What to Automate First (and What to Leave Alone)

Frequency first. A task that happens forty times a week is worth automating even if each instance is small. A quarterly task almost never is, however annoying it feels.

Consistency second. Automation handles predictable shapes well and judgement calls badly. Booking an appointment is consistent. Deciding whether to waive a fee for a long-standing customer is not.

Cost of failure third. Some tasks lose revenue when they're skipped, like following up a quote, requesting a review, or responding to an enquiry. The lost revenue costs you far more than the time they take.

For most local businesses the first automation is answering enquiries and booking them in, because it scores highest on all three: it happens constantly, follows a pattern, and is expensive to miss.

Second is usually follow-up: the quote that never got chased, the review that never got requested, the customer who lapsed. Individually trivial, collectively significant, and reliably neglected because they're never urgent.

Neither of these is exciting, which is why they get skipped in favour of something more novel. They're also where the money is.

Anything where being human is the product. Complaint handling, difficult conversations, the judgement call on an unusual job. Automating these saves minutes and costs relationships.

Anything that rarely happens. Setup and maintenance overhead will exceed the benefit, and automations you rarely use break without anyone noticing until you need them.

And anything you can't currently describe as a set of rules. If you can't write down how the decision gets made, you can't hand it over. That's not a technology limit; it's a sign the process isn't defined yet, and defining it is the actual work.