P2P bottlenecks cluster at handoffs, and four of them account for most of the stalled work: approvals, receiving, the exception queue, and supplier back-and-forth. Each has a different owner, which is why "speed up AP" projects keep missing.
Find the queue, name its owner, and fix the step rather than pressuring the people staffing it.
Approvals are the first place work stops
Requisitions and invoices wait on people with day jobs. Thresholds set too low drag executives into trivial purchases; unclear delegation strands documents when an approver is out. These are policy bugs, covered under coding and approval exceptions, and they are fixed by policy, never by reminder emails.
Receiving delays punish accounts payable
Unposted receipts stop correct invoices, and the people who feel the delay sit in a different department from the people who can fix it. The dock is busy, the receipt waits, and the quantity mismatch queue grows. Receiving speed belongs on the P2P scorecard, next to invoice cycle time.
The exception queue is the deepest bottleneck
Every other bottleneck delays one document at a time. The exception queue accumulates, because inflow scales with volume and outflow is limited by how many people investigate. It is the only P2P bottleneck that gets structurally worse as the company grows, which makes it the one worth solving structurally.
Supplier back-and-forth burns the calendar
Half of exception resolution is an email to a supplier and a wait. A two-minute question costs two days of latency, multiplied across every open exception. Cutting the number of questions, by resolving from records instead of correspondence, cuts more cycle time than answering faster ever will.
Stop staffing the queue, and resolve it instead
The standing answer to every bottleneck has been more people in the queue, and the software market has been happy to agree, since seat-licensed tools earn more when more people work the backlog. The better standard: exceptions resolve autonomously and end to end, approvals and receipts get policy fixes, and people handle the judgment calls that remain. The manual vs automated resolution page makes the comparison directly.
Fragment clears the deepest bottleneck by resolving the exception queue autonomously on top of your existing systems. See how it works or request a demo.