Supplier Management Software: Why On-Time Scores Differ

Supplier management software can show different on-time scores for the same deliveries when the reports use different promise dates, receipt rules, or groups of orders. Before acting on a percentage, open the contributing purchase schedules and check the denominator. A revised promise must not silently replace the original commitment, and an overdue order must not disappear merely because nothing arrived.

This guide is for small manufacturing purchasing teams evaluating supplier reporting. It works through four fictional delivery schedules and gives you a repeatable software demonstration. For the daily process of chasing commitments and recording receipts, start with our supplier delivery tracking guide.

Define the score before comparing suppliers

A label such as on-time delivery is incomplete. Does it count purchase orders, lines, scheduled releases, receipts, or pieces? Is the deadline the factory's requested date, the first supplier-confirmed date, or the latest revised promise? Does arrival mean a truck at the gate, a verified goods receipt, or material released after inspection?

Write down those choices with the reporting period, cutoff time, receiving location, time zone, and allowed tolerance. A line with three scheduled releases needs three identifiable commitments if the score counts schedules. Three partial receipts against one commitment must not become three separate successful deliveries.

For a named-product example, Microsoft's supplier performance reporting documentation describes filtered delivery measures and visibility into the number of orders and lines in scope. Product labels and calculation rules vary. The test below defines its own policy; it is not a claim about Microsoft's formula or a universal industry standard.

Audit four delivery schedules

Hypothetical demonstration, not a customer result: one supplier has four scheduled commitments, each for 100 pieces of the same component. The first-confirmed due dates fall between October 5 and October 8, 2026. Each deadline is 4 p.m. in the receiving site's America/Chicago time zone. The report is evaluated at 4 p.m. on October 9.

For this test, a schedule passes only when its complete 100-piece quantity has a verified physical receipt by its first-confirmed deadline. There is no grace period, cancellation, return, quantity amendment, or early-delivery restriction. All four schedules stay in the comparison group even if a later promise moves outside the original date window.

• Schedule A, due October 5: all 100 pieces are received at 2 p.m. on October 5. It passes the defined delivery test.
• Schedule B, due October 6: 60 pieces arrive at 2 p.m. on October 6 and 40 at 10 a.m. on October 7. It fails the full-quantity-by-deadline test.
• Schedule C, first confirmed for October 7: the supplier revises its promise to October 9. All 100 pieces arrive at noon on October 9. It misses the first promise but meets the revised one.
• Schedule D, due October 8: no receipt is recorded by the October 9 report cutoff. Receiving confirms that nothing has arrived. It is overdue and remains in the denominator.

Against the first-confirmed dates, one of four schedules passes: 1 / 4 = 25%. A is the pass; B, C, and D are the misses. This is a schedule-based full-quantity delivery measure, not the percentage of pieces received on time.

Explain how the same records produce 66.7%

Now change only the deadline reference to the latest revised promise, while retaining the same four-schedule comparison group. C becomes a pass, so the result is 2 / 4 = 50%. The physical receipts have not improved. The report answers a different question: did deliveries meet the latest promises?

Next filter that group to fully received schedules only. D disappears because it has no receipt. A and C pass, B fails, and the displayed result becomes 2 / 3 = 66.7%, rounded to one decimal place.

The complete bridge is therefore 25% using first promises and all four due schedules, 50% using revised promises and all four, then 66.7% using revised promises and only the three fully received schedules. Two definition changes explain the difference; no additional delivery occurred between calculations.

A received-only view can serve a clearly labeled historical question. It is not a substitute for a due-schedule review that includes outstanding late work. Keep the report's purpose visible instead of presenting different populations as an improvement trend.

Keep original promises and changes visible

Store the internal required date, first confirmed date, and current confirmed date separately. Preserve each change, its recorded time, reason, owner, and supporting communication. If the first commitment was never captured, mark it unknown rather than reconstructing a favorable date from memory.

Supplier delays and buyer-requested changes need different explanations. If your team postpones a delivery because its drawing changed, preserve that event. You may choose an adjusted performance view that excludes an approved buyer-caused delay, but show the exclusion policy, affected schedules, and unadjusted comparison alongside it.

Do not change historical scope quietly. If a commitment is split, amended, or canceled, retain the original schedule identity and the decision trail. Define how that event affects future reports and previously issued snapshots. Our partial-delivery purchase order example covers the separate quantity balances; a reporting adjustment must not create or delete a physical receipt.

Separate delivery performance from usable material

Suppose ten of A's 100 received pieces are still awaiting inspection. Under the example's explicitly physical-receipt rule, A still passes delivery. That does not mean all 100 pieces are available for production or that the supplier passed a quality measure.

Show receiving time, inspection status, released quantity, and any supplier-caused nonconformance separately. If your business instead measures usable material by the required date, name that measure and define how internal inspection delays are treated. Do not silently replace one event with another in the same trend.

Link disputed results to evidence and an owner. A missing receipt record is not always proof that a delivery is absent: it may be an entry delay. D's example explicitly confirms non-arrival. Where evidence is incomplete, show an unresolved data-quality exception and the known population rather than guessing a pass or blaming the supplier.

Make the supplier scorecard explainable

When adding supplier scores to manufacturing dashboard software, display the reporting policy and counts beside the percentage. A manager should be able to open the exact schedules behind the numerator and denominator, including open misses and exclusions.

Keep a saved snapshot of what the review meeting used. If a late-posted receipt corrects a result, distinguish the original view from the restated view and retain approval of the correction. Show the last successful data update, not just the time somebody opened the dashboard.

Compare suppliers only after checking scope and sample size. One pass out of one schedule does not provide the same evidence as a sustained record across many comparable commitments. Different part families, delivery frequencies, or agreed receiving rules may warrant separate views rather than one blended ranking.

Use the score to choose a factual follow-up: clarify a commitment, investigate a recurring delay, correct an internal handoff, or review an unresolved record. It should support a supplier conversation, not substitute for it.

Run seven supplier software acceptance checks

1. Reproduce the baseline: load A through D and show one pass out of four against the first-confirmed deadlines.
2. Change the reference: switch to revised promises and show two out of four, with an explicit label and retained original dates.
3. Expose the filter: apply the received-only view and explain why D leaves the denominator and the score becomes 66.7%.
4. Retain partial receipts: keep B as one schedule with two receipt events, not two independent deliveries.
5. Separate quality: add A's ten-piece inspection hold without falsely making all received material usable.
6. Review exceptions: demonstrate a buyer-requested date change and a missing-data case with reasons, owners, and visible inclusion rules.
7. Export the evidence: retrieve schedule IDs, deadline versions, receipt references, counts, exclusions, and the report snapshot needed to reproduce the result.

These are proposed evaluation criteria, not promises that Ploqy or a named product already provides this exact workflow. Begin with one supplier and one component family. Agree on the measure with purchasing, receiving, and production before asking software to rank a broader supplier base.