Procure to pay problems are strikingly similar across organisations of very different sizes and industries, which is useful because it means the diagnosis is usually quick. Five challenges account for most of the difficulty, they interact, and four of the five are addressed by decisions rather than by purchases.
Buying without orders, and receipts nobody records
The two upstream failures, and the two that cap everything downstream. An invoice with no order cannot be matched; an order with no receipt cannot complete a three-way match. Both are behaviours owned outside payables, both are free to fix in principle, and both require somebody senior enough to arbitrate between departments.
Slow approvals and unowned exceptions
Requests and invoices waiting on people who are busy, away or not actually the right approver. Exceptions sitting in a shared queue that everybody can see and nobody owns. Both are configuration and ownership problems rather than capability problems, and both respond quickly to named owners, automatic cover and visible ageing.
Supplier data, which underlies all of it
Duplicate supplier records, wrong legal entities, stale bank details and inconsistent naming. This is the least discussed challenge and it makes the other four worse: matching fails, duplicate detection breaks, statements will not reconcile and fraud has more places to hide. Cleaning it is unglamorous and repays repeatedly.
Questions people ask about procure to pay challenges
Which should we tackle first?
Whichever your exception reason mix says is largest, and usually that is missing receipts, which is also the cheapest to fix.
Is software the answer to any of these?
To approvals and exception visibility, largely yes. To the upstream two, no: software cannot match documents that were never created.
How long does fixing the upstream take?
A quarter or two, because it is habit change involving people outside finance rather than configuration.