The Yearly Price of a Manual Process
Three numbers about a process — and the annual cost of manual work becomes visible, along with the share of the team's working time and the hours automation takes off.
Your numbers
Items that pass through the operation: invoices, requests, inquiries, statements.
Pure hands-on time spent on an item, excluding waiting in a queue.
Everyone who touches these items at least occasionally, including reviewers.
The full cost of an employee — taxes and contributions included.
Only employee time spent on the items themselves is counted. Waiting in a queue, the cost of errors and development are not included here.
A typical range for a single production operation. With a removable share of 40% or more, the annual saving covers the one-off investment within the system's first year of operation.
- Hours per month
- 276 hours
- Hours per year
- 3 312 hours
- People fully occupied
- 1,7
- Share of the team's time
- 28,0%
- Hours removed per year
- 1 656 hours
- Removable share of costs
- 1 111 000 RUB
- Left to people
- 1 110 000 RUB
The calculation gives an order of magnitude based on average norms. Exact hours only become visible after timing the work on real documents.
How the calculation works
Three numbers form the basis: the volume of items per month, the minutes per item and the cost of an hour. Everything else is built on top of them.
Volume is multiplied by minutes and by the rework coefficient. That gives minutes per month, from which come hours, and then the year. The cost of an hour is derived from the salary at a norm of 164 working hours per month.
Then the hours are converted into two understandable figures. The first is how many people the process occupies fully, in full-time-equivalent terms. The second is what share of the team's working time it eats up.
What affects the result most
Minutes per item. An error in this field is multiplied by the whole annual volume, so imprecision costs more here than in salary or headcount. The difference between "about ten minutes" and "about fifteen" on a flow of a thousand items is roughly a thousand hours a year.
The second strongest lever is volume. It stretches the result linearly and is usually known precisely: the figure sits in the accounting system. That is the one to plug in, while minutes are measured separately.
Headcount does not affect the final sum. It is needed for the share of working time — the indicator that answers the question "is the team half occupied by the process, or fully".
The share automation takes off
How uniform the flow is determines what part of the hours goes to the machine. The rough guides: a heterogeneous flow — about 30%, a flow with a core of standard cases — about half, a uniform conveyor — up to 70%.
The numbers can be checked against your own data in an hour. A sample of 50 items from last month is taken, and each is tagged: standard, with a deviation, unique. The share of the first group plus half of the second is a workable estimate of the removable hours.
The remainder still requires a person. In the calculation it is shown on a separate line, so that the annual saving is not read as zeroing out the operation.
When the arithmetic turns into a project
The first condition — the hours are concentrated, not scattered. A thousand hours in one task pays back automation noticeably faster than the same hours smeared across fifteen small actions.
The second — the rules of the process can be written down. Where the decision is made from a document and a regulation, automation works. Where the decision rests on understandings inside an experienced employee's head, the rules are first extracted and written down, and that is a stage of its own.
The third — volume is stable or growing. A seasonal spike lasting two months a year is counted through the average annual flow, otherwise the picture is distorted in both directions.
How to read the share of working time
A value of around 30% means the process is one of several tasks at the operation. Automation here frees up hours inside existing roles.
A value above 70% says the operation is built around this process. A rollout like that changes the structure of the work, and preparing people is planned in advance — together with the technical steps.
A value above 100% occurs more often than it seems. Usually it is overtime or hidden involvement of adjacent departments that were not in the original headcount.
When it is too early to count
The process has been running less than three months and is still changing. A measurement will capture a temporary state, and a quarter later the numbers will be different.
Items arrive as scans and photos with no single format. The annual sum will be calculated correctly, but the removable share will be overstated: the input is put in order first, and automation is counted after.
Volume is under a hundred items a month. Here the result usually lands on a compact operation, and the first step is a description of the process, with the calculation repeated after six months of growth.
What to do with the number you get
The annual sum is the upper bound of what makes sense to invest in a project over the same period. The removable part is a realistic guide to the return.
It is convenient to repeat the calculation for two or three processes in a row. Comparing the annual sums shows which operation to start with, and that choice affects the outcome more than the choice of technology.
FAQ
Where does the hourly cost come from
The full monthly cost of an employee is divided by 164 hours — the average monthly norm of working time under the production calendar. Taxes and contributions are built into the salary, otherwise the hour comes out understated by roughly a third.
Why is the result shown per year rather than per month
The monthly figure is comparable to current expenses and looks familiar. The annual figure is comparable to a project budget and the operation's payroll — and that is what the decision to launch is based on.
How can minutes per item be measured without special tools
A sample of 20 items over an ordinary week is taken. Pure working time is timed, without pauses and without waiting for a reply from adjacent teams. The median of the sample is closer to the truth than the average: a couple of heavy cases pull the average up sharply.
How is the rework share calculated
Over two weeks, count how many items came back for rework at least once. Twenty returns per hundred is a coefficient of about 1.2 on time, because the second pass is usually shorter than the first.
What does a share of working time above 100% mean
The volume does not fit into the stated headcount. Usually this means overtime, help from adjacent departments, or participants in the process who were not counted. A useful signal: the list of everyone who touches the items is worth clarifying before the project starts.
What the calculation does not account for
The cost of errors, the customer's waiting time and the one-off investment in development. These items are counted separately and often turn out to be larger than the payroll part itself.
Check it on your case?
We will look at your process and give numbers for it, not for averages.
Related tools
Updated: August 7, 2026