Why a Supplier Scorecard Matters When FOB Is the Only Number on the Table
The third hoodie delivery from the cheaper supplier has landed late, the final inspection has failed twice this season, and the costing review still shows that supplier ahead by forty cents a unit. Which number should the buying director believe? A supplier scorecard exists to answer that question with evidence rather than memory: it turns delivery, quality, communication and follow-through into scores that sit next to the FOB in every sourcing conversation.
Opinions about suppliers form in the weeks when something goes wrong, they are rarely written down, and they differ between buyer, technologist and planner. A scorecard captures the same facts in one place, applies the same rules to every supplier and makes trends visible across seasons. It gives judgement something to stand on.
The figures above describe the template that follows, not industry results. Weights, bands and example data are illustrative; set your own.
What a Supplier Scorecard Measures: The Five Core Metrics
A useful supplier scorecard is short. Every extra metric dilutes attention and adds data collection work that someone has to own. The five below cover the four areas in the brief: quality, delivery, communication and corrective action, with sample performance separated from bulk performance.
Illustrative metric weights (out of 100)
Share of the total score
Illustrative starting weights for a hoodie or jersey programme. Weights express your priorities; change them to suit the category, the season and how much risk the programme can carry.
Delivery carries the highest weight because a late hoodie in a seasonal window cannot be recovered with a markdown, while a quality defect can sometimes be reworked. A development-heavy programme would shift weight towards sample performance; a later section tests that choice.
Compliance and safety sit outside the weighted score. A lapsed social-compliance audit or a failed restricted-substance requirement is a gate: weighting it would let a strong delivery score offset a failure that no score should offset. See social compliance audits for how those gates are usually evidenced.
Define the KPIs: OTIF, AQL Failures, First-Pass Approval, Responsiveness and Corrective Action
Every KPI needs a formula, a data source and a rule for what counts, or two buyers will score the same supplier differently. Definitions vary by retailer, so write down which one your programme uses and keep it fixed.
Scorecard KPI definitions
| KPI | Formula | Data source | Watch out for |
|---|---|---|---|
| OTIF (on time in full) | Order lines delivered by the agreed date in the agreed quantity, divided by order lines due, times 100 | PO log, ex-factory or goods-received dates | State the date measured (ex-factory or arrival) and the PO quantity tolerance; a line missing either condition is a miss |
| AQL failures | Final inspections failed at first attempt, divided by final inspections held, times 100; also record major and critical defects found against the accept number | Inspection reports | AQL is the acceptable quality limit sampling standard; a pass after rework still counts as a first-attempt fail |
| First-pass approval (first-time-right) | Samples approved without rework, divided by samples submitted, times 100 | Sample tracker | Count each sample type separately; ignore changes the buyer requests after approval |
| Responsiveness | Buyer queries answered fully within the agreed response time, divided by queries raised, times 100 | Email or ticket log time stamps | A quick holding reply is not an answer; score only replies that move the issue forward |
| Corrective-action closure | Corrective actions closed by the agreed date with the fix verified, divided by corrective actions raised, times 100 | Corrective action log | A closed action that recurs on the next order is reopened, not counted as a success |
Definitions used for the worked example in this guide. Adapt the wording to your own programme and keep it consistent across all suppliers.
Quality: AQL failures and defect findings
Bulk quality is best measured from independent final inspections rather than from the supplier's own reports. The headline figure is the first-attempt failure rate, because it shows how often the factory is ready for inspection when it says it is. Record the failure type too (measurement, construction, appearance, labelling) to see whether an issue returns. Our guide to AQL inspection explains the sampling plan and defect classes.
Delivery: OTIF
OTIF is demanding because it fails a line for a late or a short delivery. Measure it by order line rather than by whole PO, otherwise one late colour hides inside a mostly good order. Record the cause of each miss (fabric, trims, approval delay, capacity, shipping) to see whether the supplier or the programme is the constraint. Production lead times covers the stages where delay usually originates.
Communication: responsiveness
Responsiveness is the hardest KPI to keep honest, because it depends on a log that busy teams neglect. Agree the response time in advance, for example one working day for routine queries, then sample the log rather than score every message.
Follow-through: corrective action
When something fails, the scorecard records whether the supplier found the root cause, fixed it, and proved it on the next lot. That is corrective-action closure. A supplier with a few defects and a disciplined response is often safer than one whose problems are simply not recorded.
