Open Purchase Order Value Is a Commitment, Not a Cost
Why this matters
Open purchase order value is the one purchasing number nobody questions, because it looks like a balance and balances look factual. It is not a balance. It is two different things added together, one of which is an intention nobody has acted on, and it carries no time window at all, which makes almost every comparison a shop reaches for with it invalid. Worse, the headline is the least useful thing in it. The money the figure exists to protect is not in the total, it is in the handful of orders that were sent, were never received, and that nobody is chasing - and those are exactly the rows a single summed number hides.
What the number is made of
- Drafts: orders raised in your records and not transmitted to anybody. A shopping list with a header on it.
- Submitted: orders the supplier has, not yet received into stock. Money committed and a delivery somebody is counting on.
- Scoping: none. It is a live snapshot of whatever is open right now. It has no start date and no end date.
Cancelled orders sit in neither half, and neither do received ones. So the total answers exactly one question: what is outstanding at this moment. Every other question people ask it has a window in it, and it does not have one to give.
A draft and a submitted order are different objects with different jobs
Summing them is like adding the jobs you quoted to the jobs you booked. Read them apart and each half becomes usable.
Drafts are a planning queue, and they should be read as a count and an age, not as a value. Nobody owes anybody anything. What a draft tells you is that somebody intended to buy something and stopped. A draft older than about 10 calendar days is one of three things: a decision nobody made, a job that moved or died, or a price somebody is still waiting on. Work the queue weekly and either transmit it or kill it. A draft that survives a month has already failed at being a plan.
Submitted orders are a liability and a delivery expectation, and they are read as value against terms and as dates against your schedule. The value half tells you what falls due over your payment terms once it lands. The date half is the part a field shop actually lives on: which of next week's jobs depend on material that is still somebody else's.
Those are two reviews with two audiences. Rolling them into one figure means the office reads it as money and the dispatcher reads it as delivery, and both are reading the wrong half.
The snapshot has no window
A level measured at an instant cannot be divided by a flow measured over a period without naming the period, and this is the single most common misuse of the figure: comparing the open order total to a month of parts spend, or to a quarter of it, and concluding the shop is over-committed or under-committed. Change the window and the answer changes, with nothing about the business having moved. That mismatch has its own article rather than another derivation here.
What the submitted half CAN legitimately do is sit against its own trailing history. Submitted value expressed as weeks of your normal parts consumption is a fair reading as long as you say which weeks, and it is most useful as a trend. A shop whose submitted value climbs from about two weeks of its stated parts consumption to about four over a quarter, with no change in workload, is not buying ahead. It is accumulating orders that are not arriving. Trend the draft half on its own: a draft pile that climbs is a planning queue nobody is working, which is a different fault with a different fix.
The cut that finds the problem: age, from the right date
The one slice of this number that reliably pays is age, and the datum matters more than the threshold. Age a submitted order from its EXPECTED RECEIPT date, not from the date it was placed. An order placed three weeks ago on a four-week lead time is not late; an order placed four days ago on a next-day item is. Measuring from the order date puts those two in the wrong order.
A starting set of triggers, to be tuned to what your suppliers actually do. Every figure below is calendar days past the expected receipt date, and stating the unit is not pedantry: a ladder that mixes business days and calendar days cannot be cut into aged bands afterwards, because the band edges stop lining up with the triggers that set them.
- Stocked item, 5 calendar days past expected: somebody calls. Not an email.
- Special order, 10 calendar days past expected: somebody calls, and the job depending on it gets looked at in the same call.
- Anything past 30 calendar days with no contact recorded: it stops being a chase and becomes an investigation, because at that age the order is as likely to have arrived as to be missing.
One field decides whether any of this works. An order with no expected receipt date recorded cannot be aged, so it drops out of the only cut that finds anything, and the cut then improves the less you record. That shape shows up across a lot of shop metrics and has its own article; here the practical form is simply that the expected date is not an optional field, and an order transmitted without one gets a date put on it the day it goes.
One aged row, two opposite answers
Take a shop with 11 open orders, headline value indexed to 1.00.
- Drafts: 4 orders, 0.46 of the headline value. That is 36 percent of the open rows carrying 46 percent of the headline, and the only thing that gap earns is the decision to lift them out of the total before reading it: one speculative seasonal order is most of the 0.46 and nobody is owed any of it. Then read the four the way the rule above says, as a count and an age. Two are past 10 calendar days, which is the transmit-or-kill point, and one of those two has been sitting five weeks, which is a draft that has already failed at being a plan.
- Submitted: 7 orders, 0.54 of the headline value. Inside that: 3 orders still within their expected date at 0.21, 2 orders between 5 and 10 calendar days past expected at 0.09, and 2 orders more than 30 calendar days past expected at 0.24. Those three add to 0.54.
