SurfingBearHome › Blog › Where to start with automation
Automating repetitive work: where to actually start
Almost no organisation is stuck because it has nothing to automate. They are stuck because there is too much and no clarity on what comes first. Here are four criteria for choosing.
4 criteria Picking the first project What to avoid
The short answer
The purpose of a first automation project is not maximum impact — it is success. One successful project is what creates the second budget and the organisation’s cooperation. So half the selection criteria are about whether it is easy.
Judge on four things: does it repeat often, is the processing sequence fixed, are the inputs and outputs text or files, and is the person who does it cooperative. Work that satisfies all four is your first target.
Criterion 1 — how often it repeats
Automating something done once a month saves almost nothing; build and maintenance cost exceeds the saving. Pick work that repeats at least weekly, ideally daily.
When checking frequency, count actual occurrences rather than relying on how it feels. Work people describe as "constant" is frequently twice a month, and work they dismiss as "nothing much" frequently happens five times a day.
Criterion 2 — how fixed the sequence is
Work that always gets the same processing for the same input is easy to automate. Work where a person makes a different judgement each time is a candidate for standardisation, not automation.
The test is simple: ask whoever does it whether a different person would produce the same result. If the answer is no, you first have to establish why it differs — and that work is half of the automation design.
Criterion 3 — the shape of the inputs and outputs
Criterion 4 — the attitude of the person doing it
Even where it is technically easy, a first project fails if the person who does the work will not cooperate. If they feel automation is eliminating their job, they will keep producing exceptions, and the conclusion will be that this work cannot be automated.
Avoid it by sharing the policy before selecting targets: that recovered time goes to other work rather than reducing headcount, and that the first target will be the task they least enjoy. The second point is particularly effective — remove work nobody wants to do and cooperation follows naturally.
Work to avoid as a first project
Frequently asked questions
We have twenty candidates — how do we narrow them?
Score them against the four criteria, then plot impact against difficulty. Your first project sits in the medium-impact, low-difficulty quadrant. The very highest impact is almost always the very highest difficulty too.
Can we check with a free tool first?
Yes. There is a free assessment tool that takes a description of the work and identifies which steps are automatable and the time that would save. Bring the result straight into a consultation.
How long should a first project run?
A scope you can prove within two weeks. If a scope will not produce a result in that time, narrow it further.
What happens if it fails?
If the scope was narrow, so is the loss — which is exactly why a first project starts small. What matters is less the failure itself than whether you still have the capacity to try again.
Describe the work and we will judge it
Which of your candidates is the first project?
Describe the work you want automated, step by step, and we will set out which steps are automatable, the time saved and the tools required. You can also check with the free assessment tool first.