Purchase Order Fulfilment Rate and the Orders It Never Counts

Why this matters

Fulfilment rate is the number a shop reaches for when it wants to know which supplier to trust, and as normally composed it will hand you the wrong one. The rate is computed over orders that were received and orders that are still waiting. An order that was cancelled sits in neither half, so when a supplier comes back with no stock and no date and you source the part elsewhere, that failure does not lower their score. It removes itself from their score. The supplier who forces you to cancel five times a quarter can finish the quarter with a better record than the one who delivered everything and happens to have four orders in transit.

What the rate is made of

• Numerator: orders RECEIVED.
• Denominator: orders received plus orders still submitted and not yet received.
• Population: orders RAISED inside the window.
• Excluded from both halves: drafts, which were never sent to anybody, and cancelled orders, whatever caused the cancellation.

Excluding drafts is correct. A supplier cannot be judged on an order they never saw. Excluding cancellations looks like the same kind of hygiene and is the opposite: it is the removal of the one event that most reliably marks a supplier failure.

What "received" has to mean, or the numerator drifts

The numerator counts orders, not lines, so a partial delivery has to be forced into one bucket or the other and the choice decides the supplier's score.

• Left open until every line lands, an order delivered 90 percent complete and short one fitting scores as a total failure, which overstates the harm considerably. A supplier who reliably short-ships one line an order will bottom out on this rate while keeping your jobs running.
• Closed off as received with a short-ship note, the same order scores as a clean success, which understates it, and the shorted line silently becomes somebody's problem on a job.

Pick one and hold it; a mixed convention makes the rate incomparable between suppliers and between quarters. For a field shop the defensible choice is the second, closed as received with a separate count of short-shipped lines, because the rate exists to decide who gets the next order and a supplier who gets the truck to you is not in the same category as one who sends nothing.

The exclusion runs backwards

Think about what actually causes a purchase order to be cancelled in a field shop. The job moved. Somebody keyed it twice. The price came back above quote. And, most often in any trade with supply volatility, the supplier answered with a backorder and no date, so the buyer killed the order and sourced the part somewhere else that afternoon.

That last case is a supplier failing at the only thing you buy from them for, and the metric's response is to delete the row. Which means the rate has a property nobody intends and nobody would defend if it were written down: the faster you react to a supplier's failure, the cleaner that supplier's record looks. A buyer who cancels promptly and re-sources is doing the right thing operationally and erasing the evidence at the same time. A buyer who leaves a dead order open for six weeks is doing the wrong thing and is the only reason the number ever moves.

Two suppliers, one rate, opposite truths

One quarter, two suppliers, 20 orders raised with each.

Supplier A. 5 orders cancelled, all five because A came back with no stock and no date and the shop bought the part elsewhere. Of the remaining 15: 14 received, 1 still submitted. That 1 was raised eight days before the quarter closed on a fourteen-day lead time, so it is not late.

Rate as composed: 14 divided by (14 plus 1) is 14 over 15, which is 93 percent.

Supplier B. 0 cancelled. 15 received, 5 still submitted. Four of those five were raised in the last two weeks of the quarter and sit inside the lead time B quoted. The fifth is genuinely overdue.

Rate as composed: 15 divided by (15 plus 5) is 15 over 20, which is 75 percent.

So the metric puts A ahead of B by 18 points, 93 against 75, and the shop that reads only this number moves more volume to the supplier that left it standing on five jobs.

The honest version

Two corrections, and they pull in opposite directions, which is why applying only one makes things worse.

First, put supplier-caused cancellations back in the denominator as failures. The question the rate exists to answer is "if I send this supplier an order, do I get the part", and a cancellation forced by their answer is a no.

Second, take out the orders that could not physically have arrived yet. An order raised four days before the window closed on a two-week lead time is counted as a non-receipt against a supplier who has done nothing wrong. This is the ordinary cohort problem, outcomes counted in the same window as the opportunities, and it punishes every supplier and punishes the busiest ones most. It has its own article; here it is one exclusion rule: an order still inside its quoted lead time at the close of the window leaves the denominator, and it rejoins the population next window.

Corrected, on orders raised in the quarter:

• Supplier A: 20 raised, 1 removed as still inside lead time, denominator 19. Received 14. That is 14 over 19, or 74 percent, and the entire 5-order shortfall is cancellations A caused.
• Supplier B: 20 raised, 4 removed as still inside lead time, denominator 16. Received 15. That is 15 over 16, or 94 percent, and the single failure is one genuinely overdue order.

