A managing director sent me his calculation last month. Not for our work. His own math of what manual work had cost him before automation. One employee who copied data between three systems for three to four hours a day. He had used 25 euros an hour. The right figure was double that. That flips every payback calculation, in both directions.
That is the thesis of this piece. Any automation that does not pay back within three months is the wrong automation. And most companies do not know whether theirs pays back, because the first number is wrong.
According to Destatis, an hour of work in Germany cost 45.00 euros on average in 2025, 3.6 percent more than the year before. In the transport and storage sector the figure is lower, at 35.80 euros. These are employer costs per hour worked: gross pay plus social contributions plus paid absence. Office space, IT workplace and the manager's time come on top.
Whoever calculates with gross hourly pay halves the manual work on paper. Whoever uses 36 to 45 euros is close for dispatch. For a dispatcher on 3,500 euros gross you land at 42 to 55 euros depending on region. (Sounds high? Spread the vacation and sick days over the hours actually worked.)
Three figures, multiplied: how often a day, how long per case, full hourly rate. That gives the annual cost of the manual work. Against it stand build and operation of the service.
Three worked examples, two from projects and one as a model calculation, all rounded:
Arrival notices, forwarder with 40 trucks. Three to five calls per dispatcher and day, about 45 minutes with callbacks, two dispatchers. An hour and a half a day times 45 euros times 250 days: about 17,000 euros a year. Build: three weeks. Payback in month two.
Order capture, wholesaler. Three and a half hours a day, one person, plus two to four wrong orders a day with 20 minutes of rework each. About 40,000 euros a year, without the customer side of the errors. Build: two weeks. Payback in month one.
Portal confirmation, model calculation. Assume 40 confirmations a day, four minutes each, one person: just under three hours. About 30,000 euros a year. Build: three weeks. Payback in month two.
The pattern: with well-chosen processes, break-even sits between one and three months. Company size barely matters. After two years the net effect per process is in the high five figures. That is not an outlier, it is the normal case when the right process is picked.
Most vendors show build against hours saved. Four items are almost always missing:
Three cases where we advise against it, and we do so in every third first call:
A caveat on the examples: two come from companies with 20 to 200 employees, the third is a model with typical values. The hourly rates are averages. Payback depends on the rate of the person doing the work today. An apprentice changes the math compared with a dispatcher with 15 years of experience.
Between fast and slow payback lies neither the tool nor the budget. It is documentation. If it is clear before the build what the process does, exceptions included, the project runs in two to three weeks. If the provider has to discover the process first, it takes six to eight. There is a separate piece on that: Optimizing processes: count first, automate second.
If you want your own number: the process cost calculator turns role, time per day and hourly rate into annual cost. No sign-up. Send me the result and the process behind it. I will tell you whether it fits the three-month rule. Ask for a process check: 30 minutes, no sales pitch.
