Accounts payable KPIs: what good looks like

Most finance leaders can tell you their AP team is busy. Far fewer can tell you what a single invoice costs to process, or how many days it sits before it clears. Those numbers exist in every accounts payable function. They are just rarely measured, so they rarely improve.

A handful of accounts payable KPIs turn a vague sense of how AP is doing into something you can benchmark, target and put in front of a board. This is the short list that carries most of the signal, what good looks like against published benchmarks, which one to fix first, and where the data comes from.

The KPIs that matter

You could track dozens of AP metrics. Five carry most of the weight. Each one measures a different stage of the same journey, from the moment an invoice arrives to the moment it posts to your ERP.

• Cost per invoice. The fully loaded cost of processing one invoice from receipt to posting, with labour, software and overhead included. It is the headline efficiency number and the one most teams have never actually calculated.

• Invoice cycle time. The elapsed time from when an invoice arrives to when it is approved and posted, measured in days. It is what a supplier feels when they chase you, and what decides whether you can hit payment terms.

• Touchless rate. The share of invoices that travel from capture to posting with no human keying or intervention, also called the straight-through rate. It is the clearest single measure of how much of your AP is genuinely automated.

• Exception rate. The share of invoices that fail a check, a price mismatch, a missing purchase order, a suspected duplicate, and get pulled out for manual review. Exceptions are where the cost and the delay concentrate.

• Early-payment discount capture. The share of available early-payment discounts your team actually collects before they lapse. A slow AP function leaves this money on the table every month without anyone noticing.

Benchmarks: what good looks like

A KPI on its own is a number without a verdict. The value comes from a benchmark that tells you whether the number is good. The most widely cited AP benchmarks come from Ardent Partners, whose annual research separates the best performers from everyone else.

On invoice cycle time, Ardent puts best-in-class teams at 3.1 days to process an invoice, against 17.4 days for all other organisations in the study. That is a gap of roughly two working weeks per invoice between the teams that have instrumented their process and the teams that have not.

On cost, the pattern holds. Ardent finds that a best-in-class AP function processes an invoice for roughly a fifth of what the average team spends. The saving does not come from paying people less. It comes from touching each invoice far fewer times.

Exception rate sits underneath both of those. Ardent reports best-in-class exception rates of around 9%, against about 22% for the rest. That difference explains most of the gap in cost and cycle time, because an exception is an invoice that stops moving and starts consuming someone hour by hour. It is worth understanding what an invoice costs to process before you set a target, so the number you commit to is grounded in your own volumes.

Read those benchmarks as a direction of travel. Best-in-class figures come from teams that have invested in AP for years, and a mid-market finance function will still have ground to cover from a standing start. The useful question is whether each of your five numbers is moving the right way, quarter on quarter, and how far the gap to the benchmark still is.

Which KPI to fix first

It is tempting to target cycle time, because it is the one suppliers complain about and the one a board understands instinctively. Chasing cycle time directly usually fails, though. Cycle time is a symptom. It is slow because invoices keep stopping, and they keep stopping because they become exceptions.

So the KPI to attack first is the exception rate. Every invoice that clears its checks on the first pass is an invoice that never joins a queue, never waits for a reply, and never lapses an early-payment discount. Bring the exception rate down and cycle time, cost per invoice and discount capture all move in the right direction on their own.

Most exceptions trace back to two causes: data captured wrongly from the invoice, and a purchase order or goods receipt that does not line up. Fix capture accuracy and matching discipline, and the exception rate falls at the source. That is why the touchless rate is the KPI to watch as your leading indicator. It rises as exceptions fall, and it tells you how much of the work your team no longer has to do by hand.

Cost per invoice is a useful headline, but on its own it can mislead. A team can shave the cost by rushing approvals or skipping checks, and pay for it later in duplicate payments and strained supplier relationships. Watching cost alongside the exception rate keeps the picture honest, because a genuinely cheaper invoice is one that needed less handling while still passing every check.

Instrumenting them

None of these KPIs are useful if the data behind them is a guess. Each one has a specific source, and getting the measurement right matters as much as the target.

The table sets out each KPI with its formula, where the numbers come from, and the trap to avoid when you measure it. It adds a sixth, on-time payment rate, which is worth tracking alongside the five because late payments usually trace back to slow approval.

To see the formulas at work, take a team that processed 1,000 invoices last quarter at a total running cost of £14,000. Cost per invoice is £14,000 ÷ 1,000, or £14. If 620 of those invoices posted with no one touching them, the touchless rate is 62%, and the exception rate is the 38% that needed a person. Run the same sums each quarter and the trend tells you whether a change in process actually worked.

• Cost per invoice. Take the fully loaded cost of running AP over a period, salaries and on-costs, the software, a share of overhead, and divide by the number of invoices processed. Measure it the same way each quarter so the trend is honest, even if the absolute figure is rough.

• Cycle time. This comes from system timestamps, which are more reliable than a team memory of how long things take. Your ERP or workflow tool records when an invoice was received and when it posted. The gap between those two events, averaged across the period, is your cycle time. If you cannot see those timestamps, that is the first thing to fix.

• Touchless and exception rates. These come from your capture and matching layer. It knows which invoices posted without a human touching them and which it flagged for review. The accuracy of the underlying extraction sets the ceiling here. Stratas reads invoices at 89% field-level extraction accuracy, measured on real customer documents, and the cleaner that read, the more invoices clear straight through.

• Early-payment discount capture. This lives in your payment data. Compare the discounts available on supplier terms against the discounts actually taken. The gap is money your cycle time is costing you, quantified.

The point of measuring is not the dashboard. It is the decision the dashboard lets you make: where the time is going, which invoices are costing you the most, and what a realistic target looks like for a team of your size and volume.