What does automation cost and what does it save?
The quick estimate tells you what implementing automation would cost you and how long it would take to pay for itself. You also see how much the repetitive tasks you select cost today per year and how much of that work can be automated. You answer eight questions in about two minutes, and the estimate is calculated in your browser. It is an order-of-magnitude estimate: it shows which task to look at first, not what exactly you should implement.
2 minutes
eight questions
Free
no contact details needed
Order of magnitude
in euros and months
Step 1 / 4: Company
About two minutes
What industry are you in?
How many people work in the company?
We don't store your answers until you give us your email. The estimate is calculated in your browser.
Price
How much does implementing automation cost?
Based on Empirica's implementation history, implementing one automation target costs €5,000–11,000 when the work is moving data or rule-based processing. When part of the content needs interpreting, the price is €8,000–18,000, and when interpreting content is the core of the work, €12,000–28,000. The biggest cost drivers are where the data lives, how often the process changes, and how much judgement is involved. In the hardest case the information is on paper, the process changes constantly and the work needs judgement almost every time. The price is then nearly double that of the simplest case. Each further target costs less than the one before, because the environment, integrations and monitoring are implemented only once. That makes a set of several targets clearly cheaper than the same number of separate implementations.
| Nature of the work | Implementation per target | Examples |
|---|---|---|
| Moving data and rule-based processing | €5,000–11,000 | Moving data from one system to another, Manual data entry and checking, Compiling reports by hand |
| Part of the content needs interpreting | €8,000–18,000 | Recording orders and assignments, Processing and coding purchase invoices, Scheduling and resourcing by hand |
| Interpreting content is the core of the work | €12,000–28,000 | Classifying and routing customer messages, Reviewing contracts and documents, Screening applications and applicants, Preparing quotes and pricing |
The prices are ranges for one target, before data location, rate of change and amount of judgement are taken into account. They are based on Empirica's own implementation history, not on a public source.
Calculation
How is the estimate calculated?
The calculation starts from the weekly hours you report, which the tool multiplies by 45 working weeks and the hourly cost of the work. The default hourly cost is €45, meaning salary including employer contributions, and you can change it in the results view. Every task has its own automation rate: moving data from one system to another can largely be automated, pricing a quote cannot. The implementation cost is calculated one target at a time and summed, and it is affected by where the data lives, how stable the process is, and how much judgement is involved. The payback period is the implementation cost divided by the annual saving, calculated from the midpoints of the ranges.
Calculation assumptions
- 1
45 working weeks per year
after holidays and public holidays
- 2
€45 per hour
salary including employer contributions, changeable in the results view. Anchored to the Statistics Finland Labour Cost Survey, according to which an hour worked cost an average of €34.4 in 2020
- 3
Automation rate per task
higher for rule-based work, lower for work that needs judgement
- 4
Implementation cost one target at a time
the second and later targets are cheaper because the groundwork is shared. The prices are based on Empirica’s own implementation history, not on a public source
- 5
All results as ranges
a single precise figure would be a promise a quick estimate cannot keep
Limits
What does the estimate not tell you?
The estimate does not tell you what you should implement or with which technology. It also knows nothing about your volumes, your exceptions, or the state of your systems, because it is based on eight answers rather than measured data. The tool splits the hours you report evenly across the tasks you selected, which is a crude assumption: in reality one task usually takes considerably more time than the others. Measuring that split from real data is exactly what the Automation Assessment does.
Comparison
What is the difference between the quick estimate and the Automation Assessment?
The quick estimate is free, takes two minutes, and is based on your own estimates. It gives you the order of magnitude and the ranking: which task is the biggest. The Automation Assessment is fixed-price and takes 10 working days. It covers 3–5 interviews, measures the actual volumes and turnaround times, and produces a prioritised roadmap where every target has a value in euros and a payback period. The quick estimate tells you whether this is worth looking into, the assessment tells you what to do.
| Quick estimate | Automation Assessment | |
|---|---|---|
| Price | Free | Fixed, agreed in advance |
| Duration | 2 minutes | 10 working days |
| Based on | your own estimates | measured data and 3–5 interviews |
| Output | order of magnitude and ranking | a prioritised roadmap with euro figures |
| Answers | whether this is worth looking into | what to implement and in what order |
| Credit | not applicable | the fee is credited when implementation starts within 60 days |
If the quick estimate shows that automation is not worth it, that is a perfectly good outcome. It saves you both the cost of the assessment and the cost of the implementation.
Questions
Frequently asked questions about automation cost and the quick estimate
What does automation cost after go-live?
After go-live you pay for using the AI model and the infrastructure. Running one automation typically costs tens of euros per month, because AI usage is billed in tokens, meaning the amount of text read and written. That is a different figure from the implementation price. The payback period in the quick estimate does not include running costs, because they are usually small compared with the annual saving. On an off-the-shelf automation platform, run, tool and integration fees often come on top of tokens, so ask the vendor about them separately.
What does running AI cost per month →What is a good payback period for automation?
A well-chosen automation target pays for itself in under 12 months, which is also the typical level for the top targets in an Automation Assessment. The quick estimate treats 24 months as the upper limit: if the payback period is longer, the tool does not propose an assessment. The tool never shows a payback period under 2 months, because go-live, testing and parallel running always take their own time. The payback period is the implementation price divided by the annual saving, multiplied by twelve.
When is automation not worth it?
Automation is not worth it when the task changes constantly, when the information exists only on paper and the volume is small, or when the work needs expert judgement almost every time. In those cases the implementation cost rises and the achievable saving stays small. If the payback period in the quick estimate exceeds 24 months, the tool says so directly and does not propose an assessment. An automation nobody uses six months later costs more than doing nothing.
When an off-the-shelf tool is enough and when it isn’t →What if the company is too small for a paid assessment?
Then the tool routes you to the self-service path and does not offer an assessment. You still see the load map and the per-target findings on where to look. At a smaller scale an off-the-shelf tool is usually enough, and a custom implementation only becomes relevant once volume grows or the off-the-shelf tool runs out of road. You can run the estimate again once the workload has changed.
Is the tool free, and what happens to my answers?
The tool is free, and seeing the estimate does not require contact details. The calculation happens entirely in your browser, and no answers are sent anywhere until you give us your email address. If you ask for the per-target findings by email, we store your answers and your address so we can get in touch. We don't send anything else without separate permission.
Which tasks can you estimate?
The tool covers the 10 most common repetitive tasks in Finnish SME and mid-sized B2B companies: moving data from one system to another, manual data entry and checking, compiling reports by hand, recording orders and assignments, processing and coding purchase invoices, classifying and routing customer messages, reviewing contracts and documents, scheduling and resourcing, screening applications, and preparing quotes and pricing. You can select at most four at a time.
Sources
Where do the input figures come from?
- An hour worked cost the employer 34.4 euros in 2020 (in Finnish)
Statistics Finland, published 19 December 2022
Labour Cost Survey, average across all sectors. The tool's default of €45/h is Empirica's estimate of what white-collar work costs today, derived from that figure: white-collar roles sit above the average and costs have risen since 2020.
The other figures in the calculation are Empirica's own assumptions: 45 working weeks per year, the per-task automation rates, and the implementation cost ranges and target multipliers. They are based on our implementation history.
These sources were last checked on 4 September 2026.
The quick estimate gives a rough number. The assessment gives the exact one.
The Automation Assessment walks through your operations and tells you which targets are worth implementing first and how much they save. The fee is credited in full when implementation starts within 60 days.