Wiki/Source-to-pay/
What is source-to-pay (S2P)? A practical guide

What is source-to-pay (S2P)? A practical guide

Source-to-pay
·
4 min read
·
Updated July 2026
Joshua Kurian
Joshua Kurian
On this page

Source-to-pay (S2P) is the complete process a company runs to buy goods and services: sourcing and selecting suppliers, contracting, requisitioning, creating purchase orders, receiving, matching and approving invoices, and paying. It spans procurement, accounts payable, and finance, from the first sourcing decision to the cleared payment. If you already think in terms of procure-to-pay, S2P is that plus everything that happens before a requisition exists; our P2P vs S2P vs source-to-settle comparison covers the terminology in detail.

The definition is the easy part. What deserves a procurement leader's attention is how the process behaves: S2P is one connected chain, and almost every organization manages it as two disconnected halves, sourcing and procurement on one side, AP on the other. That split is where the money leaks.

S2P covers ten workflow families, and the invoice is only one of them

In a large physical-goods enterprise, S2P work clusters into ten families:

  • RFQ administration and supplier selection
  • Requisition-to-PO conversion and PO creation
  • Change orders and PO amendments
  • Supplier master data upkeep and exceptions
  • Maverick spend and contract leakage
  • Invoice exceptions and 3-way match
  • Duplicate invoice detection
  • GL coding
  • Credits, rebates, and deductions
  • Tax holds and withholding

Half of these run before AP ever sees an invoice. Yet walk through most transformation roadmaps and the investment concentrates on the bottom of that list, because that is where the backlog is visible and where the metrics live. Exceptions show up in every one of these families, and they are connected in a way most maturity models miss.

One upstream error routinely becomes several downstream exceptions

Take a single PO created against an outdated contract price. Every receipt against it produces an invoice that fails 3-way match. AP short-pays or holds; the supplier resubmits; the resubmission trips duplicate detection; someone issues a credit; the credit sits unapplied for a quarter. One stale field in one document just generated five pieces of exception work spread across three teams, and the analyst clearing the fifth piece has no mandate, and often no visibility, to fix the contract price that started it.

The same fan-out follows a supplier record carrying an old remit-to address, a requisition coded to the wrong entity, or a change order that never made it into the ERP. Most of what lands in AP's exception queue was created weeks earlier, upstream, by a sourcing decision, an incomplete PO, or stale supplier data, and it compounded on the way down.

The real maturity signal is how few exceptions you create

Most S2P scorecards measure the back half: cost per invoice, touchless rate, days to resolve an exception. These reward speed of cleanup. A team that clears exceptions in two days looks mature on every dashboard. A team that prevents a third of them from ever existing looks like nothing happened, because prevention has no queue to photograph.

Our position: exception creation rate, traced to the upstream stage that caused it, beats any resolution-speed number as a maturity measure. It also changes where the budget goes. Automation money follows visible pain, which is why it keeps landing on matching and coding while the front half of S2P stays manual and keeps refilling the queue. Our guide to procurement automation breaks down what is actually automatable at each stage.

Root causes live across systems, so the fix has to reason across systems

The chain above ran through an ERP, a supplier portal, a contract repository, and somebody's inbox. That is why the problem has survived twenty years of tooling. RPA breaks on it because rules cannot handle work that varies transaction by transaction and lives in five systems. ERP suite modules only see their own data, and the document that caused the exception is usually somewhere else. BPOs move the cleanup offshore and leave the creation rate untouched.

Agentic automation is the first approach shaped like the problem: agents that read and reason across the ERP, MES, portals, email, and spreadsheets the way an analyst would, with a human approving the calls that matter. (What is autonomous procurement covers the category.) Fragment builds these agents for the full S2P lifecycle, from PO creation, change orders, and supplier master data through matching, deductions, and tax holds, working inside your existing systems with nothing ripped out or replaced and live in production in weeks. See the workflows Fragment covers across S2P, or book a demo built around your own systems and workflows.

From Fragment
See exception resolution on your own data
Fragment resolves invoice exceptions autonomously across your existing ERP and documents.
Request demo