The question Australian finance teams rarely ask before implementing AP automation is whether faster processing and stronger governance can coexist in the same workflow, or whether one always comes at the cost of the other. Most implementations resolve this tension by defaulting to speed. What gets sacrificed is the control layer. What gets exposed, eventually, is the business.
Why is the speed-versus-governance tension misdiagnosed?
The standard framing of the speed-versus-governance problem treats it as a trade-off: the business can process invoices quickly, or it can process them carefully, but not both at the same time. This framing produces implementations that try to find the right balance point on a spectrum, and end up sacrificing governance incrementally to keep processing times down.
The actual tension is different. It’s between automation that reduces the number of decisions humans make, and automation that redirects human attention to the decisions that matter. The first version is faster and creates risk. The second version is also fast, but concentrates control where it belongs.
When controls are added manually to an automated workflow - “we’ll check the bank details before we approve” - the finance team is running automation and doing manual checks alongside it. They have the cost of automation without the full benefit. The checks get skipped when volume is high or the approver is distracted. Built-in internal controls are different: the system runs supplier verification, duplicate invoice detection, and ABN checking on every invoice before it reaches the approval queue. The AP officer reviews exceptions. The approver sees a validated invoice. Neither step is slower because the human time is concentrated correctly.
Where do speed-first configurations create risk?
The ACCC’s National Anti-Scam Centre reported AU$152.6 million in payment redirection losses across Australian businesses in 2024. A significant proportion of those cases involved businesses with some form of automated invoice processing configured to prioritise throughput. The automation moved fraudulent invoices toward payment faster than a slower manual process would have. The fraud check wasn’t built into the workflow.
This is the structural risk of speed-first implementation. The business has automated the routing. It hasn’t automated the verification. An invoice that arrives with a changed bank account number is processed at the same speed as a legitimate invoice - because the check that would catch the difference isn’t part of the process. This is a core AP fraud vulnerability that speed-first configurations leave open.
The same logic applies to duplicates. A supplier resubmits an invoice from three months ago with a slightly different reference number. In a manual process, the AP officer might notice. In a speed-first automated process, the invoice clears the approval queue, is approved, and is paid - because duplicate detection runs at the ledger level after the fact, not at intake before approval.
What the tension looks like when resolved
The resolution isn’t finding a compromise between speed and control. It’s designing the automation so controls run automatically, and speed applies to the invoices that have already passed them.
In a well-designed workflow, every invoice - regardless of volume - goes through bank account change comparison, duplicate detection, ABN verification, and PO matching (where applicable) before any human sees it. This takes seconds. Invoices that pass all checks enter the approval queue as validated invoices. Approvers confirm that the expenditure is authorised. They don’t perform verification; that’s already done.
Invoices that fail a check go to an exception queue, with a specific flag describing exactly what triggered the hold. The exception queue is where the finance team’s attention concentrates - and it’s a fraction of total invoice volume. At 80 invoices per month with a 5% exception rate, four invoices per month need active review. The other 76 move through validated.
That’s the design where processing speed and governance aren’t in tension. Routine invoices are fast because the automation has already done the verification. Exceptions get appropriate attention because the system has isolated them from the routine flow.
The approval decision versus the verification decision
The clearest way to understand the resolution is to separate two functions that manual AP conflates: approval authority and invoice verification. Approval authority is a judgment call - is this expenditure authorised given who’s requesting it, what it’s for, and what the business has agreed to spend? Invoice verification is a structured check - does the bank account match, is this a duplicate, does the ABN correspond to the entity on the invoice?
In email-based manual AP, the approver does both. They scan the invoice, recall whether it looks right based on memory, and approve. Under volume, the verification step gets compressed.
In a correctly automated workflow, verification is a system function. Approval authority is a human function. The system handles the repeatable, structured checks on every invoice without variation. The human handles the judgment call on invoices that have already passed those checks. Neither function is slower; they’re just in their correct place.
For businesses evaluating whether their current configuration has resolved this tension or deferred it, the AP controls stack overview covers the five layers - capture, coding, validation, approval, and ledger - and what happens when any of them is absent.
Sources: ACCC National Anti-Scam Centre - Targeting Scams Report 2024 · ATO - E-invoicing and invoice processing in Australia · ATO - Record-keeping requirements for business
Further reading: How to Build an Audit-Ready Approval Matrix · Why Delegation of Authority Matters More Than Automation Speed · Best AP Automation Software Australia 2026