Most teams know that manual work is expensive, but few can say what one specific task costs. Without a number, the discussion stays at the level of "it takes a lot of time" and nothing gets decided. The good news is that a single task can be priced in about ten minutes, with a pen and a spreadsheet.
This article is for a managing director or head of operations who wants to decide whether one concrete task is worth changing. It is not about the total effect of manual work across a company; for that, see the true cost of manual processes.
Pick one task
Choose a task that repeats and that you can describe in one sentence, for example "copying order details from e-mail into the order system" or "preparing the weekly stock report". If you cannot describe it in one sentence, split it. A good candidate is something that someone does at least weekly and that follows the same steps each time.
Four numbers
You need four numbers. Estimates are fine for the first pass.
- Minutes per run. Ask the person who does it, and if possible watch one run. People usually forget the small steps: opening files, looking for the right version, sending a confirmation.
- Runs per year. Multiply runs per week by the number of weeks the task actually happens. A task that runs daily on working days is about 220 runs a year in most countries; check your own calendar.
- Loaded hourly cost. Use the cost to the company, not the salary on the contract. It includes social and health contributions, equipment, software licences and a share of office costs. Ask finance for a single figure per role; if that is not available, use gross salary plus a fixed percentage that your finance team agrees with.
- People involved. If two people touch the task, count both, including the person who checks it.
The calculation
Put the numbers in a table:
| Item | Example value |
|---|---|
| Minutes per run | 15 |
| Runs per week | 20 |
| Working weeks per year | 46 |
| Loaded hourly cost | 30 (in your currency) |
The example values are invented to show the method; replace them with your own. With these values the task takes 15 minutes × 20 runs × 46 weeks = 13,800 minutes, which is 230 hours a year. At 30 per hour that is 6,900 a year for one task that nobody described as a problem.
Add the hidden part
The direct time is rarely the whole cost. Add three more lines to the same table:
- Corrections. How often does a run contain a mistake, and how long does it take to find and fix it? Even a mistake in one run out of twenty can add up when each fix takes an hour.
- Waiting. If other people cannot continue until this task is done, the waiting time has a cost too. You do not have to price it exactly; note the typical delay in hours and who is waiting.
- Dependence on one person. If only one person knows the steps, note it as a risk. It does not change the yearly cost but it changes how urgent the task is.
Keep these lines separate from the direct cost. Then you can show both numbers: "the task costs 6,900 a year in direct time and probably more in corrections", which is honest and still useful.
When automation is worth it
Compare the yearly cost with the one-off cost of changing the task. The change can be a small script, a rule in an existing tool, a simple internal application or a better template. A simple rule of thumb is to ask for a payback within twelve months. If the change costs less than the yearly cost of the task, it is worth a closer look. If it costs several times more, check whether the task can be made shorter by changing the process first.
Two cautions. First, do not count the whole yearly cost as savings. The time of the person does not disappear; it moves to other work, and that only has value if there is other work waiting. Second, count the cost of maintaining the change. A small automation still needs an owner who notices when it stops working.
If you have several candidates, calculate each one in the same way and sort them by yearly cost divided by estimated effort. The article on measuring manual process cost in a week shows how to collect the minutes when you cannot rely on estimates, and an automation plan for 90 days describes how to schedule the first changes.
Common mistakes
- Using salary instead of loaded cost. This understates the cost by a large margin and makes every change look unattractive.
- Estimating from memory. People often under- or over-estimate by half. Watch one run with a timer.
- Pricing the task but not the mistakes. Rework is where a small task often becomes an expensive one.
- Calculating once and forgetting. Volumes grow and costs change. Put a reminder in the calendar for six months later.
- Starting with the biggest task. A medium task with a clear process is a faster win than a large one with unclear steps. If the choice is between buying and building a tool, see a build-versus-buy decision framework.
A short checklist
- Choose one repeated task and describe it in one sentence
- Write down minutes per run, runs per year and people involved
- Get one loaded hourly cost from finance
- Calculate the yearly direct cost and add corrections and waiting separately
- Compare it with the cost of the change and a payback of twelve months
- Set a date to re-run the calculation
Sources
- Our related articles linked above.
- The method is a general working practice, not a standard; adapt it to your organisation. All example values are invented to illustrate the arithmetic. No client results or market figures are claimed in this article.