What Do Manual Mistakes Actually Cost?
Manual errors look small one at a time, but together they can cost a lot. Here's how to work out what the errors in your admin actually cost.
The errors nobody counts
Most businesses know that mistakes happen in admin now and then. An invoice with the wrong amount, or an email that ends up with the wrong customer. Each one looks small, which is why hardly anyone puts a number on them. But the cost often doesn't stop at the time it takes to fix the mistake. A large part comes from what the mistake leads to later.
Where errors happen
Anyone entering information by hand will make mistakes sooner or later, however careful they are. The errors usually happen in certain situations:
- The same information has to be moved between several systems
- Input arrives in different formats, such as PDFs and spreadsheets
- It's the end of the month and the pace is high
- Someone gets interrupted in the middle of a task
- The rules have exceptions that aren't written down anywhere
This is about how the work is set up. Asking people to be more careful doesn't change that.
Four kinds of cost
It's easier to see what errors cost if you split the cost into four parts, since each one is underestimated in its own way.
Fixing the error
This is the visible part. Someone spots the mistake and has to find out what happened, correct it in the system where it started, update other systems that were affected and talk to the people involved. If the error has spread to several systems, it can take a long time to sort out.
What the error leads to
An error that isn't caught right away has consequences. A wrong invoice leads to a credit note and a new invoice. A wrong price in a quote leads to an awkward negotiation when the customer notices. Because the consequences come later, they're rarely linked back to the original mistake.
Customer trust
A customer will usually forgive a one-off mistake. If it happens again and again, the customer starts to wonder whether you're on top of things, and that can affect whether they keep buying from you. That cost is hard to measure, but it's there.
Risk
Some errors have consequences outside the company. If VAT is wrong in a return, the Swedish Tax Agency can charge a tax surcharge. If personal data is sent to the wrong recipient, it may be a personal data breach, which under Article 33 of the GDPR generally has to be reported to the supervisory authority within 72 hours. Errors like these are rare, and that's exactly why there's seldom a routine for them when they do happen.
An illustrative calculation
The numbers below are made-up assumptions to show how the calculation works. They aren't based on any measurement.
Imagine a services company that handles 600 admin records a month, such as invoices and time sheets. We assume 2 percent of the records contain an error, that each fix takes 25 minutes and that dealing with the consequences takes three times as long as the fix itself.
- 600 records × 2% = 12 errors a month
- Fixing: 12 × 25 minutes = 5 hours a month
- Consequences: 3 × 5 hours = 15 hours a month
- Customer trust and risk: a cautious estimate of 5 hours a month
That's 25 hours a month. At an hourly cost of €45, it comes to €13,500 a year. That's what the errors cost in the example, and no automation removes all of it. If you want to calculate what an automation saves, you also need to subtract what it costs to run.
Why the cost is hard to see
The errors are spread across several people. Whoever fixes the invoice doesn't know customer support also got a call about it, and the salesperson who loses a customer doesn't know it started with an admin error long before.
Few businesses keep a running list of errors and what they led to, either. Without that, it becomes an estimate from memory. And the person fixing the errors is already on the payroll, so the time doesn't feel like a cost, even though it could have gone to other work.
What happens when the steps are automated?
When information moves between systems without anyone typing it in again, the errors that come from manual entry go away.
Automatic checks can also flag a problem right away when a price is outside the usual range or a total doesn't add up. That stops the error before it spreads.
An AI agent can handle the steps where the input looks different every time. It can read details from a PDF or an email, compare them with earlier data and flag anything unusual to a person instead of guessing. How much it may do on its own is decided by what a mistake in that particular step would cost.
When every step is logged, you can also see exactly where an error started. That makes both the fix and the improvement easier. What's left for people is more often the hard cases, like incorrect input and exceptions that need judgment.
How to calculate your own error cost
You can get a first estimate yourself over a month.
- Ask the team to note every error that gets fixed, with a short description and roughly how long it took.
- Ask them to also note what the error led to, such as a credit note or an extra call, and how long that took.
- Add up the time and multiply by your hourly cost.
- Multiply by 12 to get an annual figure.
The list also shows where it's worth starting. The workflows with the most errors and the biggest consequences are where an automation can do the most good.
Errors come from how the work is set up, and that's what can be changed. Want help working out what your errors cost and where it's worth starting? Get in touch and we'll go through your workflows together.