Accounts Payable Metrics: 15 AP KPIs Every Team Should Track

Puts a dollar value on your AP process. APQC uses total AP cost per invoice as a core benchmark measure.

Example: AP spends $100,000 in a quarter on 20,000 invoices. $100,000 ÷ 20,000 = $5.00 per invoice.

What to include: fully loaded staff cost (salary plus benefits), software, scanning or OCR, printing, postage, payment fees, storage, and outsourced services. Leaving out big costs makes AP look cheaper than it is.

Watch out: one blended number hides a lot. PO invoices, non-PO invoices, and exceptions usually cost very different amounts, so split it by type. To estimate your current manual cost, try the AP cost calculator.

How much volume your team handles relative to its size. APQC uses it as a core AP productivity measure.

Example: 20,000 invoices a year handled by 10 FTEs = 2,000 per FTE, about 167 a month.

Watch out: count FTEs honestly, since two people spending half their week on AP equal one FTE. Don't use it to rank individuals. Invoice complexity, exceptions, entities, and currencies all change what "productive" means.

Calendar days an invoice takes between two defined points in AP. It's sometimes called invoice processing time, and it's the metric people most often calculate inconsistently.

Example: 1,000 invoices take a combined 8,000 days from receipt to approval. 8,000 ÷ 1,000 = 8 days.

Pick your end point and write it down. Receipt to approval shows process speed. Receipt to payment also includes your payment terms: if you pay on day 30 on purpose, that number will sit near 30 even when approval took a day. APQC benchmarks receipt to payment transmission. Track both, labeled clearly.

Watch out: cycle time isn't touch time. An invoice can need 6 minutes of work and still wait 9 days. Touch time drives cost; cycle time drives late payments and missed discounts. See where the time goes in invoice processing in accounts payable.

The wait between sending an invoice for approval and getting final sign-off. It's often the biggest part of cycle time, and it mostly happens outside the AP team.

Example: routed at 10:00 a.m. Monday, approved at 2:00 p.m. Tuesday = 28 hours.

Report the median, not just the average. A few invoices stuck for weeks drag the average up and hide what normally happens:

Common causes: approvers who are out, invoices sent to the wrong person, unclear limits, email routing, and no reminders. Break approval time down by approver and level, and consider rule-based invoice approval software.

The share of invoices that can't move through the normal flow and need someone to review or fix them.

Example: 500 exceptions out of 10,000 invoices = 5%.

Make it useful with reason codes: PO mismatch, missing PO, price variance, quantity variance, unknown vendor, missing approval, tax error, suspected duplicate. If 60% are "missing PO," the fix is in purchasing, not AP.

How often invoice data is captured correctly the first time, whether by a person or by OCR.

Example: 9,800 of 10,000 invoices needed no corrections = 98%.

Watch out: invoice-level and field-level accuracy differ. One wrong field on a 20-field invoice is still 95% field-accurate, but someone has to fix it. Invoice-level is stricter and closer to real rework.

For PO-backed invoices, the share that match on the first try without manual correction.

Example: 8,500 of 10,000 PO invoices match first time = 85%.

Watch out: only count PO invoices in the denominator. A low rate usually points upstream: outdated PO prices, unrecorded partial receipts, or vendor data that doesn't match the invoice.

How often the same invoice gets paid more than once.

Example: 6 duplicates in 12,000 payments = 0.05%. Small as a percent, so report it in dollars too.

Track two numbers: duplicates caught before payment (your controls working) and duplicates paid (losses already made). Regular accounts payable reconciliation catches the ones that slip past capture.

The share of payments that go out right the first time: correct vendor, amount, bank account, and date, with no reversal or correction. APQC tracks first-time error-free disbursements as a benchmark measure.

Example: 24 of 12,000 payments needed correction. (12,000 - 24) ÷ 12,000 = 99.8%.

Why it matters: it's the control check on every speed metric. A team that gets faster while this falls isn't improving. Bank-detail errors deserve extra attention, since they can signal fraud, not just a typo.

Also called straight-through processing rate: the share of invoices that move through your defined AP workflow without manual AP work.

Example: 7,000 of 10,000 invoices flowed through untouched = 70%.

Define "touch" in writing. Most teams don't count a required business approval as a touch, since it's a control, not rework. An AP clerk fixing a flagged field does count. Without a written rule, the number can be inflated by changing what counts.

Watch out: read it next to data accuracy and error-free payments. Touchless only helps if the data flowing through is right. More on how it works: touchless invoice processing.

How often invoices are paid by their due date.

Example: 9,500 of 10,000 paid on time = 95%.

Late isn't the only problem. Paying a Net 45 invoice on day 10 with no discount gives up cash for nothing. The goal is paying on time, not as early as possible. Consistent timing is also one of the easiest ways to build stronger vendor relationships.

How much of the available early-payment discount value you actually take.

Example: $40,000 captured of $50,000 offered = 80%.

Terms like "2/10 net 30" offer 2% for paying 20 days early. Annualized, that's a high return, roughly 37% using the common simple-interest method, though the exact figure depends on how it's calculated. Few uses of short-term cash beat it.

Watch out: measure in dollars and record why each discount was missed (late approval, invoice arrived late, cash decision). Taking a discount can be the wrong call when cash is tight, so it's a finance decision, not just an AP one.

The average number of days a company takes to pay suppliers. It sits alongside days sales outstanding (DSO) in the cash conversion cycle, explained in accounts payable vs accounts receivable.

Example: average AP $2 million, annual COGS $12 million. ($2M ÷ $12M) × 365 = 60.8 days. Some companies use purchases instead of COGS; state which you use and stay consistent.

Watch out: higher DPO isn't automatically better. It keeps cash longer, but DPO well beyond your terms often means late fees and unhappy suppliers. Read it with on-time rate and discount capture.

How many times a company uses and settles its average AP balance in a period.

Example: $12 million in credit purchases and $2 million average AP = 6 times a year. 365 ÷ 6 ≈ 61 days, matching the DPO example when both use the same base.

Tip: DPO and turnover are two views of the same behavior. For leadership, pick the one your CFO already uses.

How often suppliers contact AP about invoice or payment status. Many dashboards skip it, but it captures work that never shows up in invoice counts.

Example: 300 inquiries against 10,000 invoices = 3 per 100 invoices.

Why it matters: every "where's my payment?" email is time not spent processing. A rising rate is often the first sign of slow approvals or poor visibility, before on-time rate drops. Tag inquiries by reason.