Score Samples and Bulk Separately
A factory that is excellent at making one beautiful sample may still struggle to repeat it across 5,000 pieces, and the reverse is also true. Averaging the two hides the pattern. Keep two views of the same supplier: a sample-stage view and a bulk-stage view, and read them side by side.
Sample stage versus bulk stage
| Question | Sample stage | Bulk stage |
|---|---|---|
| What is being tested? | Technical understanding, pattern making, hand-feel and finish | Process control, capacity, consistency and discipline on dates |
| Main scorecard metrics | First-pass approval; responsiveness; sample turnaround against the plan | OTIF; AQL failures; corrective-action closure; responsiveness |
| What a weak score means | Specification gaps or limited technical staff; see how to improve sample approval rate | Planning or capacity problems; see how to prevent production delays |
Illustrative structure. Add your own sample types, such as proto, fit and pre-production samples, and keep their results visible separately.
Separate scoring also protects fairness: a supplier on a first trial has only sample data, so state that bulk evidence is not yet available and use a trial order to fill the gap. The garment sampling process sets out the sample types you can score.
How to Build the Supplier Scorecard in Five Steps, with Scoring Bands
Most scorecards fail because nobody owns the data, not because the template is wrong. This sequence gets a usable first version running within a season.
Building the scorecard
- 1Choose the five metricsUse quality, delivery, sample approval, responsiveness and corrective action, or swap one for a measure that matters more in your category.
- 2Fix the definitionsWrite the formula, date measured, tolerance and data source for each metric before any scoring starts.
- 3Set weights and bandsAgree weights that total 100 and scoring bands from 1 to 5, using your own risk appetite rather than a borrowed benchmark.
- 4Name the data ownerOne person collects and scores after every order and posts the result to the supplier within an agreed time.
- 5Review each seasonShare the scorecard with the supplier, agree actions, and compare to the previous seasons before allocations are set.
A scorecard the supplier sees and discusses changes behaviour; one kept privately mainly records disappointment.
Share the template with suppliers before the first PO is placed. Agreeing the rules in advance removes most of the arguments that follow a poor score, and it supports the weekly rhythm in how to stop chasing suppliers after the PO.
Scoring bands: turning raw results into a 1 to 5 score
Raw percentages are converted into a 1 to 5 score so that very different metrics can be combined. The bands below are the ones used in the worked example. They are illustrative: a short-run fashion programme and a core replenishment programme should not use identical thresholds.
Illustrative scoring bands
| Score | OTIF % of lines | AQL first-attempt fail % | First-pass approval % | Responsive within agreed time % | Corrective actions closed on time % |
|---|---|---|---|---|---|
| 5 (excellent) | 95 or above | 0 | 80 or above | 95 or above | 95 or above |
| 4 (good) | 90 to 94.9 | Up to 10 | 65 to 79.9 | 85 to 94.9 | 85 to 94.9 |
| 3 (acceptable) | 80 to 89.9 | Up to 20 | 50 to 64.9 | 70 to 84.9 | 70 to 84.9 |
| 2 (weak) | 70 to 79.9 | Up to 35 | 35 to 49.9 | 50 to 69.9 | 50 to 69.9 |
| 1 (unacceptable) | Below 70 | Above 35 | Below 35 | Below 50 | Below 50 |
Bands are illustrative thresholds for a worked example, not benchmarks. Set your own from the risk the programme can absorb, and confirm them with suppliers before use.
Weighted score = sum of (score x weight) divided by 100, giving a result between 1 and 5. Multiply by 20 to show it out of 100. With few order lines per season, a single miss moves a percentage sharply, so always show the counts (for example 17 of 24 lines) next to the percentage.
Worked Example: Scoring Two Suppliers Beyond FOB
This worked example is illustrative and uses hypothetical data. Supplier A quotes a hoodie FOB of USD 9.60 and Supplier B quotes USD 10.00, both for the same specification. The buyer tracks one season of 10,000 units with each, then fills in the scorecard. All inputs are assumed.
