Automation saves time only when the work you automate is already in order. If a process depends on who happens to do it, a tool will just repeat the same mess faster. This check helps you see whether a process is ready before you pick a tool.
Short answer
A process is ready for automation when it repeats often, follows clear rules, receives data in a consistent digital form, has an owner, and you know how long it takes today. If every case is handled differently, sort out and write down the process first, then automate it.
Work worth automating
- Work that repeats often. The same task is done several times a week, the same way.
- Work that follows rules. A decision can be written as a sentence such as “an enquiry about service A goes to person B”.
- Work with structured input. Data arrives through a form, a system or a spreadsheet with the same fields.
- Work where time gets lost. Retyping, forwarding and reminders are where time and data usually slip away.
Work not to rush
- Rare tasks. If something happens a few times a year, setup and upkeep can easily cost more than they save.
- Decisions that need judgement. Negotiations, complaints and special customer requests need a person.
- A process that keeps changing. If the rules change every month, automation will always lag behind practice.
- Scattered data. When data sits in emails, on paper and in people’s heads, the first step is to get it in order.
1. Describe the process as it is
Write down the steps from start to finish, the way the work is really done today, not the way it should be. For each step, note who does it, in which tool and where the work waits for someone. The biggest gains usually sit exactly where work waits or where data is copied by hand from one system to another.
2. Measure the baseline
Before you change anything, record how many times a month the task is done, how long one case takes, how many errors there are and how long a customer waits for a reply. Without those numbers you will not be able to show later that automation helped. We describe the same principle in Before you buy an AI tool.
3. Check the input data
Look at the form in which data enters the process. A form with required fields is easy to automate, free text in an email is harder, and scanned paper is hardest. Often the first and cheapest step is to tidy up the input, for example by having the contact form ask for the type of enquiry.
4. Write down the rules and the exceptions
For every decision in the process, write down the rule. Then write down the exceptions: what happens when data is missing, when a case fits no rule, or when a customer asks for something special. If there are almost as many exceptions as rules, the process is not ready yet.
Put the process in order first, then automate it.
5. Name an owner and a human check
An automated process still needs someone who is responsible for it: who checks that it works, who changes the rules and who handles the cases the system cannot. Decide also where a person reviews the result before it reaches the customer, especially when the tool uses AI.
6. Check personal data
If the process handles personal data, the Serbian Law on Personal Data Protection applies. Three things are worth checking before rollout:
- A contract with the vendor. If the tool processes data on your behalf, the vendor is a processor. That processing must be governed by a contract or another legally binding act in writing, which includes electronic form (Article 45).
- Automated decisions. A person has the right not to be subject to a decision based solely on automated processing if it has legal effects on them or significantly affects their position. The law sets out exceptions and safeguards, including human involvement in the decision (Article 38).
- Impact assessment. Where processing, especially with new technologies, is likely to result in a high risk to people’s rights, the controller carries out an impact assessment before processing starts (Article 54).
This is an overview, not legal advice. For a specific process, check your obligations with a lawyer or your data protection officer.
Example: routing website enquiries
The example is generic, with no client data. Enquiries from the contact form land in one shared inbox. Someone reads them, decides who they belong to and forwards them. When that person is away, enquiries wait.
- Process. Form, shared inbox, manual forwarding, reply to the customer.
- Baseline. Number of enquiries per month and time to first reply.
- Input. The form gets a required “Type of enquiry” field.
- Rules. Each enquiry type goes to one person. An enquiry with no clear type goes to the process owner.
- Owner. One person checks that enquiries reach the right place and updates the rules when needed.
- Data. The form collects a name and contact details, so you check where the tool stores data and whether there is a contract with the vendor.
After a few weeks, compare time to first reply with the baseline. If it is not shorter, forwarding was not the problem.
When it is better to wait
If the process has no owner, if you do not know how long it takes today, or if the team does not agree on how it should work, a tool will not fix that. One meeting where the process is written down and agreed is worth more than another subscription.
How we approach it
We start from your processes and data, not from whatever tool is popular right now. First we document goals, constraints and the current state, then we compare options by impact, effort, cost and risk and give a clear recommendation. You can end the engagement after any stage and keep a document you can act on. More on our Consulting & Technology page.
In short
Automate work that repeats often, follows rules and has structured input. Before that, describe the process, measure the baseline, write down the rules and exceptions, name an owner and check your personal data obligations. If you would like to go through your process together, get in touch.
