Accounts payable processes thousands of invoices, and even a few percent of errors means hundreds of exceptions to find, correct, and re-process each month. Because the process has clear steps and countable defects, it lends itself well to the tools of quality engineering.
This guide covers the common error types and their causes, the measures to use, and a worked example that turns a month of 372 logged errors into a Pareto chart, a DPMO and sigma level, and a cost saving from focusing on the top two causes.
Before You Start
Why Accuracy in Accounts Payable Matters
Errors Cost Money Twice
An error costs the payment itself if it is wrong, and the time to find, correct, and re-process it.
Rework Hides in Volume
In a high-volume process, a 6% error rate means hundreds of exceptions a month, each handled by a person.
Suppliers Notice
Late or wrong payments damage supplier relationships and can cost early-payment discounts.
Controls Depend on Clean Inputs
Matching, approvals, and fraud checks work better when the data entering them is right the first time.
Common Error Types and Their Causes
| Error type | Typical causes | Typical countermeasure |
|---|---|---|
| Coding errors (wrong account or cost center) | Free-text entry, unclear coding rules, requesters unfamiliar with the chart of accounts | Default coding by supplier, drop-down lists, coding at the requisition |
| Missing or invalid purchase order | Purchases made without a PO, PO closed or exhausted | Require a PO before commitment, alerts for PO balance |
| Duplicate or near-duplicate invoices | Same invoice received by email and mail, small differences in reference numbers | Duplicate checks on supplier, amount, and date; standard invoice numbering |
| Wrong amount or tax | Price not as ordered, tax not applied correctly, unit-of-measure mismatch | Three-way match (PO, receipt, invoice) with tolerances |
Measures
| Measure | Formula | Meaning |
|---|---|---|
| First-pass yield | Invoices processed without correction / invoices received | Share right the first time. |
| Error rate | Invoices with an error / invoices received | The complement of first-pass yield. |
| DPMO | Defects / (units × opportunities per unit) × 1,000,000 | Defects per million opportunities. See the DPMO guide. |
| Touchless (straight-through) rate | Invoices processed with no human intervention / invoices received | Automation effectiveness. |
| Cost per exception | Handling time × labor rate (plus any overpayment) | What an error really costs. |
Worked Example: A Month of Invoices
An AP team processes 4,800 invoices in a month and logs every error found before or after payment, classified by type. It checks five points on each invoice: supplier, amount, coding, PO match, and approval. The figures are illustrative.
| Measure | Calculation | Result |
|---|---|---|
| Errors logged | 132 + 96 + 60 + 54 + 30 | 372 |
| Opportunities | 4,800 invoices × 5 checks | 24,000 |
| DPMO | 372 / 24,000 × 1,000,000 | 15,500 |
| Approximate sigma level | Inverse normal of (1 − 0.0155) plus the conventional 1.5 shift | about 3.66 |
| Cost of handling errors | 372 × $18 per exception | $6,696 per month |
The team focuses on the top two types. Default coding by supplier and coding drop-downs address coding errors, and a rule that a PO must exist before a purchase is committed addresses missing POs. If those countermeasures cut both types by 60%, errors fall by (132 + 96) × 0.6 = 136.8, to about 235, and handling cost falls by 136.8 × $18 = about $2,460 per month. That is a target, not a result. The team confirms it by tracking the error rate by type on a control chart. See the Transaction Quality guide.
Handling cost is not the whole picture. A duplicate payment, even when recovered, ties up cash and takes vendor follow-up, and unrecovered ones are a direct loss. Weigh these separately from processing time. Changes to approval or matching controls should be reviewed by finance leadership and, where relevant, the auditors.
Preventing Errors at the Source: The Procure-to-Pay Chain
Most invoice errors are created upstream of accounts payable. An invoice for the wrong amount may reflect a purchase order with the wrong price; a duplicate payment may reflect a vendor record duplicated in the master file. Looking at the whole chain shows where to place checks.
Vendor master. Duplicate vendors cause duplicate payments. Control who can create or change vendor records, verify bank detail changes by a call to a known number, and periodically review the file for duplicates and inactive vendors.
