Business process automation —
what can actually be automated.
Process automation means that repetitive work — copying data between tools, sending reminders, producing reports — happens without a person doing it. In small and medium companies this usually covers five areas: enquiry handling, follow-up, customer service, onboarding and reporting.
Below we break it down: which processes are worth automating, how to tell whether one is suitable, what automation will not fix, and where to start.
You are not buying software.
You are connecting what you already have.
The most common misunderstanding: automation is not about rolling out a new program. It is about connecting the tools you already have, so that data moves between them by itself.
A client fills in the form on your site, the enquiry lands in an inbox, somebody reads it, retypes the details into the CRM, replies to the client and adds a task to the calendar. Four actions, fifteen minutes or so, and the whole thing depends on whether that one person happens to be at their desk.
The form submission goes into the CRM assigned to the right salesperson, the client gets a confirmation within a minute, a notification appears in Slack and a dated task appears in the calendar. Nobody retypes anything.
Nothing new was bought. What changed is that the tools started talking to each other.
The five processes automated most often.
Five areas where automation pays back fastest. Each can be implemented on its own.
How it usually goes: enquiries arrive from the form, by email, through Facebook and by phone. Some end up in the CRM, some stay in somebody’s inbox. Response time depends on who happens to look.
What changes: all channels converge in one place, every enquiry gets an owner and an automatic confirmation, and nothing is left without a next step.
When it is most urgent: when you cannot answer the question “how many enquiries did we get last month and what happened to them”.
Lead management automation →How it usually goes: the quote went out, the client never replied, the salesperson was going to chase it in a week. They didn’t, because they had three other things on.
What changes: once the quote goes out, a reminder sequence starts and stops by itself when the client replies. The salesperson doesn’t have to remember — they react when something happens.
When it is most urgent: when you have more quotes out than you can keep track of in your head.
Sales follow-up automation →How it usually goes: the same handful of questions comes back every day. The team answers them by hand, from scratch each time, and the difficult cases drown among the simple ones.
What changes: tickets are classified, the repetitive ones are answered automatically from the knowledge base, and those that need a person reach one with full context.
When it is most urgent: when most of your support team’s work is answering questions that have already been asked before.
Customer service automation →How it usually goes: once the contract is signed the emails begin — documents, access, invoicing details, agreeing a start date. Every client goes through it slightly differently.
What changes: once the quote is accepted, one sequence runs: documents to sign, payment, kick-off form, team tasks, welcome message. Every client starts the same way.
When it is most urgent: when the first two weeks of a new engagement look different every time.
Client onboarding automation →How it usually goes: someone spends Friday afternoon pulling numbers from three tools into a spreadsheet and assembling a report. It takes half a day and nobody is sure the data is current.
What changes: the data arrives automatically, the report appears on Monday morning without anyone touching it, and anything out of the ordinary triggers an alert straight away rather than a month later.
When it is most urgent: when you make decisions based on figures that are two weeks old.
Reporting automation →How to tell whether a process is suitable.
It repeats regularly
At least a few times a week. Something done once a quarter rarely pays back the cost of implementing it — even if it takes half a day each time.
It has clear rules
It can be described in sentences like “if the enquiry is about X, it goes to Y”. If every case calls for judgement, it is not a process — it is expert work.
The data is digital
The information lives in an email, a form, a spreadsheet or a CRM. If the key data is on paper or in somebody’s head, it has to be got out of there first.
A mistake can be undone
An email sent to the wrong person is a five-minute problem. A wrongly issued invoice or a misdirected payment is a different category of risk, and calls for safeguards that raise the cost of implementation.
If you can, it is suitable for automation. If after the third sentence you start saying “well, it depends, because sometimes…” — it is not a process yet. It is a habit shared by a few people who do similar things in different ways. Automating that habit will cement the mess rather than remove it. First someone has to decide what the process should look like — and that is part of the work we do at the start of every implementation.
What automation will not fix.
If the company has no agreed way of handling enquiries, automation will give you a faster mess. Before we connect anything, somebody has to decide how it should work.
