A supplier scorecard should help a buyer decide what to discuss, what to change and where a backup source is needed. Start with four measures: fill rate, delivery reliability, price variance and substitutions. Keep the underlying orders visible so the score is explainable.
Choose a period and preserve the original promise
Use a period long enough to include representative orders and show the sample size. Five orders can expose a problem, but they cannot establish a stable long-term reliability estimate. Report missing receipts or unverified prices instead of silently excluding them.
Record both the original order and accepted changes. Measuring only against a revised promise hides what changed. Measuring only against the original request can unfairly classify a buyer-approved schedule change as supplier failure. Show both where the distinction matters.
1. Fill rate: units and lines are different
For an item with one consistent unit:
Unit fill rate = accepted original-item units received / original-item units ordered × 100
Define the measurement cutoff and how you treat permitted substitutes. For a mixed catalog, avoid adding incomparable units such as kilograms, bottles and pallets into a misleading single ratio. Use item-level rates, a consistently defined line-fill measure, or a disclosed weighting method.
An illustrative order for 100 bottles with 90 accepted bottles has 90% unit fill. That does not mean five of fifty SKUs were missing: a unit rate cannot establish the number of affected lines.
Track late balances separately. A supplier may ultimately deliver every unit but leave the business short when it actually needed the stock.
2. Delivery reliability: compare dates explicitly
Measure the elapsed time from order release to usable receipt, and compare the receipt with the agreed delivery window. One simple measure is:
On-time rate = receipts within the agreed window / receipts due in the period × 100
Keep the window in the definition; a café's morning delivery requirement is different from a retailer's one-day tolerance. For partial deliveries, show the first receipt and completion date.
Also review the distribution of actual lead times. An average of four days can hide a mix of reliable three-day deliveries and occasional ten-day delays. This information can improve planning, but a percentage of late orders is not a direct formula for safety stock.
3. Purchase price variance: distinguish changes from errors
This guide uses positive PPV for a favorable difference:
PPV = (baseline unit price − invoiced unit price) × comparable quantity
For 100 units with a baseline of $5 and an invoice price of $5.20, PPV is negative $20, or 4% unfavorable against $500 expected spend. Use consistent currency, pack conversion and treatment of freight and tax.
Keep two comparisons when needed: invoice versus accepted price, and accepted price versus the original plan. An invoice can be correct while the cost has risen. A difference can result from an agreed surcharge or changed specification rather than a supplier error.
A cost increase only affects the relevant purchases. Do not apply one supplier's percentage to the entire company's COGS unless the supplier covers that whole cost base.
4. Substitution rate: measure the decision as well as the event
Orders with at least one substitution / total relevant orders × 100
Separate proposed, accepted and rejected substitutions. Include which item changed, the buyer's decision, pack size, usable yield and receipt. A substitute may preserve service or create a quality issue; its existence alone does not tell you which.
For ingredients and components, a supplier proposal does not approve recipe suitability, engineering specifications or regulated use. The responsible operator must make that decision. An accepted mapping can then support the inventory and cost calculation.
Use OTIF as a summary, then inspect its parts
On-time-in-full requires both timing and completeness under the agreed definition. If 18 of 20 orders due were complete and on time, OTIF is 90%. It does not tell you whether the other two were one day late, mostly missing or both.
Do not increase a buffer by the “OTIF gap,” average two suppliers' lead times and assume risk has disappeared, or authorize automatic purchasing from a good score alone. Planning and approval decisions need the actual demand, delay, capacity, item-criticality and spending context.
Turn the review into a supplier conversation
Choose thresholds with the supplier and the internal users who depend on delivery. They are service targets, not universal industry benchmarks. Review critical or deteriorating relationships more frequently than stable low-risk ones.
For each issue, bring an example and ask for a concrete change:
| Finding | Useful next action |
|---|---|
| Repeated shorts on one critical item | Confirm capacity, ask for earlier shortage notice, qualify a backup |
| Late arrivals on one delivery route | Agree a realistic window and review the order cutoff |
| Invoice price differs from accepted price | Resolve the lines and document the credit or corrected price |
| Frequent rejected substitutes | Record permitted alternatives and who can approve them |
A backup supplier is useful only if its items are qualified and its capacity, pack sizes, prices and lead times work for the business. An 80/20 split is one possible trial allocation, not a standard that guarantees resilience. Two vendors can share the same upstream source or freight bottleneck.
What to collect in the buying workflow
The supplier order record can hold the original PO, supplier messages, accepted revisions, receipt evidence and invoice references. These records reduce reconstruction, but a scorecard still needs consistent definitions and complete inputs.
In a LineNow evaluation, ask the team to demonstrate the exact report or export used for each measure. Recalculate several orders manually, including a partial receipt and a rejected substitute. That establishes more trust than a polished score with an unclear denominator.