Home/Resources/First automation

How to choose your first automation project

Your first automation project decides the fate of your whole programme. Pick well and it pays for itself. Pick badly and the second project never gets approved.

4 min read · 2 September 2026

The most common reason automation programmes stall is not a lack of ideas. It is the first project. Teams pick a process that is dramatic but rare, or broad but vague, and the project drags, costs more than expected and quietly dies. Then nothing else gets approved for two years. The fix is a deliberate choice, and three criteria scored simply will take you to the right project.

The worst first project

The worst first project touches every department, affects every record and needs six data sources to work. It fails for a simple reason: it cannot be tested step by step, and nobody agrees on what done looks like. The second worst is painful but rare, a process that happens twice a year. It stretches the approval cycle, shows little visible benefit and demoralises everyone. Both feel important, and both teach the organisation that automation does not work, which is the opposite of what a first project should do.

Criterion one: pain

Automate something people hate, ideally something they complain about at least weekly. Pain matters twice. Painful processes have the clearest measured savings, because the people doing the work can describe exactly how long each task takes. And pain creates support: when the people who do the work want the change, they test it properly, report problems honestly and defend it in meetings. Automating something nobody notices is how you end up with a flow nobody uses.

Criterion two: frequency

Frequency turns pain into money. A process run daily by twenty people costs far more than the same process run monthly by two, and it offers many more chances for error. For a first project, choose something that runs at least weekly, and daily is better. Frequency also makes results visible fast. New volumes flow through the automated version within days, the dashboard fills with real data, and the project proves itself while the business case is still fresh in everyone's memory.

Criterion three: risk

Choose a process where a mistake is embarrassing but not dangerous. First projects are not the place for payroll, safety or anything near a regulated boundary. They are the place for records, requests, notifications and reporting. Low risk matters because the team is still finding its confidence. A gentle mistake in an internal log is a training moment; a mistake in a customer invoice is a committee meeting. Keep the first project somewhere the organisation can afford to watch it improve.

A simple scoring method

Score each candidate process from one to five on each criterion: the pain felt by the people doing the work, the frequency of the process, and how low the risk is. Add the scores. The process with the highest total is your first project, provided its score stands clearly above the others. If two processes tie, prefer the one that draws on data already held in one place, because data access is the hidden fourth criterion. Write the scores down. It takes ten minutes, and it takes the politics out of the decision.

What a realistic first delivery looks like, and where to start

A strong first project is small: one process, one team, live within a few weeks. Delivery should include the automated flow, training for the users and one review meeting a month later to see how it is running and what to adjust. That pattern builds confidence, and confidence is what funds the bigger second project. A full list of what we automate is on our services page, and the fixed packages on our pricing page cover the size of project we are talking about.

Ready to pick yours? A free assessment walks the same scoring exercise with you, against your real processes, and ends with a written recommendation you can act on immediately.

Get a free assessment