Season inputs (hypothetical)
| Metric | Supplier A | Supplier A result | Supplier B | Supplier B result |
|---|---|---|---|---|
| OTIF | 17 of 24 lines on time in full | 70.8% | 22 of 24 lines on time in full | 91.7% |
| AQL failures | 3 of 6 final inspections failed first time | 50.0% | 1 of 6 final inspections failed first time | 16.7% |
| First-pass approval | 8 of 10 samples approved without rework | 80.0% | 6 of 10 samples approved without rework | 60.0% |
| Responsiveness | 57 of 60 queries answered in time | 95.0% | 50 of 55 queries answered in time | 90.9% |
| Corrective-action closure | 4 of 8 closed on time | 50.0% | 3 of 3 closed on time | 100.0% |
Illustrative data for a worked example. Not a benchmark and not based on any real supplier.
Applying the scoring bands gives the scores below. The pattern is common in practice and worth reading carefully: Supplier A is the stronger sample-maker and the faster responder, but it delivers late, fails more inspections and closes fewer actions in bulk.
Scores from the bands (1 to 5)
1 dot = weak, 5 dots = excellent
| Supplier A | Supplier B | |
|---|---|---|
| OTIF (weight 30) | ●●●●● | ●●●●● |
| AQL failures (weight 25) | ●●●●● | ●●●●● |
| First-pass approval (weight 20) | ●●●●● | ●●●●● |
| Responsiveness (weight 15) | ●●●●● | ●●●●● |
| Corrective action (weight 10) | ●●●●● | ●●●●● |
Scores come from applying the illustrative bands above to the hypothetical results. They are an illustration, not a rating of any real supplier.
Weighted calculation (illustrative)
| Metric | Weight | A score x weight | B score x weight |
|---|---|---|---|
| OTIF | 30 | 2 x 30 = 60 | 4 x 30 = 120 |
| AQL failures | 25 | 1 x 25 = 25 | 3 x 25 = 75 |
| First-pass approval | 20 | 5 x 20 = 100 | 3 x 20 = 60 |
| Responsiveness | 15 | 5 x 15 = 75 | 4 x 15 = 60 |
| Corrective action | 10 | 2 x 10 = 20 | 5 x 10 = 50 |
| Total | 100 | 280 / 100 = 2.80 (56 out of 100) | 365 / 100 = 3.65 (73 out of 100) |
| Bulk stage only (OTIF, AQL, corrective action; weight 65) | 65 | 105 / 65 = 1.62 | 245 / 65 = 3.77 |
Calculation checked arithmetically. Bulk stage only divides the sum of the three bulk metrics by their combined weight of 65.
The cheaper supplier ranks well below the other once delivery and quality are counted, and the bulk-only view is starker than the total, because Supplier A's excellent samples lift its overall score. That gap is exactly what separate stage scoring is designed to reveal.
Does the Cost Gap Survive the Scorecard?
A score is only useful if it can be set against money. Continue the hypothetical example: on 10,000 units the FOB gap of USD 0.40 is worth USD 4,000 in Supplier A's favour. Now put a cost on each late or short order line. Assume, for illustration, that each late line costs USD 1,500 on average across expediting, air freight, internal chasing and lost sales. That figure is an assumption for the example; use your own measured cost.
Break-even view (hypothetical)
| Item | Calculation | Result |
|---|---|---|
| FOB saving with Supplier A | 10,000 units x USD 0.40 | USD 4,000 |
| Late or short lines: A versus B | 7 late lines (24 minus 17) against 2 (24 minus 22) | 5 extra lines |
| Extra cost of lateness with A | 5 x assumed USD 1,500 per line | USD 7,500 |
| Net position | USD 7,500 minus USD 4,000 | USD 3,500 against A |
| Break-even | USD 4,000 divided by USD 1,500 | About 2.7 extra late lines |
Illustrative arithmetic. It excludes inspection failures, rework and markdown exposure, which would widen the gap, and it should be rebuilt with your own measured costs.
In this scenario, a gap of just three extra late lines would erase the saving. The same logic applies to quality failures, which is why the scorecard sits alongside, not instead of, a landed-cost view. For the transition costs of leaving a supplier altogether, see the real cost of changing suppliers.
Do the Weights Change the Answer?
Weights encode strategy, so test whether your conclusion depends on them. Alternative weight sets applied to the same scores show how far the ranking moves; all are illustrative.
Weight sensitivity (hypothetical weights, same scores)
| Weight set (OTIF, AQL, first-pass, responsiveness, corrective) | Supplier A | Supplier B | Leader |
|---|---|---|---|
| Base: 30, 25, 20, 15, 10 | 2.80 | 3.65 | Supplier B |
| Bulk-heavy: 35, 35, 10, 10, 10 | 2.25 | 3.65 | Supplier B |
| Sample-heavy: 10, 10, 50, 20, 10 | 4.00 | 3.50 | Supplier A |
Illustrative weights applied to the hypothetical scores above. Each total is the sum of score x weight, divided by 100.
