The most expensive automation project is not the one with the big invoice. It is the one that fixes a process nobody experienced as a problem. It gets delivered, it works technically, and two months later nobody uses it.
So the first question is not “which technology” but “which process”. Here is how we choose one.
Three signs a process is worth automating
- It repeats. Dozens of times a week, not three times a year. The more often, the faster it pays back.
- It has rules. There is a correct answer that can be explained to a new hire in five minutes.
- Delay costs money. The customer who waits leaves, the invoice processed late holds up a payment, the report that slips postpones a decision.
If all three are true you have a good candidate, even if the process looks dull. The best automations are boring.
Three signs it is not
- It needs judgement. If two experienced employees would answer differently and both would be right, this is not work for an agent.
- It is written down nowhere. A process that lives only in one person’s head gets documented first and automated second. The reverse order produces a bot that is guaranteed to be wrong.
- It changes every month. If the rules keep being rewritten, you will pay for adjustments every time and never come out ahead.
How to find it in 30 minutes
You do not need an external audit for the first step. Take an ordinary week and answer three questions:
- Where do the people who are waiting pile up? Customers without a reply, colleagues waiting for an approval, orders sitting still.
- What does an employee do several times a day with the same words or the same clicks?
- What gets copied by hand from one system into another?
The point where all three overlap is usually the process worth starting with.
The four-question test
Before signing anything, answer in writing:
- How many hours a week does the process consume now, and whose hours are they?
- Who would notice if it disappeared tomorrow: a customer, an employee, or nobody?
- Where does the data it needs come from, and does all of it exist in written form?
- What does success look like in numbers three months from now?
If the answer to the last one is “we will see”, the moment has not arrived. A project without a number to check against can be neither succeeded nor failed, only extended.
Common mistakes on a first project
- Picking the most visible process instead of the most painful one. Visibility impresses management, pain produces the savings.
- Automating everything at once. One process delivered and used is worth more than five half-done.
- Leaving the team out of the conversation. The people doing the work every day know where it hurts; if they learn about the agent after the rollout they will sabotage it politely, by simply not using it.
- Counting saved hours without deciding what they turn into. Freed-up time with no destination fills itself with different routine.
What changes after the first process
The second project goes twice as fast, because the data is already in order and the team has seen that what is being taken away is not their work but the unpleasant part of it. That trust is the real result of the first process, more than the hours saved.
Which is why we suggest starting with something small and indisputable, rather than the project that would look best in a presentation.
If you want to walk the list above through your own processes with us, we run a free process audit: 30 minutes online. And if the conclusion at the end is that nothing is worth automating yet, we will say so plainly.