Most accounts payable problems are not independent problems. High cost per invoice, a slow month-end close, missed early payment discounts, duplicate payments, exceptions eating senior time: these are symptoms of where an operation sits on a maturity curve.
Here are the five stages, the evidence that tells you which one you are in, and why the next stage is the only one worth planning for.
The five stages
Stage 1: Manual
Invoices arrive by email and paper. Data is keyed by hand, tracked in a spreadsheet, approved by forwarding messages.
Evidence you are here: cost per invoice at or above the industry average, no reliable period cutoff at month end, exceptions handled by whoever has capacity, and no measured exception rate.
What unlocks the next stage: getting documents into one intake path instead of several.
Stage 2: Digitized
Documents are electronic and stored centrally. Search improves. Keying and decisions remain manual.
Evidence you are here: PDFs in a shared drive or portal, searchable history, but someone still types data into the ERP and chases approvals by email.
What unlocks the next stage: extraction and validation at the point of entry, rather than after.
This stage deserves a warning. It feels like progress and frequently delivers little, because a faster filing cabinet is still a filing cabinet. Teams can sit here for years believing they are automated.
Stage 3: Captured and routed
Invoices are captured on arrival, data is extracted and validated at entry, approvals route by rule with delegates and escalation.
Evidence you are here: invoices no longer wait in inboxes, duplicates are flagged automatically, approvals do not stall on one person's absence.
What unlocks the next stage: vendor master data quality and an ERP integration that posts, rather than exports.
This is where the measurable gains begin, because the waiting and hunting that consume cycle time disappear.
Stage 4: Integrated
Validated documents post into the ERP with vendor matching and coding applied. The team works in one system instead of reconciling two.
Evidence you are here: no manual export or import step, ERP rejections surface with an owner, straight-through rate measured and reported.
What unlocks the next stage: reducing the exception rate, which is the ceiling on straight-through processing.
Many operations believe they are here because a tool is installed. The test is whether anyone moves data between systems by hand — the heart of the ERP integration test.
Stage 5: Touchless at scale
The majority of invoices flow from receipt to posted with no human touch. People are not removed; they handle exceptions, supplier strategy, controls and analysis.
Evidence you are here: straight-through rate is the primary operational metric, exception rate approaches best-in-class levels, and AP time is spent on judgment rather than data handling.
Why stages cannot be skipped
The common mistake is buying Stage 5 and installing it on a Stage 1 process. The failure is mechanical, not philosophical.
Touchless processing requires clean inputs: maintained vendor master data, purchasing discipline that raises POs, approval rules built on roles rather than individuals. Without those, automation routes the same chaos faster, the exception queue becomes the process, and the organization concludes that automation did not work.
Each stage creates the preconditions for the next. Capture at entry makes validation possible. Validation makes straight-through posting possible. Straight-through posting makes touchless scale possible. Skipping a stage relocates manual work rather than removing it.
Three questions that locate your stage
- What share of invoices posts without a human touch? If the number is not measured, you are at Stage 3 or below regardless of the software installed.
- Where does data enter the process? If a person keys it, you are below Stage 3. If a person moves it between two systems, you are below Stage 4.
- What is your exception rate, and who handles exceptions? Senior staff on routine cleanup is the clearest signal that the stage below was never completed.
Pick one stage, not three
Name the single blocker preventing the next stage — PO policy, vendor master data, approval rules, or integration — and fix that one thing. Teams that advance one stage deliberately outperform teams that buy five stages at once.
The sequence that works is the same at every stage: measure, fix the inputs, then automate.
If you are choosing what to evaluate next, the honest checklist of questions to ask any AP automation vendor is a good place to start.