Supplier Scorecards: How to Evaluate and Rank Your Factories

A supplier scorecard turns sourcing from a relationship business into a measurable one. Most brands that buy activewear from overseas factories rely on memory, price and a vague sense of “they always deliver” — until a season collapses because one factory quietly subcontracted half a shipment, or a dye lot arrives four shades off. A scorecard replaces intuition with a repeatable, shared scoring method that both sides can see, argue about with data, and improve against over time. This guide explains how to build one for apparel manufacturing, what to measure, how to weight those measures for sportswear specifically, and how to run the quarterly review so it changes behaviour rather than gathering dust in a spreadsheet.

Why Apparel Suppliers Resist Scorecards — and Why They Should Not

Factories are used to being audited for compliance, not evaluated for performance. Compliance audits feel like a threat; performance scorecards feel useful when they are transparent and consistent. The difference is that a compliance audit is binary, while a scorecard is comparative and directional. It tells a factory exactly where they stand against their peers and against their own past results, which is precisely what a factory owner needs to justify investment in new equipment or additional QC staff.

For buying teams, the operational value is even more direct. A scorecard lets you:

• Allocate future volume objectively instead of by habit.
• Spot a declining supplier two months before it becomes a delivery crisis.
• Justify a higher unit price for a supplier that saves you money elsewhere.
• Compare factories across countries without re-running the same conversations.
• Document due diligence when a brand, investor or retailer audits your supply chain.

The Five Scoring Pillars

A workable apparel scorecard uses five pillars. Fewer than four and you miss something critical; more than six and nobody can act on the output. Each pillar contains two to five sub-metrics with a defined measurement method.

• Quality (30%): AQL defect rate at final inspection, first-pass yield, in-line rejection rate, measurement accuracy against spec tolerance, and rework volume.
• Delivery (25%): On-time-in-full percentage, average days late when late, production lead time versus quoted lead time, and response time on sample rounds.
• Cost and commercial (20%): Price competitiveness against a benchmark basket, price stability over twelve months, willingness to quote transparent breakdowns, and payment term flexibility.
• Capability and responsiveness (15%): Sample turnaround, engineering suggestions, technical problem solving, ability to handle new constructions, and communication quality.
• Compliance and risk (10%): Valid audit reports, chemical test history, subcontracting disclosure, labour documentation, and environmental permits.

These weights suit a brand in a stable growth phase buying mid-volume technical apparel. A young brand launching its first collection should shift weight toward capability and responsiveness, because getting a workable product matters more than shaving cents off a unit price.

Choosing Metrics That Cannot Be Gamed

The fastest way to ruin a scorecard is to score something the factory can manipulate without improving anything real. Defect rate measured by a factory’s own QC team, for example, will trend toward zero as reporting gets softer, not as quality gets better. Every metric needs an independent measurement source, a fixed definition, and a denominator.

A few rules that hold up in practice:

• Measure at your inspection, not theirs. Use your own AQL results or a third-party inspection report. If you have no inspection budget, photograph a random sample from every shipment and score visually against the same checklist each time.
• Define late precisely. Late means the ex-works date is later than the confirmed date on the purchase order, not the date someone mentioned in a chat.
• Use a rolling window. Score the last twelve weeks or the last three shipments, whichever is larger. A single bad order should dent the score, not destroy it.
• Record sample rounds separately. A factory that needs five sample rounds for a legging is telling you something about their pattern room that delivery data will never capture.
• Never score price alone. A supplier quoting 6% below the benchmark but delivering a 4% defect rate is more expensive per sellable unit than one quoting at benchmark.

The last point deserves its own calculation. Compute landed cost per sellable piece: unit price divided by first-pass yield, plus freight, duty and inspection cost. That single number is the most honest cost metric in apparel sourcing, and it belongs in the cost pillar as a supporting figure.

Sportswear-Specific Scoring Factors

Activewear has failure modes that general apparel scorecards miss. Leggings fail at the crotch seam and the waistband elastic. Sports bras fail at the strap attachment. Any compression garment fails at stretch recovery after repeated laundering. A scorecard for a yoga or performance brand should therefore add technical sub-metrics to the quality pillar:

• Seam strength test results on the actual seam types used, measured in newtons per centimetre.
• Stretch recovery after ten home-laundry cycles, expressed as percentage growth.
• Opacity under flash photography on light colourways.
• Pilling grade after a defined abrasion cycle.
• Colourfastness to washing, perspiration and rubbing, scored separately.
• Elastane content verification through a fibre composition test on finished goods, not just on the fabric roll.