Automation is very good at “if this, do that”. It cannot handle “this client matters, let’s make an exception”. The best implementations don’t remove the person — they move them from doing the work to overseeing it.
If a salesperson isn’t calling leads because they don’t want to, a Slack reminder will change nothing. All it changes is that it becomes visible that they aren’t calling — which can be valuable, but it is a different kind of benefit.
A faster reply improves your chances of getting a conversation. You still have to have the conversation.
A process run twice a month that takes fifteen minutes adds up to six hours a year. Automating it will never pay back. Do the maths before you start.
How long it takes and what it costs.
| Scope | Price net of VAT | Time |
|---|---|---|
| One process step, 1 integration | PLN 3,600 | 5 working days |
| A full process, up to 3 integrations | PLN 6,900 | 10 working days |
| System — 3–6 connected processes | PLN 16,000 – 35,000 | 3–6 weeks |
| Architecture — multiple departments | PLN 36,000 – 120,000 | 2–4 months |
Timelines are in working days, counted from when we receive access to the tools. Four things drive the price: the number of tools to connect, the complexity of the rules, the state of the data in the company and any personal-data protection requirements. On top of the implementation cost come the tools themselves — usually from PLN 0 for a self-hosted setup to a few hundred zloty a month on cloud platforms.
An example. An employee loses 6 hours a week retyping enquiries and sending first replies. That is 26 hours a month, at PLN 70 an hour — PLN 1,820. Subtract PLN 300 for tools and PLN 1,520 remains. An automation costing PLN 6,900 pays back in just under five months. If your result comes out above twelve months, you are probably automating the wrong process.
What it is built with.
Three platforms cover most of what small and medium companies need.
Make
Fastest to get going, a visual editor and a large library of integrations. A good choice when implementation time matters and you have no technical team.
n8n
Open-source software you can run on your own server. Data never leaves your infrastructure, which matters with sensitive data and GDPR requirements. No licence fees, but the server has to be maintained.
Zapier
The simplest, the most expensive at higher volumes, and the best fit for straightforward connections between popular tools.
Claude AI
A language model we use where the process has to read something, understand it and make a call — classify a ticket, extract data from an email, write a summary. It does not replace the automation platform; it performs individual steps inside it.
Choosing the platform is a technical decision, not a business one — and we make it. What matters far more is whether the process was designed well. A badly designed one will run badly on all of them — including with the best language model inside it.
The costliest mistake is building an agent where a single "if-then" would do — the second costliest is the reverse. One question settles it: would two people doing this step make the same decision? A well-built process is usually 80% rules and 20% model. The step-by-step test and the cost of each option →
Four of the five are not technical mistakes but preparation mistakes: an undocumented process, everything at once, no owner on the company side and no measurement before the start. The fifth is skipping the team at launch. How to avoid each of them →
Where to start.
Start with the simplest process that meets three conditions at once:
That one process takes two weeks to implement, you see the effect in its first week of running, and only then do you decide whether to go further. It is a far cheaper way of finding out whether automation makes sense in your company than a three-month project.
Most companies that start with a big implementation end up with a system the team doesn’t use — because it was built before anyone understood how the process actually works.
Frequently asked questions.
No. We work with the tools you already use. If something is missing we suggest a simple, proven option — we don’t force anyone to replace their whole toolset.
5 to 10 working days from the moment we receive access. Larger systems connecting several processes: 3–6 weeks.
In small companies, usually not. It takes the repetitive work off the team so there is time for the work that needs a person. The real effect is usually hours recovered, not jobs cut.
We build systems so they can be modified. Small changes are covered by the monthly care package; larger ones are quoted separately.
It depends on the platform chosen. With sensitive data we recommend a self-hosted setup — the data then never leaves the company’s own infrastructure.
The one that can be described most briefly. A shorter description means clearer rules, which means faster implementation and less risk of surprises.
Yes, and it is the recommended approach. You launch the first process on its own, check the effect, and only then decide about the next one. We design systems so they can be joined up later.
Let’s find the process that will pay back
fastest for you.
30 minutes, an online call, no obligation and no sales deck. We will go through the processes in your company and point to one to start with. Including when the answer is “none of them yet”.
Book a free consultation →