There is a version of accounts payable management where the team processes invoices, posts them to Xero or MYOB, and moves on. Nobody tracks how long each invoice takes. Nobody knows the error rate. The only metric anyone looks at is whether the bills got paid.
This works until it doesn’t. And the moment it stops working — a duplicate payment slips through, month-end takes an extra two days, a supplier complains about a late payment — there is no data to diagnose what went wrong or where the process is breaking down.
If you are processing more than a hundred invoices a month, you need to measure your AP function the same way you measure any other operational process. Not with vanity metrics. With numbers that tell you whether your process is efficient, accurate and under control.
The metrics that matter
1. Cost per invoice
This is the total cost of processing a single invoice from receipt to posting in your accounting system. It includes the time spent by every person who touches the invoice — the person who downloads it from email, the person who enters the data, the person who codes it, the person who approves it and the person who reconciles it at month-end.
The commonly cited Australian benchmark for manual processing is around $27 to $30 per invoice. That number comes from industry surveys and it holds up in practice. If you have two finance staff spending a combined 60 hours a month on AP and you process 300 invoices, the maths is straightforward.
Businesses using structured AP automation typically bring this below $10. The exact figure depends on invoice complexity — a single-line office supply invoice costs less to process than a multi-line freight bill with mixed GST treatments and a three-way match requirement.
Why it matters: Cost per invoice is the baseline efficiency metric. If you cannot calculate it, you cannot make a credible business case for changing anything about your AP process. It also exposes hidden costs — the senior accountant spending two hours a day reviewing invoices is an expensive line-item coder.
2. Invoice cycle time
This is the elapsed time between an invoice arriving and that invoice being approved and posted to your accounting system. Measure it in business days.
For most manual AP teams processing over 200 invoices a month, average cycle time sits between five and eight business days. The bottleneck is rarely data entry — it is approval routing. Invoices sit in email inboxes waiting for someone to respond. The approver is on site. The invoice got forwarded to the wrong person. Nobody realised it needed a second approval because it was over the threshold.
A well-structured AP process should bring average cycle time below three business days for routine invoices. Complex invoices — new suppliers, PO mismatches, unusual amounts — will take longer, and that is fine. The goal is not to rush every invoice through. It is to stop routine invoices from sitting idle.
Why it matters: Long cycle times mean late payments, missed early payment discounts and strained supplier relationships. They also mean your month-end close takes longer than it should because invoices from the last week of the month are still in the approval queue.
3. Straight-through processing rate
This measures the percentage of invoices that move from receipt to approval to posting without any manual intervention. The invoice is captured, the data is extracted, the account codes are applied, the GST treatment is validated, the PO is matched and the invoice is routed to the right approver — all without a human touching it.
A mature AP automation setup targeting industrial businesses should aim for 60 to 80 percent straight-through processing on recurring supplier invoices. New suppliers, unusual line items and invoices with exceptions will always need human review. That is the point — automation handles the routine work so the finance team focuses on the invoices that genuinely need attention.
Why it matters: This is the clearest measure of whether your automation is actually working. If your straight-through rate is below 40 percent after three months of using an AP tool, either the tool is not learning from your coding patterns or it was not built for your invoice complexity.
4. First-pass coding accuracy
When your system — or your team member — codes an invoice for the first time, how often is that coding correct? Correct means the right account code, the right tracking category, the right tax treatment and the right entity.
In a manual process, first-pass accuracy is difficult to measure because errors are often caught and corrected silently during month-end review. The correction happens, but nobody records that the original coding was wrong.
In an automated process, you can measure this directly: how often does the AI-suggested coding match what the reviewer approves without changes?
Why it matters: Low first-pass accuracy means your team is spending time reviewing and correcting invoices that should have been coded correctly the first time. It also means your month-end adjustments are higher than they need to be. If you are seeing first-pass accuracy below 85 percent on recurring suppliers, something is wrong with your coding logic or your chart of accounts structure.
5. Exception rate
What percentage of invoices trigger an exception — a PO mismatch, a duplicate detection flag, a GST discrepancy, a bank detail change, an amount that falls outside normal range for that supplier?
A healthy exception rate for an established AP process is between 10 and 20 percent. Below 10 percent and you may not be catching enough. Above 30 percent and your exception rules are too aggressive or your upstream processes — purchase orders, supplier onboarding — need attention.
