Business Basics
How to reduce human error in the workplace

How to reduce human error in the workplace

illustrated button and cursor with the words business basicsHeader imageHeader imageHeader imageHeader image

How much can human error cost a company? On August 11, 2020, a single error cost Citigroup USD 900 million when someone in Delaware accidentally paid off a loan on Revlon's behalf instead of making an incremental payment. As human errors go, that one was a doozy.

The company asked for its money back, of course, but the payment had gone out to many different lenders. Some of them returned the funds; others didn't.

In the end, Citigroup had to go to court, lose, then finally win on appeal on September 8, 2022, more than 2 years later, before proving its legal right to recoup that error.

What is the risk of human error?

We're all human. We all make mistakes. But some errors cost more than others.

The FBI reports that the simple mistake of opening and clicking the wrong email costs businesses billions of dollars every year in data breaches and other cybersecurity consequences.

In other words, many of our most costly mistakes don't come from major accidents. They come from very common, very human factors, including routine tasks that don't feel all that risky.

What are some of the causes of human error?

When it comes to human decision-making, there are many types of human error, but don’t be too quick to blame your people for the foibles of complex systems.

Risk management and error reduction don’t just come down to addressing mishaps or poor training. In fact, human error prevention tends to be more successful when it acknowledges the unintended consequences of systems that don’t take our human biological nature into account.


The simple fact is that boredom gives the human brain a false sense of security, lulling us into momentary lapses of judgment. "Routine" feels like the opposite of "risky," no matter how dangerous that routine might be.

Change blindness

Another thing the human brain tends to be blind to is any small change in the details we habitually ignore. If the address on a monthly invoice changed, would you notice that detail? What if the PO box was only different by one number?

Like boredom, change blindness is the kind of human performance error that happens when routine takes over and we stop thinking actively about what we're doing.

Manual, ad hoc systems

There are plenty of ways in which the risk of error pops up in the workplace, but at the end of the day, the worst culprit often comes down to poorly designed systems.

Human error investigations into workplace accidents often reveal systemic flaws. Take, for example, the Citigroup error of almost 900 million dollars. In that case, the software system forced people to "trick" the loan software, checking a box for the whole amount even when they wanted to send a smaller payment.

Three people saw that checked box, and all three approved it because of how the system worked.

Bottom line, what the Citigroup case study shows is simple: if we want to reduce human error, we can't ask people to overcome bad systems. Counting on people to do everything correctly, all the time, just doesn't work.

Seven principles for preventing human error

Instead of making do with manual systems and repetitive tasks, businesses can reduce human error by implementing good systems—ones that streamline dull routines and catch the mistakes we're most likely to miss.

The right initiatives, of course, depend on what you're trying to fix—from equipment failure to financial errors—but these 7 principles can point your systems in the right direction.

1. Track what you care about

Start by defining the errors you want to avoid. Do you want to reduce the number of exceptions in your accounts payable processing? What about lowering the overages in your departmental budgets?

Whatever you want to address, put a risk-assessment system in place so you can track your results. Sometimes, the simple act of focusing on the problem is half the battle.

2. Create formal systems

Many startups begin with ad hoc systems that work well on a small scale but become tough to maintain as a company grows. Standard operating procedures can reduce human error while preparing for expansion.

Managing bills and invoices by email, for example, might be fine when you have a handful of clients and only a few bills every month, but as that number grows, things can easily slip through the cracks—especially when there's an issue.

Without a standard operating procedure for handling exceptions, bills pulled out of the stack often end up right back in it, getting paid anyway, even if they're wrong.

3. Keep your systems simple

Ideally, your formal systems should streamline your operations, reducing dozens of ad hoc systems to a few simple templates and workflows.

After all, if you create a different system for every possible thing that can happen, your solution checklist will be just as complex as the original problem. Instead, try to treat everything that flows through your system the same way.

4. Automate manual tasks and routines

Simplifying your operating systems may reduce active errors, but it also runs the risk of human boredom kicking in. To minimize that risk, automate your new systems as much as possible, marrying human action with the benefits of automation and artificial intelligence (AI).

Let computers and other machines handle the repetition while people oversee those systems with the benefit of human judgment.

5. Practice separation of duties

Because of things like inattentional blindness and change blindness, the human brain has a hard time catching certain kinds of errors. We tend to see what we expect to see.

As a result, it's better for one person not to play several roles in a given process. For example, the person who performs a machine repair should not also sign off on that repair. Someone else should double-check that the problem has been fixed.

The same is true in the example of bills—generally speaking, the person who writes the checks should not be the same person who approves those payments.

6. Create smart workflows, not loopholes

The checks example brings up an interesting problem—if your check writer has to run every single payment by your CFO, no matter how small or routine that bill might be, you're likely to get bogged down in your own system.

That runs the danger of people trying to get around the system, creating loopholes in the very system that's designed to prevent errors.

Instead, create intelligent systems with simple variability—release valves for the routine things that don't require significant oversight.

For example, small, recurring bills could be approved automatically, while you could route anything over a certain amount through a formal approval process.

7. Keep an automatic audit trail

Finally, your system should keep an automatic audit trail of every touchpoint along the way. Capturing every action helps with statistical analysis—such as how many human touches the average invoice requires before getting paid, or how many exceptions your team has to deal with in a given month.

It also makes incident investigation and root-cause analysis a lot easier, helping you take corrective action should the need ever arise.

Reducing human error in your bill pay systems with BILL

If you'd like to minimize human error in your accounts payable, BILL helps you automate your invoice processing from intake to payment.

  • It reads each invoice for you, entering data automatically for your review
  • It applies your approval workflows automatically
  • It lets you communicate within the system, automatically connecting each conversation to its underlying bill
  • You can make payments through your choice of several payment options
  • It even keeps an automatic audit trail of every touchpoint along the way

Learn more about streamlining your AP process with BILL by implementing payment automation.

The information provided on this page does not, and is not intended to constitute legal or financial advice and is for general informational purposes only. The content is provided "as-is"; no representations are made that the content is error free.