Short answer: start with whatever is repetitive, rule-based, and mildly soul-destroying — not whatever feels most impressive. The tasks that make the best first automation are rarely the ones people expect, and that's actually good news, because it means you don't need a big, complicated project to get real value on day one.
When people ask what to automate first, they usually start by thinking about their biggest, most business-critical process. Understandable instinct — wrong place to start. A better filter is to ask, for any given task:
Does it follow the same steps every time? If the answer involves "well, it depends," it's not a great first candidate. Automation loves predictability. Save the judgment calls for later.
Does it happen often enough to matter? A fiddly task you do twice a year isn't worth automating, even if it's genuinely annoying. A slightly less fiddly task you do fifteen times a week absolutely is.
Does it eat time without adding much value? Copying data from one place to another, re-typing the same email, chasing the same status update — this is where automation earns its keep. Nobody's proud of doing it manually; nobody will miss doing it.
If a task ticks all three, it's a strong first candidate. If it only ticks one, it can wait.
A handful of tasks show up again and again as good starting points, almost regardless of industry:
None of these are exciting. That's rather the point — they're exactly the sort of thing nobody wants their time going on anyway.
Some tasks look tempting to automate first but tend to cause more trouble than they save, early on:
Anything that still involves real judgment. If a human needs to weigh up context, tone, or exceptions, automating it too soon usually means building something that gets it wrong often enough to lose trust. These can often be partly automated later, once you understand the pattern of exceptions properly.
Your most complex, highest-stakes process. It's tempting to go straight for the workflow that would save the most time if it worked. But it's also the one with the most edge cases, the most stakeholders, and the most cost if something breaks while you're still learning. Build confidence on something smaller first.
Anything you haven't mapped out properly yet. If you can't describe the process step by step, including the exceptions, it's not ready to automate — it's ready to be written down.
If you've got a shortlist of candidates, weighing them up on two things — how often the task happens, and how easy it would be to automate — usually sorts the order for you:
| Quick to automate | More involved to automate | |
|---|---|---|
| Happens often | Do this first | Worth doing, plan properly |
| Happens rarely | Nice to have, low priority | Usually not worth it yet |
Almost everyone's first automation lands in the top-left box: something frequent, and something simple. That's not a coincidence — it's the combination that proves the value fastest with the least risk.
Don't start with your biggest headache. Start with your most repetitive one. The best first automation is usually boring, happens constantly, and follows the same steps every single time — which also makes it the safest place to learn what automation can (and can't) do for you, before you point it at something that really matters.
Not sure which of your own repetitive tasks would make the best starting point? That's exactly the kind of thing we help figure out — get in touch and we'll help you spot the quick wins in your own workflow.