Why it matters: Exception rate tells you two things. First, whether your controls are working — if you never flag exceptions, you are either processing perfectly clean invoices or you are not checking. Second, whether your exception rules are calibrated — too many false positives and your team stops taking exceptions seriously.
6. Duplicate invoice rate
How often does a duplicate invoice make it through to payment? This includes exact duplicates (same invoice number, same supplier, same amount) and near-duplicates (different invoice number but same amount, same date, same supplier — often a re-sent invoice).
Any duplicate rate above zero percent is worth investigating. In practice, businesses processing over 200 invoices a month manually will typically see one to three duplicate payments per quarter. At an average invoice value of $2,000 to $5,000, that adds up.
Why it matters: Duplicate payments are pure cash leakage. They are also a governance failure — if a duplicate can get through your process, so can a fraudulent invoice. Measuring this KPI is a baseline control health check.
7. Early payment discount capture rate
If your suppliers offer early payment terms — 2/10 net 30 is common — what percentage of those discounts are you actually capturing?
Most manual AP teams capture fewer than 20 percent of available early payment discounts because the invoice does not clear the approval process fast enough. The 10-day window expires while the invoice is sitting in someone’s inbox.
Why it matters: A business processing 500 invoices a month with an average value of $3,000 and 10 percent of suppliers offering 2/10 terms is leaving $36,000 a year on the table by missing those discounts. That is real money — and it is directly tied to invoice cycle time.
Metrics that look useful but are not
Invoice volume on its own tells you nothing about efficiency. Processing 500 invoices a month is not an achievement if it takes five people and the error rate is 15 percent.
Approval turnaround time in isolation is misleading. A fast approval might mean the approver is rubber-stamping without reviewing. What matters is whether the right invoices are being reviewed by the right people — approval quality, not approval speed.
Percentage of invoices processed on time depends entirely on how you define “on time.” If the target is seven business days and your average is six, you look fine — but six days is still too slow if your suppliers offer early payment discounts with a 10-day window.
How to start tracking
If you are not measuring any of these today, do not try to instrument everything at once. Start with two metrics:
-
Cost per invoice — estimate your total AP processing hours per month, multiply by the average hourly cost of the people involved, divide by invoice volume. This gives you a baseline.
-
Invoice cycle time — pick 20 invoices at random from last month. For each one, note when it arrived and when it was posted to Xero or MYOB. Calculate the average. This is your current reality.
These two numbers will tell you whether your AP process is where it should be or whether there is a case for changing how you work.
From there, the remaining metrics — straight-through rate, coding accuracy, exception rate — become measurable once you have a system that tracks what happens to each invoice at each stage. That is one of the structural benefits of moving from email-and-spreadsheet AP to a purpose-built AP layer: the data exists because the system recorded every step.
What good looks like
For an Australian SMB processing 200 to 1,000 invoices a month with a mix of simple and complex supplier bills:
| KPI | Manual process | Automated process |
|---|---|---|
| Cost per invoice | $25–$35 | $5–$12 |
| Average cycle time | 5–8 days | 1–3 days |
| Straight-through rate | 0% | 60–80% |
| First-pass coding accuracy | 70–80% | 90–97% |
| Exception rate | Unknown | 10–20% |
| Duplicate payment rate | 1–3 per quarter | Near zero |
| Discount capture rate | Below 20% | 60–90% |
These are not aspirational numbers. They are what well-implemented AP automation delivers in practice for businesses with industrial invoice complexity — freight bills, multi-line supplier invoices, mixed GST treatments and cross-entity structures.
The connection between measurement and control
The reason most AP teams do not track these metrics is not that they do not care. It is that their process does not generate the data. When invoices move through email chains and spreadsheets, there is no system recording when each step happened, who touched the invoice, what was changed and why.
Measuring AP performance requires a process that captures data at every stage. That means moving invoice processing into a system that tracks receipt, extraction, coding, matching, approval and posting — and makes that data available for reporting.
Once you can measure, you can manage. And once you can manage, you can make informed decisions about where to invest time, where to tighten controls and where your AP function is costing more than it should.
Further reading: Best AP Automation Software Australia 2026 · What a Modern AP System Needs to Do · The Real Cost of Manual AP