Supplier B leads under two of three weight sets, and Supplier A leads only when sample performance carries half the weight. A programme that ships bulk should not use sample-heavy weights, and anyone proposing a weight change should say which supplier it favours.
Compare Trends Across Seasons, Not Single Orders
One season is a snapshot. Three seasons show direction. A supplier at 3.2 and rising is a different proposition from one at 3.2 and falling, yet a single scorecard cannot tell them apart. Keep the weighted score for each season and compare it, together with the individual metrics, so you can see which dimension is driving the change.
Weighted score by season (hypothetical)
| Season | Supplier A | Supplier B |
|---|---|---|
| Season 1 | 3.20 | 2.90 |
| Season 2 | 3.00 | 3.30 |
| Season 3 (worked example) | 2.80 | 3.65 |
| Three-season average | 3.00 | 3.28 |
| Change, Season 1 to Season 3 | -0.40 | +0.75 |
Illustrative values chosen to show how a trend reads. The Season 3 figures match the worked example above.
The average alone suggests a small gap; the trend shows Supplier B improving and Supplier A sliding, which matters more for allocation. Note known causes such as a change of fabric mill, and treat one bad season after two good ones as a question to ask rather than a verdict.
Link Future Allocations to Performance
A scorecard that does not influence volume becomes paperwork. Suppliers notice quickly whether a poor score costs anything or a good one earns anything. Agree the link in advance and apply it consistently, using a small set of rules that the commercial team and the supplier can both read.
Illustrative allocation rules
| Weighted score and trend | Typical action | Evidence needed |
|---|---|---|
| 4.0 or above, stable or rising | Eligible for additional volume or new styles, subject to capacity checks | At least two seasons of data and a clean compliance gate |
| 3.0 to 3.99 | Hold current share; agree an improvement plan on the weakest metric | Plan with owners, dates and a re-score after the next order |
| 3.0 to 3.99 and falling | Hold, no new styles, and begin to qualify an alternative | Two consecutive declines or a failed corrective action |
| Below 3.0 | Reduce exposure, route new development elsewhere, set a probation period | Written notice of the gap and a defined recovery target |
| Compliance gate failed | Stop placing orders until resolved, regardless of score | Evidence from the audit or test body |
Thresholds are illustrative. Add a minimum number of order lines before any rule applies, so that one order cannot trigger a decision.
Do not move volume on one score. Require a minimum evidence base, two consecutive reviews pointing the same way, and a conversation with the supplier before any reduction. Evidence that would change the decision includes a documented root cause that has been fixed and verified on the next lot.
A new supplier earns allocation through a controlled trial with a scorecard, not because its price is attractive. The method is in how to test a new hoodie supplier, and for the risk of leaning on one factory see backup supply chain planning.
Common Scorecard Mistakes and a Checklist Before You Launch
The usual failures are about data, definitions and follow-up, and a short check before the first score is issued avoids most of them.
Pre-launch scorecard checklist
- Each KPI has a written formula, a date measured and a named data source.
- Weights add up to 100 and have been tested against at least one alternative weight set.
- Scoring bands were agreed with the supplier before the PO was placed.
- Samples and bulk are scored separately and reported separately.
- Compliance and safety are gates, not weighted scores.
- Counts are shown next to every percentage, with a minimum number of lines before a rule applies.
- One person owns data collection and the scorecard is shared within an agreed number of days.
- Allocation consequences are written down in advance and reviewed each season.
Resist adding metrics until the card is unreadable, and resist scoring only problems, which leaves no way to recognise a supplier that is improving. Which supplier's next score would change your sourcing plan, and do you already collect the KPI that would drive it?
How Ruhrose Approaches Supplier Scorecards
Ruhrose is a supplier, so we sit on the side being scored, and we think that is healthy. For hoodies, jersey, denim and woven casualwear produced in Bangladesh with UK buyer support, we are comfortable working to a buyer's own scorecard, sampling before bulk and agreeing definitions up front. Where a buyer wants a controlled trial, the proposed pilot quantity of 500 pieces per style and colour, where commercially and technically feasible, is a way of generating the evidence a scorecard needs; it reduces risk but does not remove it. See our test and scale model for how that works.