After writing about what AI automation actually is, the question we hear most often is always the same: “okay, got it, but where do I start?”
It’s a fair question. There are dozens of tasks that could be automated in any business, and diving into all of them at once usually ends up as a half-finished project that convinces no one. The key is choosing the first one well.
Here are the criteria we use with a client to decide where to start.
Criterion 1: Do you do it more than once a week?
If you do a task once a month, automating it probably isn’t worth the effort of setting it up. But if you do it several times a week — or several times a day — every minute you save multiplies fast.
Ask yourself: how many times this week have I done exactly the same thing I did last week?
Criterion 2: Is the outcome always similar?
The easiest tasks to automate well are the ones that follow a recognizable pattern: answering the same common questions, confirming an appointment, moving a piece of data from one place to another. Tasks that require heavy case-by-case judgment — deciding on a strategy, resolving a conflict with a client — aren’t good candidates yet.
Ask yourself: if I explained this task to someone new on the team, would I give them a clear, repeatable instruction, or would I say “it depends, you’ll see”?
Criterion 3: Is it actually costing you money?
Sometimes the cost is direct time (hours you could spend on something else). Other times it’s an indirect cost: missed appointments because reminders weren’t sent, clients who go to a competitor because you’re slow to respond, copy-paste errors that create extra work to fix.
Ask yourself: if I stopped doing this task by hand, what would happen to my income or my free time?
Criterion 4: Do you already have the data you need, somewhere?
Automating is much easier when the information already exists somewhere — your calendar, your spreadsheet, your online store — even if it’s disorganized. When the information only exists “in someone’s head,” it first has to be written down somewhere before anything can be automated.
Ask yourself: if I had to explain to an outsider how I do this task, could I show them where every piece of data I need actually lives?
Putting it together: a real example
One of our clients, the owner of a small dental clinic, had half a dozen candidate tasks. Running each one through these four criteria, the winner was clear: appointment reminders. They were done daily, always the same way, missed appointments cost real money, and all the information already lived in the calendar.
We started there. It worked well within two weeks, and from that foundation we added more automations with confidence, instead of trying to do everything at once from the start.
The short rule
Look for the task that’s repetitive, predictable, costs money or time, and whose data already exists somewhere. That’s your starting point — not the most impressive one, but the one that gives you a clear, quick win to build on.
If you’d like us to go through your candidate tasks together, tell us how you work and we’ll help you find yours.