What is a process “worth automating”
Not every repetitive activity is a good candidate for automation. A process is worth automating when it combines three characteristics: it repeats frequently, it follows clear rules, and it consumes disproportionate time relative to its actual complexity.
Signs a process is ready for automation
- It repeats daily or weekly, with constant volume.
- The steps are the same every time — they don’t depend on subjective, case-by-case judgment.
- The process can be described completely, from input to output, with no unclear areas.
- It consumes significant cumulative time over the course of a month, even if each instance seems small.
How to prioritize — the impact-effort matrix
Don’t automate everything at once. Rank candidate processes along two axes: impact (time saved, errors reduced, compliance risk) and implementation effort (technical complexity, number of systems involved, how well the process is defined today).
| Low effort | High effort | |
|---|---|---|
| High impact | Priority 1 — automate this first, results show up quickly | Project — plan it separately, in stages, don’t mix it with the quick wins |
| Low impact | Optional — automate only if you have spare capacity | Avoid — the cost almost certainly outweighs the benefit |
A process with medium impact but low effort delivers results faster and more visibly than one with high impact but extremely complex to automate correctly — and it builds the internal trust needed for larger projects.
How much time you’re actually losing, in numbers
Estimating “it takes us some time” doesn’t convince anyone to allocate budget. The basic formula, applicable to any repetitive process:
Monthly time lost = No. of executions / month × Time per execution × No. of people involvedFor example: a manual invoicing process run 3 times a day (~65 executions/month), taking 12 minutes each, by a single person, adds up to roughly 13 hours a month — over a week and a half of work per year dedicated to a single repetitive task. At an internal hourly cost of 60-80 RON, that’s an annual cost that’s easy to calculate and set against the cost of automation.
Beyond execution time, the cost of manual errors also matters: a process entered manually into multiple systems typically has a human error rate of a few percent per execution (fatigue, interruptions, ambiguity). The cost of an error isn’t just the correction time — it’s also the time of the person who received the wrong result (client, colleague, accountant), plus the compliance risk if the error ends up in an official document. Even a 2-3% error rate becomes significant at volume: at 500 executions/month, that’s 10-15 corrections, each with its own time cost.
An example from Leadpro’s practice
Illustrative scenario, typical of the companies we work with — not the financial data of any specific client. An operations team received orders by email and entered them manually into an internal invoicing system, then copied the same data into a reporting file for management. The process had clear rules (the same fields, every time), but it was carried out manually three times: once when reading the email, once when entering it into invoicing, once when reporting it. The automation targeted only the step with the fewest exceptions — extracting data from the email and entering it into invoicing — leaving reporting manual where case-by-case adjustments were still needed. The result wasn’t the complete elimination of manual work, but reducing it to the part that genuinely required human judgment.
Concrete examples
- Repeatedly entering the same data into multiple systems.
- Generating documents (invoices, reports) from templates and existing data — see document automation.
- Taking and confirming appointments through conversational channels — see appointment automation.
Quick decision table
| Criterion | Automate | Don’t automate (yet) |
|---|---|---|
| Frequency | Daily/weekly, constant volume | Occasional, under 10 executions/month |
| Rules | Clear, the same on every execution | Depend on case-by-case judgment |
| Exceptions | Few and predictable | Frequent and hard to anticipate |
| Process definition | Can be described completely, no unclear areas | The process itself is unclear or internally disputed |
| Stability | Doesn’t change often | Changes frequently (organizationally or legally) |
When NOT to automate
If the process has many exceptions, depends on subjective judgment, or changes frequently, full automation can cost more than the benefit — sometimes the right answer is to simplify the process first, or to automate only the clear part, leaving the rest manual.
Resources and references
- McKinsey — The State of AI in 2025: Agents, innovation, and transformation — 88% of companies already use AI in at least one business function, but only a third have reached organization-wide scaling — the difference between experimenting and getting real impact comes down to choosing the right processes, not just adopting the technology.
- Download the process prioritization checklist (.txt) — the same criteria from this article, in a format you can fill out for each candidate process in your company.
Checklist
- Does the process repeat at a predictable frequency (daily, weekly)?
- Are the steps the same every time, or do they vary significantly from case to case?
- Can the entire process be described in clear steps, without ambiguity?
- How much time does it consume monthly, cumulated across the whole team?
- What happens when an exception occurs — who decides?
Risks to consider
- Automating a poorly defined process multiplies the problem instead of solving it — clarify the process first, then automate.
- A process with many rare exceptions can cost more to fully automate than to leave partially manual.
- Automation without monitoring can run incorrectly for a long time before anyone notices.