Restaurant Supplier Performance: Review Delivery Reliability Across Outlets

Agree the metric dictionary first

Before counting anything, agree a shared definition with its scope, cut-off, data source and owner. This guide uses a local measure called the on-time, in-full due-line share. It is not a universal on-time-in-full definition, and the business must decide its own rules.

The denominator is every order line that was due in the period, including lines the supplier confirmed it would not deliver. Counting only lines that arrived hides the worst failures.

• Whether you measure lines, whole orders or units.
• How the agreed delivery window works, including early arrivals.
• How partial deliveries, acceptable quantities and substitutes are treated.
• How cancellations, returns and order cut-offs are handled.

Protect the original commitment

Record why a deadline changed. Changes made by your own team, changes made by the supplier, and changes approved before the deadline are different things. Keep the original commitment and record any revised version with its effective approval. Do not quietly move a date to erase a miss.

Received is not the same as accepted. Quality and food controls at receiving stay exactly as they are; nothing should be accepted to improve a score.

Choose whether the review reports performance against the original commitment, the latest properly approved commitment, or both. State that choice in the dictionary and retain the change history. An early arrival outside the agreed receiving window needs the same explicit treatment as a late one; do not assume any earlier delivery meets the agreed service.

Worked example: a fixed cohort of 20 lines

In this original illustration, 20 order lines were due across the outlets during a defined week, and every outcome was checked at the agreed cut-off. Of these, 16 were accepted in full within the agreed window. Two were eventually accepted in full but late. One arrived on time but short, and one was confirmed as not received.

The on-time, in-full due-line share is 16 of 20, or 80%. Leaving out the line that never arrived would give 16 of 19, about 84.2%. That changes the denominator, so it is not valid for this measure.

The status hierarchy used here first asks whether the line met the agreed on-time, in-full test, then records the reason separately. In this example the four outcomes happen not to overlap, but in practice a delivery can be both late and partial; the reason field captures that.

Handling partial and uncertain lines

Take one illustrative line: 100 units ordered, 80 accepted by the deadline and the remaining 20 accepted the next day. The line was eventually complete, but it fails the original deadline test. It is still one line; do not count each partial receipt as a separate order line.

If the evidence for a line is unknown, it is neither a pass nor a fail. Mark the share as incomplete until the line is checked, or report coverage separately. A confirmed non-delivery is different: it is a known failure, not missing data.

If the cohort is zero or any count is missing, leave the share blank. The category counts must be whole numbers and must add up to the cohort.

For a broader register, make outcome categories mutually exclusive and keep timing and quantity reasons alongside them. A short receipt can be assigned to the short category while its reason record also notes lateness. Include an explicit other-known-failure category where needed and a separate unknown count. Unknown cases keep the overall result incomplete until evidence is checked; they must not disappear from the cohort.

Read the number in context

One percentage should not rank suppliers. Twenty lines is an illustration, not an adequate sample for every decision. Disclose differences in order size, product mix and service scope, and list any cancelled lines that were excluded.

Track quantity reliability, quality failures and the supplier's responsiveness and corrective actions as separate things. A high delivery share does not erase a critical incident.

Review fairly with the supplier

Before blaming the supplier, check your own side. Was the order changed late? Did receiving record the delivery on time? Then review the findings together. For each issue, record its severity, the evidence for the cause, an owner, an action, a due date and when it will be rechecked.

Use receipt evidence tied to the actual agreed window, rather than an invoice date or the time someone later entered the record. If one outlet repeatedly records deliveries late, investigate that recording process separately from the supplier arrival. Share the relevant factual line history for review, then check whether the agreed action changed the next comparable deliveries.

Keep an approved backup sourcing process in place for critical items so that a performance problem does not become a service failure while it is being resolved.

Guidance on supply chain management from the US National Institute of Standards and Technology describes using supplier metrics and scorecards to measure performance and improve coordination. It also describes segmenting and evaluating suppliers as part of procurement strategy. That is general manufacturing context, not restaurant law or a metric standard, and it does not prescribe the formula used here.