What Metrics Belong on a Bag Supplier Audit Scorecard?

Every sourcing organisation eventually builds a supplier scorecard and most of them build the wrong one. The common failure is to design it as a report card — a set of grades issued periodically to a production partner — when what the buyer actually needs is a decision instrument: something that answers, on a fixed cadence, whether to place the next order, increase the volume, hold the current level, or start looking elsewhere.

The distinction drives every design choice downstream. A report card can be subjective, because its only job is to summarise. A decision instrument cannot, because a number that moves a commercial decision has to be traceable to a source. This piece sets out the dimensions worth scoring on a bag programme, how to weight them, where the data comes from, and what turns a score into an action.

What Is a Programme Scorecard For?

Three decisions, in order of importance: whether the next order goes to this source, whether the volume on it can increase, and whether a second source needs to be qualified. A scorecard that cannot answer all three is under-designed. A scorecard that tries to answer anything beyond them — a ranking for its own sake, a league table across a supply base — is over-designed and will not be maintained past the second quarter.

The second purpose is to make a disagreement short. Without a score, a conversation about whether performance has slipped is an exchange of anecdotes: the buyer remembers a late shipment, the production partner remembers a specification change. With a score built from named sources, the same conversation is a check of two numbers against a definition, and it ends in minutes.

Which Dimensions Belong on a Bag Programme Scorecard?

Five cover the ground. Quality is the inspection result over the period, expressed as a rate rather than a pass or fail. Schedule is delivery against the quoted window, measured from the trigger event rather than from the order date. Documentation is whether the agreed document set arrived complete and correct the first time. Communication is response performance against the protocol. Change discipline is whether substitutions and process changes were notified in advance.

Those five are chosen because each is measurable from a document that already exists, and because together they cover the failure modes that actually damage a programme. Cost is deliberately excluded from this scorecard: price is negotiated per order and belongs in the commercial file, and folding it into a performance score makes the score move for reasons unrelated to performance.

The fifth dimension, change discipline, deserves an explanation because it is the one buyers omit and the one that predicts the most. It measures whether a material substitution, a process change or a tooling change was notified in advance and in writing. A change that improves the product is still a change, and an unannounced change to a food-contact material or to a restricted-substance input is a compliance event rather than a courtesy.

How Should Each Dimension Be Weighted?

Weight by consequence rather than by intuition. On a bag programme, quality carries the heaviest weight because a quality failure is visible to the end customer and expensive to unwind; schedule is close behind because a missed launch costs more than the goods. Documentation, communication and change discipline are leading indicators: they are weighted lower because they rarely damage a programme on their own, but they predict the two that do.

The weights above are a starting point and should be revisited once a programme has four quarters of data. The useful test is retrospection: apply the scorecard to the last four quarters and check whether the score moved when the buyer experience moved. If a quarter that felt bad scores well, the weights or the definitions are wrong, and the fix is to adjust the instrument rather than to argue with the number.

How Do You Score Without an Inventory System?

Most small and mid-sized programmes have no purchasing system to draw from, and the answer is not to postpone the scorecard until one exists. A single flat file with one row per lot and one column per dimension is sufficient, provided each cell carries a source reference rather than a judgement. The discipline is not the tooling; it is that every cell can be traced to a document.

Three fields make the file work: the lot identifier, the date of the measurement, and the source reference. With those, a score can be recalculated, audited and argued about on evidence. Without them, the file becomes a set of opinions with numbers attached, which is worse than no scorecard because it lends false precision to a hunch.

One rule keeps the file honest across a long programme: never leave a cell blank. Where a lot closed without an inspection, enter that as a fact rather than as a gap, because a missing inspection is itself a performance signal and a blank cell hides it. Where a document set was never requested, enter the same. A scorecard with blanks is not a partial record, it is a record that has quietly stopped being maintained.

What Does a Good Score Actually Look Like?

A score is only useful once it is banded, because a number on its own produces no behaviour. Four bands are enough. A score in the top band means the source can take more volume and the next order proceeds without discussion. The second means proceed, with the open items tracked. The third means proceed only with a named control in place. The fourth means hold new orders and start the second-source conversation.

The critical design detail is that each band has a defined action attached, written down before the first score is calculated. A band without an action is a number that generates a meeting. A band with an action is a rule that removes the need for one most of the time, which is the entire point of building the instrument.

The bands should be set once and then left alone for at least four quarters. The temptation is to move a threshold when a score lands just below it, and every such adjustment removes a little of the instrument authority. Where a threshold looks wrong after a full year, change it openly at the annual review, record the change and the reason, and apply the new thresholds prospectively rather than restating history.