Purchase order. Wrong prices, missing terms, and unclear descriptions cause exceptions later. Keep catalog prices current and require PO approval before commitment.
Receipt. Late or inaccurate receipts block matching. Record receipt at delivery, and make it easy for receiving staff to do so.
Invoice capture. Electronic invoices and clear submission instructions reduce keying errors. Reject or flag incomplete invoices at intake, and feed back to vendors.
Match and approve. Set tolerances for small price and quantity differences and route exceptions by cause, so that each type goes to the right person.
Duplicate Payments, Controls, and Measuring Accuracy
Some errors are costly enough to justify specific controls, and the process should be measured in a way that shows whether accuracy is improving.
- Duplicate detection. Check for the same vendor with the same invoice number, or the same amount and date, before payment. Review payments after the fact to catch those missed.
- Bank detail changes. Require independent verification and approval. Changes requested by email are a common fraud route.
- Segregation of duties. Different people should create vendors, approve invoices, and release payments, or compensating controls should be in place.
- Payment run review. Review the list for unusual amounts, new vendors, and round-number payments before release.
| Measure | Definition | Use |
|---|---|---|
| First-pass match rate | Invoices matched automatically with no touch | Shows process health and rework |
| Exception rate by cause | Exceptions divided by invoices, by type | Points to the causes to fix upstream |
| Cost per invoice | Processing cost divided by invoices | Shows efficiency; watch that accuracy does not suffer |
| Payment errors and recoveries | Overpayments, duplicates, and amounts recovered | Shows the cost of inaccuracy |
| On-time payment rate | Invoices paid by due date | Vendor relations and discounts |
Use Pareto and p-charts. Rank exception causes to choose the top one or two to fix, and chart the exception rate over time to see whether a change worked, as in the Transaction Quality and p-Charts Guide.
Balance speed and accuracy. Paying faster to capture discounts is good only if accuracy holds. Track both. See the Sigma Level, DPMO, and RTY Guide and the Pareto Analysis Guide. This guide is educational and is not accounting, audit, tax, or legal advice. Follow your accounting policies, the standards that apply to you, and the advice of qualified professionals. Figures in the examples are illustrative.
Self-Assessment Questions
- Do we log errors by type, and do we know the top two?
- Do we know our first-pass yield and cost per exception?
- Do we fix the source, such as coding rules and PO requirements, or only correct individual invoices?
- Do we check for duplicates before payment?
- Do we track error rates over time on a control chart?
Common Mistakes
Fixing the Invoice, Not the Cause
Correcting each error individually leaves the cause in place. Use Pareto analysis to find the causes.
Adding Approval Layers
More approvals add delay and rarely fix coding or PO problems. Prevent errors at entry.
Counting Only Handling Time
Overpayments, lost discounts, and supplier friction are real costs. Include them.
Weakening Controls to Go Faster
Automation should reinforce matching and duplicate checks, not bypass them.
Invoice and Payment Accuracy: Frequently Asked Questions
How do you calculate DPMO for invoices?
DPMO is defects divided by the number of units times the opportunities for a defect per unit, multiplied by 1,000,000. For example, 372 errors over 4,800 invoices with 5 checks each gives 372 divided by 24,000, times a million, or 15,500. Define the opportunities clearly and count them the same way each period.
What are the most common causes of invoice errors?
Common causes include incorrect coding, missing or invalid purchase orders, duplicate or near-duplicate invoices, and amount or tax mismatches. Most trace to entry practices and rules upstream, so fixing the source, such as default coding and requiring a PO before purchase, works better than correcting invoices one by one.
Does automating accounts payable remove errors?
It can remove many, especially rekeying and matching errors, but automation also changes where errors arise, for example in data capture or master data. Automate on top of clear rules and clean master data, keep matching and duplicate checks, and measure the touchless rate together with the error rate.
Sources and Further Reading
- Michael George, Lean Six Sigma for Service.
- Forrest W. Breyfogle III, Implementing Six Sigma, on DPMO and sigma levels.
- ASQ Certified Six Sigma Black Belt Body of Knowledge.
- Guidance on accounts payable controls published by professional accounting bodies.