Choosing purchase order software can absorb weeks of feature comparison and one five-minute exercise settles most of it. It tests the capability that separates a system from a form, it works in any trial account, and products that fail it differ in kind rather than in polish, which is the distinction a feature matrix cannot make.
The test
Raise an order for ten of something. Record a receipt for four. Enter an invoice for four. Ask what is outstanding on that line. The answer should be six. If the product cannot answer, or answers at document level rather than line level, it tracks orders as documents and cannot support matching, accruals or delivery chasing.
Why partial delivery is the pivot
It is the normal case, not an edge case. Deliveries split, one ordered line gets invoiced twice, and quantities are revised. Document-level tracking turns all of that into exceptions, which teaches people to override rather than investigate, and routine overriding is how a control quietly becomes a formality nobody trusts.
The second test: revisions
Change the price on an approved order and check the previous value is still visible with the date it changed. Systems that overwrite look tidier and lose exactly what a supplier disagreement turns on, which is what the order said on a particular day. Two tests, ten minutes, and they tell you more than any comparison table.
And the third: escalation
Submit an order for approval and then do nothing for a week. Whether it escalates to a named alternative or simply waits decides how much elapsed time you will actually remove, and it is never demonstrated because demonstrations do not wait for anything.
Questions people ask about best purchase order software
What if the vendor will not offer a trial?
Ask them to demonstrate exactly those sequences on a call. Reluctance to show partial receipting is itself an answer to the question you were asking.
Does every organisation need line-level tracking?
If deliveries almost always arrive complete on one date, order level is workable. Most organisations find partial deliveries commoner than they assumed once they look.
What about over-delivery?
Test it too: receive twelve against an order for ten and see whether it is flagged. Silent acceptance is how you end up paying for more than you ordered.