Build a NetSuite supplier scorecard from declared promise dates, receipt quantities and a stable denominator before ranking vendors. Show original-promise performance separately from the latest operating promise, and define how partial receipts count. A percentage without those definitions can reward date changes or penalize suppliers for internal recording delays.
This guide focuses on historical delivery-score definitions. NetSuite's Supply Chain Control Tower vendor predictions, procurement dashboards and custom analytical reports are different tools. Confirm their features and input rules before treating a native prediction as the company's on-time-in-full score.
Decide whether a scored obligation is a PO, PO line, scheduled delivery or another agreed unit. An order with ten lines should not silently carry the same meaning as one line simply because both have a PO number.
For a delivery schedule, preserve the quantity and promise associated with each scheduled portion. If a supplier agreed to two deliveries, scoring the first partial against the entire order can be unfair. If it agreed to one complete delivery, treating every partial receipt as an independent success can be misleading.
Document the choice beside the metric and keep it stable across reporting periods. A change in denominator can alter the score even when supplier behavior is unchanged.
Capture the original agreed promise and subsequent authorized revisions where the system and process support them. Record who requested a date change and why.
Original-promise performance measures reliability against the initial commitment. Current-promise performance helps the buyer manage the latest plan. Both are useful, but they should not replace each other silently.
If historical promise dates were overwritten and cannot be reconstructed reliably, disclose that limitation. Do not manufacture an original baseline from the final receipt date or assume a default PO date was a genuine supplier commitment. Begin preserving the required evidence prospectively.
On-time needs a due-date rule, a receiving event and any agreed tolerance. In-full needs an ordered or scheduled quantity and a policy for accepted quantity, overdelivery and cancelled remainders. Quality needs its own acceptance evidence and denominator.
The business may combine them into an OTIF measure, but it should still expose the components. A supplier that delivers everything late needs a different discussion from one that delivers on time with repeated shortages.
Define whether receipt means physical arrival, posted item receipt or quality-released stock. If the score uses posted receipt date, review late entry as a data-quality issue. The buyer should not accuse a supplier of lateness based on an internal posting delay that has not been investigated.
A supplier has ten scored delivery lines, each for ten units and each due on the same agreed date. Eight lines arrive complete on time. A ninth delivers five units on time and five later. The tenth delivers all ten units late.
Under a line-based rule requiring the full quantity by the promise date, eight of ten lines pass, giving 80 percent OTIF. On-time units total 85 out of 100, giving 85 percent unit delivery by the due date. After the late quantities arrive, eventual unit fill is 100 percent.
All three numbers can be correct. They answer different questions. If the due dates for the late lines are moved to their actual arrival dates without retaining the original promise, the apparent result can improve without any change in supplier behavior. These are hypothetical figures, not an industry benchmark or a customer outcome.
| Definition | Required decision |
|---|---|
| Population | Which vendors, items, entities and due periods are included? |
| Grain | PO, line or scheduled delivery? |
| Promise | Original, current or both? |
| On-time rule | Which event and tolerance determine success? |
| In-full rule | How do partials, overdelivery and cancellation affect quantity? |
| Quality rule | Which accepted or rejected quantities count? |
| Missing data | Excluded, blocked or separately reported? |
| Review owner | Who resolves disputes and approves publication? |
This contract is the reporting specification. A dashboard should implement it, not invent the definitions after the numbers are visible.
One PO line can have several receipts and several bills. Joining those records directly can repeat ordered quantity or multiply receipt rows. Test the analytical grain before aggregating by vendor.
Reconcile one simple full receipt, one partial delivery and one returned quantity manually. Compare the line-level expected result with the report output, then reconcile population totals.
Keep delivery analysis distinct from invoice volume. A vendor that bills twice for one delivery should not receive two delivery observations. Likewise, a missing item receipt may indicate a different receiving process rather than proof that goods never arrived. Mark the evidence gap and investigate it before scoring failure.
Define treatment for buyer-requested delays, cancelled orders, supplier replacements, drop shipments, service purchases and returns. Some populations may need separate scorecards because they do not share comparable receipt evidence.
Do not exclude poor outcomes merely because they are inconvenient. An exclusion needs a reason that would be applied consistently to every supplier. Retain counts and quantities excluded from each metric so readers can assess coverage.
For very small samples, show the observation count prominently. One late delivery out of two and 50 late deliveries out of 100 both produce 50 percent, but they support different levels of confidence about a recurring pattern.
Supply Chain Control Tower's documented Vendor Delivery Performance process analyzes orders and receipts over a configured interval and produces predicted early or late days and risk-confidence information. It also exposes calculation timing and observation counts.
Those outputs support planning risk review. They should not automatically be relabeled as the custom OTIF definition in this guide. Preserve the source name, calculation time and intended use when combining them in a procurement dashboard.
If a prediction disagrees with the historical scorecard, compare time windows, populations, date rules and quantity treatment first. Two different metrics can both be internally consistent while answering different operating questions.
Before sharing a score externally, have the buyer review the underlying exception lines and resolve obvious recording errors. Present the definition, period, sample size and a few concrete delivery examples.
Ask for a corrective plan tied to the observed cause: acknowledgment discipline, packaging quantity, production capacity, transport or documentation. Do not infer a supplier's intent or financial condition from a delivery percentage.
Commercial consequences such as reallocating business, imposing penalties or ending an agreement require authorized business and legal review. The report is evidence for that decision, not an automated verdict.
Save the scored population, metric version, extraction time and approved exclusions. If data changes later, identify whether the score was restated and why. This is especially important when a quarter-end review influences contract discussions.
For unclear joins or unreliable date history, CuriousRubik's NetSuite support services can help scope a reporting validation. Start with the partial-receipt example and an agreed metric definition before building a larger dashboard.
No. OTIF asks whether the defined obligation was complete by the agreed time. Eventual fill can reach 100 percent after late receipts while OTIF remains lower. Show both with their definitions.
Preserve both when reliable history is available. Original promises support reliability analysis, while current promises support today's plan. Record the reason and owner for changes so the report does not reward silent deadline movement.
Apply the declared grain and agreement. A line requiring full delivery by one date fails line-based OTIF if only part arrives, while unit-based on-time quantity can still receive partial credit. Do not mix those denominators.
No. It may reflect missing data or a different receiving process. Investigate the evidence and disclose the gap before assigning failure or making supplier-specific conclusions.
No. The native prediction uses its own configured analysis and outputs. Keep its purpose, timing and population separate from the historical scorecard definition, then compare them only after their assumptions are understood.