The composed rate ranked A ahead of B by 18 points (93 against 75). The corrected rate ranks B ahead of A by 20 points (94 against 74). Both figures are percentages of orders raised at supplier level over the same quarter, so the two rankings are directly comparable and one of them is simply wrong.

Run the floor against the example. Do not rank a supplier on fewer than 10 assessable orders in the window; roll the window forward a month at a time until it clears. A finishes at 19 and B at 16, so both qualify. A supplier you buy from twice a quarter cannot be ranked on this rate at all, and for those the useful record is not a percentage, it is the notes.

What changes as a result

The corrected numbers do not say drop Supplier A. They say something more specific: A's failures are concentrated in cancellations, which means they are failures of AVAILABILITY rather than of delivery, and availability failures are fixed by a second source on those specific parts rather than by moving the whole account. Pull the five cancelled orders, list what was on them, and find out whether the same three part families keep appearing. They usually do.

That is a different decision from changing primary supplier, which is a yearly review with price, terms, counter service and location in it, and is covered elsewhere. The rate feeds that review. It does not replace it.

Fulfilment and on-time are two rates, on two populations

Shops routinely run these together and they answer different questions on different denominators. Fulfilment asks did it come at all, over orders raised. On-time asks did it come when promised, over orders RECEIVED. A supplier can sit at either end of one and the opposite end of the other, and the pair is far more decision-useful than either alone.

Carry the same quarter forward. Of Supplier A's 14 received orders, 13 arrived on or before the expected date, which is 93 percent of orders received. Of Supplier B's 15 received orders, 11 did, which is 73 percent of orders received.

Set the two side by side, naming which base each one sits on:

• Supplier A: fulfilment 74 percent of assessable orders raised, on-time 93 percent of orders received.
• Supplier B: fulfilment 94 percent of assessable orders raised, on-time 73 percent of orders received.

These are not two readings of one thing, and they must not be averaged or compared across the diagonal. Together they describe two suppliers working two ways: A says no quickly and then delivers almost everything it accepted, on the day, while B accepts everything and runs late on a quarter of it.

Which you want depends on your work, not on which number is higher. If your jobs are scheduled to a customer promise and a slip means a second trip, B's late quarter is the expensive one, and A's refusals at least leave you time to source elsewhere. If material is bought ahead into stock and a few days does not matter, B is the better account by a distance and A's availability gap is the one that hurts. Read as a pair, the supplier conversation writes itself: A gets asked about stocking depth, B gets asked about dates.

What a good rate looks like

There is no defensible published benchmark for supplier fulfilment in a small field service shop, and a number invented to sound authoritative here would do real damage, because a shop would move an account over it. Two honest comparators, both of which you own:

• Your own trailing four quarters for that supplier. Direction matters more than level. A supplier sliding from the low nineties to the mid seventies over three quarters is telling you something regardless of where either figure sits against anyone else's shop.
• The spread between your suppliers in the SAME quarter. This is the stronger of the two, because the same quarter carries the same supply conditions, the same freight problems and the same seasonal crunch for all of them. When one supplier is 20 points below the others in a bad quarter, the quarter is not the explanation.

Classify the cancellation, or the rate stays broken

A reason on a cancelled order is what makes the corrected rate possible, and a free-text note will not do it. Use a short fixed list, because the value is in being able to add them up.

That last row is the one that decides whether any of this holds up over a year. A blank reason is not neutral, because the cancellations most likely to be left blank are the awkward ones, and the default has to run against the party the blank protects. Make it count as a supplier failure and the reasons get filled in within a month, by the same buyer who would otherwise never see the point.

And the honest caveat on the whole measure. None of this catches the supplier who never fails an order because the counter staff quietly tell your buyer not to bother ordering that item from them. That conversation leaves no order, no cancellation and no row anywhere, and the only trace it leaves is a part you used to buy there and now buy somewhere else. The rate is a floor on supplier reliability, never a ceiling.

References

• See related: Choosing a Primary Supplier, for the yearly decision this rate feeds
• See related: Estimate Conversion Rate and the Cohort Problem, for outcomes counted in the same window as the opportunities
• See related: Open Purchase Order Value Is a Commitment, Not a Cost, for the aged submitted orders this rate leaves open
• See related: Building Supplier Leverage as a Small Shop, for what to do with a supplier whose failures are concentrated in a few parts