acceptance rate = accepted lots ÷ inspected lots × 100.

With 46 of 48 lots accepted, the rate is 95.8 per cent. Show this next to PPM, because one very large good lot can bury several small failed ones in a quantity-based figure.

Corrective action. Track supplier corrective-action requests by severity, days to containment, days to closure and, most important, recurrence. A request closed in a week that comes back next month was not closed.

Cost of poor quality. Record the money your plant spent because of supplier defects: sorting, rework, line downtime, premium freight for replacements and customer containment. As an illustration, a bad lot that costs ₹18,000 in sorting, three hours of downtime at ₹12,000 an hour and ₹22,000 in air freight adds up to ₹76,000. Agree the attribution rules with the supplier in advance so the figure is not seen as an arbitrary penalty.

Measure delivery against the original promise

Line OTD = PO lines received inside the agreed window ÷ PO lines due in the period × 100.

In the example quarter, 230 of 250 due lines arrived inside the window, so OTD = 230 ÷ 250 × 100 = 92.0 per cent.

The formula is simple; the rules around it are where gaming happens. Write down:

• which date counts: the date you requested, the supplier's first acknowledged date or its latest one;
• whether you measure ship date or receipt date;
• the early and late tolerance, for example three days early to zero days late;
• how a partial delivery counts;
• how buyer-caused changes, holds and customs delays are excluded.

Measure against the first acknowledged date. If a supplier can push the acknowledgement back twice and still score "on time", the metric rewards decommits.

Add OTIF (on time and in full) for lines where a short shipment hurts. If 221 of the 230 on-time lines were also complete, OTIF = 221 ÷ 250 × 100 = 88.4 per cent. ASQ's supplier-quality body of knowledge names both OTD and OTIF as delivery measures. Track average decommit days and premium-freight incidents alongside.

Score commercial performance as total cost

Unit price on its own rewards the wrong supplier. NIST MEP's supplier-selection advice warns that focusing on cost per unit tends to raise the total cost of ownership. Include:

• price variance against an agreed baseline;
• landed cost, freight and duties;
• MOQ, standard-pack and cancellation flexibility;
• invoice and credit-note accuracy, including GSTIN and HSN details on the tax invoice;
• frequency of expedite charges and spot buys;
• payment terms and warranty support;
• cost-reduction proposals that engineering accepted.

Compare prices only for the same part, quantity, pack, Incoterm, currency and delivery terms. The quote comparison guide shows how to put bids on equal terms first.

Make the judgement dimensions evidence-based

Service and resilience scores are partly judgement, which is fine if the judgement is anchored. Use a defined scale and score from sampled records, not memory.

For resilience, look forward rather than back: visibility of manufacturing sites and sub-tiers, how early capacity constraints are reported, whether an alternate site is qualified, sole-source and single-country exposure, and how stable acknowledgements and backlog have been. NIST's advice to small manufacturers is that a balanced scorecard mixes quantitative and qualitative data and uses standard measures such as responsiveness and on-time delivery.

For compliance, verify rather than count logos: certificate status and site scope, manufacturer authorisation for each line, product-change and end-of-life notices received on time, material declarations, and audit findings closed. ISO 9001 certification is one input (the 2026 edition was published in September 2026, checked October 2026). For assemblers, an entry on IPC's J-STD-001/IPC-A-610 Qualified Manufacturers List names the qualified classes and an expiry date; record those, not just "IPC: yes".

Convert, weight and add hard gates

Convert each raw metric to a 0 to 100 score with a published curve, then multiply by its weight and add up.

Avoid cliffs. If 95.0 per cent OTD earns full marks and 94.9 per cent earns half, suppliers argue about the decimal, not the deliveries. Use a straight line instead. Suppose the delivery curve gives 0 points at 85 per cent OTD and 100 points at 98 per cent. A supplier at 94.9 per cent then earns (94.9 − 85) ÷ (98 − 85) × 100 = 76.2 points, and one at 95.0 per cent earns 76.9. Always show the raw value next to the points.

Worked weighting, using dimension scores of quality 82, delivery 90, commercial 75, service 80, resilience 60 and compliance 95:

• quality 82 × 0.30 = 24.60
• delivery 90 × 0.25 = 22.50
• commercial 75 × 0.15 = 11.25
• service 80 × 0.10 = 8.00
• resilience 60 × 0.10 = 6.00
• compliance 95 × 0.10 = 9.50

Total: 81.85.

Some events must override the average. Suspected counterfeit material, falsified certificates, an unauthorised substitution, a serious safety breach or concealment of a known problem should trigger quarantine and escalation whatever the total says. For authorised distributors, a failure to disclose non-authorised supply at quotation, which SAE AS6496 requires as summarised by ECIA, belongs in this category.

Set bands and consequences

The example supplier, at 81.85, is Approved, but its resilience score of 60 deserves a specific action. Be careful with sole sources in the bottom band: dropping one abruptly can hurt your customers more than the supplier's performance did. Contain, develop and qualify an alternative in parallel, as described in the multi-supplier strategy guide.

Run the review cycle

A monthly data cut with a quarterly review suits most strategic suppliers. A cycle that holds up:

1. Freeze the period's transaction data.
2. Check exclusions with the internal owners.
3. Send the draft score and the evidence to the supplier.
4. Correct factual errors; do not renegotiate definitions.
5. Look at the trend across several quarters.
6. Agree no more than three actions, each with an owner and date.
7. Escalate missed actions.
8. Feed the result into allocation and the approved vendor list.

If a definition turns out to be wrong, change it from the next period, with approval, and note the change on the scorecard.

Mistakes that make scorecards useless

• one template for every supplier type;
• definitions that change each quarter;
• revised promise dates that erase lateness;
• percentages from tiny samples shown without the count;
• points for certificates nobody verified;
• missing data treated as good performance;
• critical events averaged away;
• scores that never lead to an action or an award decision.

A new supplier should not get a performance score until it has enough deliveries to mean something. Until then, record its qualification evidence and pilot results as a provisional assessment and say so on the sheet.