Request these tests once per season per fabric and construction combination, and keep the reports with the scorecard. A factory that resists third-party testing on a fabric they claim to have used for years is telling you the claim may be stale.

Building the Scoring Scale

Use a simple 1 to 5 scale with written anchors for every score point. Without written anchors, two buyers will score the same factory differently, and the factory will learn to argue about the person rather than the process.

• 5 — Benchmark: meets the best-in-class threshold you have seen across your supplier base.
• 4 — Strong: consistently meets targets with minor, non-repeating issues.
• 3 — Acceptable: meets the contractual minimum but no better.
• 2 — At risk: missed targets in the current window; corrective action open.
• 1 — Failing: repeated misses or a single severe event such as a rejected shipment.

Define the thresholds numerically for each metric, then map numbers to the scale. For example: on-time-in-full at 98% or above scores 5; 95 to 97.9 scores 4; 90 to 94.9 scores 3; 80 to 89.9 scores 2; below 80 scores 1. The exact numbers matter less than applying the same ones to every factory for at least four quarters before changing them.

Data Collection Without Extra Headcount

The most common reason scorecards die is data collection. A system that needs a dedicated analyst will not survive a busy season. Design for data you already produce: purchase orders, inspection reports, shipment records and email timestamps.

• Keep one master order log with confirmed date, actual ex-works date, shipped quantity and inspection result.
• Save every inspection report in a consistent folder naming scheme, one folder per factory per quarter.
• Log sample round counts at the moment the sample is approved, not retrospectively.
• Tag email threads with the factory name so response time is measurable.
• Enter data weekly, in ten minutes, rather than quarterly in an afternoon of archaeology.

A single spreadsheet with one tab per factory and a quarterly summary tab is enough for brands buying from three to fifteen factories. Do not buy software before you have run the process manually for two quarters; you will specify the wrong tool.

Ranking and Volume Allocation

The score only matters if it changes decisions. Tie volume allocation to score bands in a written policy:

• Top band: eligible for new development, larger allocation, longer-term commitments and priority on capacity.
• Middle band: steady allocation, quarterly improvement targets agreed in writing.
• Lower band: no new styles, no allocation increase, corrective action plan required within two weeks.
• Failing band: new orders paused pending a formal review; dual-source immediately.

Publish the bands to your suppliers. A factory that knows the top band gets the next development programme will invest in it. A factory that suspects allocation is decided by whoever visits oftenest will invest in relationships instead of capability, which is not what you are paying for.

Running the Quarterly Review

A scorecard without a conversation is a report. Hold a thirty-minute review with each strategic factory every quarter, following the same agenda:

• Share the score and the underlying data, not just the total.
• Ask the factory to explain the two lowest sub-metrics in their own words.
• Agree one to three corrective actions with an owner and a date.
• Confirm the next quarter’s forecast so capacity planning is realistic.
• Record what you will change on your side — late tech packs and unclear approvals cause as many delays as factory failures.

The last item matters for credibility. If your own sample feedback arrives three weeks late, the delivery metric is partly your fault, and a scorecard that never acknowledges this will be treated as theatre.

Common Failure Modes

• Too many metrics. Twenty sub-metrics produce a report nobody reads. Six to twelve is the working maximum.
• Changing weights every quarter. Reweighting after a disappointing result destroys trust. Fix weights for a year.
• Scoring only strategic suppliers. Score everyone you buy from, even at low volume; secondary suppliers become primary during capacity crunches.
• No consequence. If the lowest scorer gets the same volume next season, the exercise is decoration.
• Ignoring lead time inflation. A factory that quotes 45 days and delivers in 60 is worse than one that quotes 60 and delivers in 58, even if its quoted price looks cheaper.

A Minimal Starter Template

If you are starting from nothing, run one quarter with four metrics only: AQL defect rate at inspection, on-time-in-full percentage, average sample rounds per style, and landed cost per sellable piece. Score them on the 1 to 5 scale, rank your factories, and hold one review meeting with each factory. Add the technical quality metrics in the second quarter, and the compliance pillar in the third. A scorecard that grows with your process will be used; one that arrives fully formed will be ignored.

The ultimate test of a good supplier scorecard is simple: if you removed the spreadsheet and kept only the conversations, would your sourcing decisions change? If not, the scorecard is working — it has made your judgment explicit, shared and defensible.