“This vendor is always late” is a common complaint and a surprisingly hard one to verify without actually looking at the data. Measuring supplier on-time delivery performance properly means moving past impression and into a small set of metrics — used carefully, with their limitations acknowledged.
The core metrics
Contractual date vs. actual date
The foundation of every other metric here. Without both dates recorded consistently for every PO, nothing downstream is reliable.
On-time percentage
The share of purchase orders where the actual date was on or before the contractual date. Simple to calculate, but only meaningful with enough purchase orders behind it — more on this below.
Average days early or late
On-time percentage alone can hide a lot. A supplier who’s “on time” 90% of the time but catastrophically late the other 10% has a different risk profile than one who’s consistently a few days late across the board. Tracking average variance — and its spread — adds the context a single percentage misses.
Sample size
Not a performance metric itself, but essential context for every other number on this list. Discussed in detail below.
NCR count
The total number of non-conformance reports raised against a supplier’s purchase orders.
NCRs per PO
NCR count alone doesn’t account for volume — a supplier with 3 NCRs across 3 POs has a very different quality record than one with 3 NCRs across 30 POs. Normalizing by PO count makes suppliers comparable regardless of how much business they’ve done.
Average NCR closure time
How long, on average, non-conformances stay open before resolution. A supplier who raises occasional NCRs but resolves them quickly is behaving differently than one whose NCRs linger.
Inspection cycle time
The time from “ready for inspection” to inspection completion. Long cycle times can indicate either an overloaded inspection process or a vendor whose work frequently needs rework before it passes.
The sample size problem
This is the part of supplier performance measurement most likely to be done badly, and it’s worth stating plainly: a supplier with a single purchase order on record cannot be meaningfully compared to a supplier with fifty.
If Supplier A delivered their one PO on time, that’s a 100% on-time rate. If Supplier B delivered 47 of 50 POs on time, that’s 94%. Read as raw percentages, Supplier A looks better. In reality, Supplier A’s single data point tells you almost nothing about what to expect next time, while Supplier B’s 50-PO history is a genuinely informative track record.
Any supplier performance report should show the sample size next to every percentage, and any comparison between suppliers should account for the fact that a small sample can swing wildly on a single good or bad delivery, while a larger sample smooths that noise out. Treating a 1-PO “100% on-time” supplier as equivalent to, or better than, a 50-PO “94% on-time” supplier is a statistical error that leads to real, avoidable sourcing mistakes.
What counts toward the calculation
On-time percentage is: (number of completed POs where actual date ≤ contractual date) ÷ (total number of completed POs) × 100. Two definitions matter here, and skipping them is how two people calculate “the same” metric and get different numbers:
- Completed means the PO has an actual (confirmed, not forecast) delivery date on record. Open POs still in progress aren’t part of the denominator yet — including them would let a batch of on-schedule-but-not-yet-delivered POs inflate a rate that hasn’t actually been earned.
- Partial deliveries and cancellations need an explicit rule, not a default one. A common approach: a partial delivery counts as on-time only once the full PO quantity has been received against the contractual date; a cancelled PO is excluded from the calculation entirely rather than counted as either a success or a failure. Whatever the rule, apply it consistently — otherwise a supplier’s rate depends on how convenient a given case was to categorize.
Worked example: a supplier has 12 completed POs. 9 delivered on or before the contractual date, 2 delivered late, and 1 was a partial delivery where the remaining balance arrived after the contractual date (counted as late under the rule above). On-time rate: 9 ÷ 12 = 75%.
Putting these metrics into a vendor review
A periodic vendor review — quarterly is common for active suppliers — is where these metrics earn their keep. Rather than an impressionistic conversation about how a relationship “feels,” a review grounded in actual numbers can address specific, verifiable patterns: an on-time rate that’s declined over the last three quarters, an NCR-per-PO ratio that’s climbing, an inspection cycle time that’s grown as order volume increased.
This also changes the nature of the conversation with the supplier itself. “Your last few deliveries felt late” invites a defensive response. “Your on-time rate across the last 12 POs is 67%, down from 91% the year before” is a specific, checkable claim that’s harder to dismiss and easier to have a genuinely productive conversation around — including, sometimes, a legitimate explanation the data alone wouldn’t show.
What these metrics don’t capture
It’s worth being honest about the limits here too. On-time percentage and NCR counts say nothing about the quality of communication during a delay, how responsive a vendor is to a genuine emergency, or how a supplier behaves on the purchase orders that don’t generate a formal NCR but still involve friction. A vendor with excellent on-time numbers who’s consistently difficult to reach for a status update is a real, different kind of risk that this particular set of metrics won’t surface on its own — worth tracking as a separate, more qualitative note alongside the quantitative record, not a reason to discard the quantitative record itself.
Using these metrics well
None of these numbers are useful as a one-time snapshot. Their value comes from tracking them over time, per supplier, and reviewing them before major sourcing decisions — renewing a contract, awarding a new PO to an existing vendor, or deciding whether a supplier’s pricing advantage is worth their delivery risk.
They’re also not a substitute for judgment. A supplier with a rough on-time record during a documented industry-wide material shortage is a different situation than one with a rough record during normal conditions. The metrics tell you what happened; understanding why still requires the context only an expediter or buyer who’s been tracking the relationship actually has.
Where this data comes from
None of these metrics can be calculated after the fact from memory. They depend on contractual and actual dates, NCR raised and closed dates, and inspection completion dates being recorded consistently, PO by PO, as part of normal expediting work — not reconstructed at review time from scattered records.
The PO Delivery Tracker Pro edition includes Vendor Performance History across recorded PO data, alongside portfolio forecasting and expediting controls. Use the underlying PO history and sample size when interpreting any supplier percentage rather than treating a headline rate as sufficient on its own.