The two badly aged orders are 2 of the 7 submitted by count, 29 percent, and 0.24 of the 0.54 committed by value, 44 percent - so nearly half the shop's genuine commitment is sitting in two rows nobody has touched. As a share of the headline total the same two rows are 24 percent, which is the number an owner glancing at the total would never guess.
They resolve in opposite directions, and one free test separates them.
The test: has anything on that order been ISSUED to a job? Not "is it on the shelf", which requires somebody to walk out and look, and which a part already installed will fail anyway. Issued to a job.
Order A, 0.15 of the headline, 34 days past expected, stocked items. Two of its lines have been issued to jobs in the last three weeks. So the material arrived. What never happened was the receipt. The order is not late, the record is.
Order B, 0.09 of the headline, 41 days past expected, one special-order assembly. Nothing issued, nothing on the shelf. The supplier put it on backorder in the second week and the confirmation went unread. This one is late, and the job it was raised for is still waiting.
Same age band, same list, opposite problems, and the summed total says neither.
What each one costs while it sits
They are not equally expensive, and neither cost is the value on the order.
The unrecorded receipt (Order A) leaks in four directions at once. Stock value is understated by the amount of the receipt, so the denominator of your turnover figure is wrong and turns read high. The supplier invoice has nothing to match against, so it either sits unapproved past terms, which is how a shop with cash in the bank ends up on a credit hold, or it gets paid a second time when a statement copy turns up. The parts read as unavailable, so somebody reorders them. And the job the parts went onto carries no cost for them, so that job's margin reads better than it was.
The unchased backorder (Order B) costs schedule, not money. The commitment is real but the exposure is a job held for material that was never coming, on a promise the customer is still holding you to. The recovery is not a better number, it is a phone call: confirm the date, and if there is not one, source it elsewhere and cancel. Which is where the next trap lives, because a cancelled order disappears from the supplier's record entirely.
The order that half arrived
The composition treats an order as open or not open, and a supplier's truck does not work that way. A partial delivery leaves a row that is genuinely both, and how your records handle it decides two separate things.
- If a partially received order stays open at its FULL value, the headline overstates the commitment by everything already delivered, and that material appears in two figures at once: as open commitment and as stock on hand. It is the one case where the open total and the inventory value really do overlap.
- If it stays open at its REMAINING value, the total is honest, but the row now ages against an expected date most of it already met. An order 90 percent delivered and waiting on one backordered fitting lands in the aged bucket looking identical to one where nothing came at all.
Neither treatment is wrong, and both need the same second field: the date of the LAST receipt against that order. An order with a receipt three days ago is a live supply relationship with one line outstanding, worth a note and no phone call. An order past its expected date with no receipt against it at all is the case in the section above. Same age, different animals, and the receipt date is the only thing that separates them.
Which numbers see an open order, and which do not
| The number | Does an open order appear in it? | What follows from that |
|---|---|---|
| Job parts cost and job margin | No. Not until received and issued | A job quoted off an ordered part carries no parts cost at all until the part lands, so its margin flatters until then |
| Inventory value | No, not until received, except for a partially received order left open at full value | Apart from that one case the two are disjoint, and even then they cannot be summed into a meaningful position |
| Cash out | No. Not until the supplier invoice is paid, on terms | Open value is an obligation with a date, not an outflow; the timing gap is the whole reason to read it against terms |
| Parts spend for the period | Drafts no, submitted usually no until received | Whether a submitted order counts here depends on whether your parts spend is measured on receipt or on payment. Name which, or the comparison drifts |
| Committed spend by supplier | Yes, the submitted half | This is the input to a supplier concentration read, which is a different article |
| Order fulfilment rate | Submitted, yes. Drafts, no. Cancelled, neither | An order cancelled after a supplier failure vanishes from both halves of that rate |
| The delivery promise on a scheduled job | The submitted half only | The one thing the number is genuinely for, and the one nobody reads it as |
Read the map top to bottom and the honest summary is that open purchase order value touches none of the four money numbers a shop already runs on - job margin, inventory value, cash out and parts spend - and shows up only in the three purchasing reads at the bottom, which is exactly why it is safe to leave unexamined for a year. The cost of doing that is not in the total. It is in the two rows.
References
- See related: What a Purchase Order Actually Obliges Both Sides To, for what the submitted half commits you to
- See related: Purchase Order Fulfilment Rate and the Orders It Never Counts, for what cancellation does to a supplier's record
- See related: Days Sales Outstanding and the Mismatch Inside It, for a snapshot measured against a windowed flow
- See related: SLA Compliance Only Counts Jobs That Carry a Deadline, for a cut that improves the less you record