Home/Guides/How to know what to automate first
Efficiency

How to know what to automate first

Most companies know they should automate more — but not where to start. Here's a simple method for prioritising the right things.

Guide6 minEfficiencyUpdated Sep 2026

Quick overview

In short: Automate first the process that is most time-consuming, recurring and error-prone – and at the same time easiest to connect with your existing systems. Automating the wrong process first is a common and costly mistake, so don't guess – rank.

How to get started:

  1. List 5–10 processes that take time every week.
  2. Score each process on the four factors: time spent, frequency, error risk and connectability.
  3. Build away the highest-ranked process first, measure the impact before and after, then move on to the next.

Count honestly: put a rough figure on the time saved, but always measure the impact yourselves – the gain only materialises if the automation is actually put to use.

Automating the wrong process first is a common and costly mistake. You spend time on something that's barely used, while the genuinely heavy work carries on as before. The solution isn't to guess — it's to rank.

Four factors that decide

For each candidate process, ask four questions. Together they give a clear order of priority.

  • Time spent — how many hours per week does it take?
  • Frequency — does it happen every day, or rarely?
  • Error risk — how often do things go wrong, and what do the errors cost?
  • Connectability — is the data already in systems that can be integrated?

A process that is time-consuming, recurring, error-prone and easy to connect is an obvious first candidate. One that happens rarely and requires human judgement belongs further down the list.

Do the maths — but honestly

Put a rough figure on the time saved. A process that takes 8 hours a week and can be halved frees up roughly 4 hours a week — but only if the automation is actually put to use.

Worked example: a quoting process that takes 6 h/week and is automated down to 2 h frees up ~4 h/week. The numbers are a worked example — always measure the impact yourselves before and after.

Start small, measure, build on

Don't automate everything at once. Take the highest-ranked process, build it away, measure the impact and move on to the next. A small first step gives you data to base the next decision on — and builds trust internally.

01

List

Write down 5–10 processes that take time.

02

Score

Rank them on the four factors.

03

Build

Automate the highest-ranked one first.

04

Measure

Compare before and after, then move on.

If you want to skip the guesswork, we do the mapping for you: a review of your systems and processes with ten ranked opportunities, cost, time savings and a 90-day plan.

FAQFrequently asked questions

Frequently asked questions.

What should you automate first?+
The process that is most time-consuming, recurring and error-prone — and at the same time easiest to connect with your existing systems.
How do you measure the impact?+
Measure time spent and error rate before and after. Use worked examples to estimate, but trust your own measurements.
Do we have to automate everything at once?+
No — start with one process, measure, and build from there.

Want the whole map? We list ten opportunities with cost and impact.