How Often Should the Score Be Recalculated?

On two cadences, because the dimensions move at different speeds. Quality, schedule and documentation are lot-level and should be entered as each lot closes, so that the score is current whenever a decision arises. Communication and change discipline are continuous and should be reviewed quarterly, because a single slow reply is noise and a quarter of them is a pattern.

The quarterly figure is the one that drives the conversation, and it should be a weighted average of the lots in the quarter rather than a snapshot of the last one. Averaging matters on a low-volume programme: with three lots a quarter, one bad lot swings a snapshot by a third and swings an average by less, and the average is the more honest description of the relationship.

A practical note on the schedule dimension, because it is the one most often measured wrongly. Measure from the trigger event named in the order, which on this programme is written sample approval, and measure to the date the goods were actually ready for the inspection gate. Measuring from the purchase order date instead flatters every lot by the length of the sampling cycle and hides the delays that matter.

Communication is the dimension most often scored badly, because buyers score the relationship rather than the protocol. The fix is to score only what the protocol defines: a first response within the stated window, a substantive answer within the stated window, and an escalation acknowledged within the stated window. Three measurable events, none of which requires a judgement about tone, and all of which are visible in a message log with timestamps.

How Do You Keep the Score Honest When One Person Owns It?

By separating the three jobs. One person enters the data, a second reviews the entries against the source documents, and the decision that follows from the score is made by the person who owns the programme. Where one person does all three, the score drifts toward whatever that person already believes, usually without any intention to distort it.

The second safeguard is to score the source rather than the relationship. Enter what the inspection report says, not what the buyer thinks about the inspection report. Where a buyer genuinely disagrees with a measurement, record the disagreement as a note beside the cell and leave the number alone. Notes accumulate into a pattern; adjustments to numbers accumulate into a fiction.

What Happens When a Score Falls Below the Line?

The band action fires, and the important part is that it fires automatically rather than after a discussion. A score in band C means the next order carries a named control and a review date; a score in band D means new orders are held and the second-source qualification opens. Those are rules, and rules are what make the instrument worth maintaining.

Alongside the action, the review should identify which dimension moved and why. A fall driven by schedule is a capacity conversation. A fall driven by quality is a specification or a process conversation. A fall driven by change discipline is a governance conversation and usually the easiest to fix, because it requires a decision rather than an investment.

Should the Production Partner See the Score?

Yes, with the underlying data. A score shared with the numbers behind it is a management tool for both sides: it turns a general impression into a specific list and it makes the next conversation about two or three cells rather than about the relationship. A score shared without the data is an accusation, and a score withheld entirely is a private grievance that will surface badly at the worst moment.

Share it on the quarterly cadence, in the same meeting as the open corrective actions, and share the definitions with it. Most disputes about a score are disputes about a definition — what counts as a late delivery, what counts as a document error — and definitions agreed in advance remove most of them before they start.

Where a score has been shared for several quarters, an interesting second-order effect appears: the production partner starts forecasting its own score. That is a sign the instrument is working, because it means the definitions have been understood and the behaviour is being managed against them rather than against a relationship. It also means the definitions have to be right, which is the argument for agreeing them slowly and in writing before the first score is circulated.

1. Dimensions

Quality, schedule, documentation, communication, change discipline

2. Weights

30 / 25 / 15 / 15 / 15, revisited after four quarters of data

3. Source per cell

A named document, never a judgement

4. Cadence

Lot-level entry on close; quarterly weighted average for review

5. Bands

A 90+, B 75-89, C 60-74, D below 60, each with a written action

6. Owners

One enters, a second reviews, the programme owner decides

7. Sharing

Shared quarterly with the underlying data and the definitions

One last design choice separates instruments that survive from ones that do not: keep the scorecard to one page visible at a time. A file with five dimensions, five weights, four bands and the current quarter on a single screen gets opened; a workbook with twelve tabs and a history sheet does not get opened after the first quarter, and an unopened scorecard is worse than none because it implies a control that is not running.

The scorecard sits alongside the commercial frame rather than inside it: MOQ 500 pcs per style and colourway, sampling 6-10 working days from a signed-off tech pack, bulk production 35-50 days from written sample approval, inspection at AQL 2.5 under ISO 2859-1 General Inspection Level II, T/T 30/70, quoted FOB Xiamen. Those figures are fixed; what the scorecard measures is whether they are being delivered.

Frequently Asked Questions

People Also Ask