99.9% Accurate, and Still Losing Millions?

You’ll often hear that Accounts Payable data processing is 99.9% accurate. It sounds reassuring. So is it true? The honest answer is yes, no, and it depends. I’ve been reviewing AP organizations’ data for more than 25 years, and here’s what I’ve learned: the accuracy number matters far less than what’s driving it.

First, do the math on 99.9%

Before we debate whether 99.9% is accurate, let’s sit with what it actually means. If your organization processes $1 billion in annual spend at 99.9% accuracy, that “tiny” 0.1% is still $1 million in errors. At $500 million in spend, it’s half a million dollars. Suddenly 99.9% doesn’t feel so comfortable.

That’s the first problem with the accuracy claim: even when it’s true, the leftover is real money. And in our experience, plenty of organizations aren’t actually hitting 99.9% at all. We’ve recovered funds from companies performing above that threshold and from companies performing well below it.

ERP was supposed to fix this. It didn’t.

Back in the early 2000s, when companies were rolling out ERP systems, the expectation was that error rates would drop dramatically. With automation, built-in controls, and standardized workflows in place, surely the mistakes would disappear.

They didn’t. And through the mid-2010s to 2020, in the data we reviewed, we actually saw error rates tick up, not down. More systems, more integrations, more complexity, and more places for errors to hide. The technology helped in some ways, but it didn’t deliver the clean, error-free AP department everyone expected.

So if the accuracy rate isn’t really about the ERP badge on the wall, what is it about? Over 25 years, three factors stand out.

1. How many ERP systems you run

This is the big one. Companies running a single, integrated ERP generally have fewer errors. Companies running more than one ERP almost always have more.

The reason is simple: the systems aren’t talking to each other. A single ERP environment will flag a suspicious or duplicate transaction because it can see everything in one place. Split that same activity across two or three systems that don’t integrate, and the controls that would have caught the problem never fire. The invoice paid in System A and again in System B looks perfectly normal to each one.

2. The strength of your control environment

The second factor is discipline. How well is the organization actually managing its controls day to day? A few questions tell us a lot:

• Vendor master file health. Is it clean, or is it full of duplicate and outdated vendor records? Duplicate vendors are one of the most reliable sources of duplicate payments.
• PO usage. Does the organization consistently use purchase orders, or do a lot of invoices come through without them? Strong, consistent PO discipline creates a matching trail that catches errors early.

The stronger and more consistent the controls, the closer an organization gets to that 99.9%, and beyond.

3. How much processing is automated

The third factor is how invoices actually get into the system. Companies that rely on manual entry see higher error rates: it’s simply more places for a keystroke to go wrong. Companies that lean on EDI and electronic processing see error rates that are significantly lower.

Automation isn’t a silver bullet, but it removes a whole category of human-entry mistakes, and that shows up clearly in the numbers.

So, what’s your real number?

Here’s the takeaway: “99.9% accurate” isn’t a fact you can assume, but an outcome that depends on how many systems you run, how strong your controls are, and how much of your processing is automated. Some organizations are better than 99.9%. Many are worse. And even the good ones are leaving real dollars on the table.

The only way to know your real number is to measure it. That’s what we do, and more often than not, what we find surprises people.

Want to know where your AP data really stands? Start a no-cost Proof of Value. We only get paid a percentage of what we recover.