Ask a business owner what re-typing orders costs and the answer is almost always “oh, a few minutes”. That is true. It is only true per order, not per year — and those two numbers are further apart than instinct says. This article shows how to calculate the real cost of manual work, which items everyone forgets, and how to read from the result whether automating is worth it.
The sum nobody does
Take those orders. Six minutes each, ten a week, two people doing it. That is 120 minutes a week — two hours, not bad. But there are 46 working weeks in a year (holidays and public holidays removed), so 92 hours. At a fully loaded hourly cost of 45 euros that is over 4,100 euros a year. For a task that costs “a few minutes”.
And that is before the item that usually tips the sum.
Rework: the hidden half
With re-typing, a small share goes wrong: a wrong house number, a quantity with an extra zero, a customer created twice. Say 5 % of orders contain an error and fixing one — calling, correcting, resending — takes twenty minutes. That is twenty extra minutes a week, 15 hours a year, almost 700 euros. And the errors the customer notices cost more than time.
Rework appears on no timesheet, which is why it is never counted. Our automation ROI calculator has it as a separate item for exactly that reason.
What an honest hourly cost is
Gross salary divided by 40 hours is not what an hour costs. An employee costs the employer roughly 1.3 to 1.4 times gross salary — pension, insurance, holiday pay — and those costs are spread over the hours someone actually works, not the contracted ones. For an administrative role that quickly reaches 35 to 50 euros an hour. If you do the work yourself, your own rate is the honest number: every hour of re-typing is an hour you do not invoice.
The other side of the sum
Automating costs money too, and that belongs in the sum just as honestly. An integration that puts web-shop orders straight into the accounts often lands between 2,500 and 7,500 euros, plus a small monthly amount for hosting or an API licence. And an automation rarely takes over everything: count on about 80 %, because exceptions — an order with an odd address, a customer who calls — remain human work.
Put the two sides next to each other and you get the question that matters: in how many months is it earned back? In the example above, with a 4,500-euro build and 80 % automation, that is about fifteen months. Not spectacular, but real: after three years there is a net of around 6,000 euros — plus two people who no longer re-type.
When it does not pay
This is the part a calculator has to say honestly. A two-minute task that happens once a week costs an hour and a half a year. However annoying it is, no build earns itself back on that. Automating is then the wrong solution — better to set the task up smarter, or simply accept it.
There is one exception: when the task produces errors customers notice, or when nobody wants to take it over and it blocks growth as a result. Then more than time counts. But make that argument explicitly, rather than hiding it in the hours.
How to go about it
- Pick the task the team complains about. Complaints are a good indicator; they are usually the tasks that recur often and produce nothing.
- Measure for a week. How often, how long, by whom, how much goes wrong. Do not estimate — people systematically underestimate frequency.
- Calculate it with the calculator, including rework and a realistic automated share.
- Read the result as it is. Earned back within a year: do it. One to two years: worth the conversation, certainly if errors are involved. Longer: wait until the task grows.
In short
Manual work is cheap per occurrence and expensive per year, and the most expensive part — rework — is written down nowhere. Calculate it with real numbers, put the cost of automating honestly against it, and let the payback decide. Sometimes it says no, and that is an answer too.
Does your task come out well? Read how we automate processes, or send us the numbers — we’ll tell you within two working days what it would cost.