How to Know If a Repetitive Task Is Worth Automating — A Simple Test

A client called me last month asking how much it would cost to automate his quarterly board reports. He hated putting them together. Took him most of a Saturday every three months. He wanted them gone.

I told him not to do it.

He paid me for the conversation anyway, but the answer was the same: this is a bad automation candidate. Not because it can't be automated — almost anything can — but because it shouldn't be. The math doesn't work, and even if it did, the thing he was trying to automate isn't the thing he should be fixing.

This comes up constantly. Small business owners get excited about automation because they're drowning in repetitive work. They hear "AI can do that" or "Zapier can do that" and they want to throw money at the loudest task on their plate. That's the wrong instinct. The loudest task is usually not the most automatable task. And the most automatable task is usually the one you stopped noticing years ago.

Here's the test I run before recommending any automation work. It's four questions. If you can't answer "yes" to at least three of them, don't automate.

Question 1: How often does it happen?

This is the simplest one and the one people get wrong most often. If a task happens four times a year, you almost never get your money back. Even cheap automation has a build cost, a testing cost, and a "what happens when the source system changes" cost. Spread that across four runs a year and the per-occurrence price is brutal.

My rough cutoff: at least weekly, ideally daily. Monthly is a maybe — depends on the other answers. Quarterly is almost always a no.

My client's board report failed here on the first question. Four times a year. Walk away.

Question 2: Does it follow the same pattern every time?

Automation is dumb. I mean that literally — it does what you tell it to do, the same way, every time. If the inputs are clean and predictable, automation flies. If the inputs are messy and the rules keep changing, you'll spend more time fixing the automation than you ever saved running it.

The test: can you write down the steps in a way that someone you've never met could follow without asking questions? If yes, it's automatable. If you find yourself writing "and then use your judgment" anywhere in the steps, you've found the part that probably shouldn't be automated.

Question 3: Does it require human judgment?

This is the question that kills the "AI will just do it" fantasy. AI is getting better at judgment, but most of what small business owners actually want to automate involves trade-offs — which customer to call first, which vendor to push back on, how to frame bad news to a board.

If the task requires you to weigh competing priorities, read a room, or make a call that affects a relationship, leave it alone. Use the time you save automating other things to do that work yourself. That's the work that actually needs you.

Question 4: How much time per occurrence — and how much over a year?

Multiply it out. A task that takes 10 minutes but happens 30 times a week is around 260 hours a year. A task that takes a full day but happens twice a quarter is 64 hours a year. The first one is a slam dunk. The second one barely clears the threshold and probably doesn't survive once you factor in maintenance.

My rough math: under three hours a year saved, walk away. Three to ten hours, only if the task is painful enough that you'll stop doing it altogether if it doesn't get fixed. Over ten hours, this is probably worth a real conversation.

The hidden cost everyone forgets

Automation isn't free to run forever. Source systems change. Vendors change their APIs. A column gets added to a spreadsheet and the whole flow breaks. Someone has to fix that. If you don't have an IT department — and you probably don't — that someone is going to be you, or me, or another consultant.

I tell clients to assume 10 to 20 percent of the original build cost per year in maintenance. Sometimes it's less. Sometimes it's more. Either way, that number has to fit inside the time savings or you've moved the cost around without reducing it.

The third option nobody mentions

The best automation conversation I had last year ended with the client deciding to just stop doing the task entirely. He'd been generating a weekly internal report for six years. Nobody read it. The person who used to read it had left in 2021. Nobody told him.

Before you automate something, ask whether anyone would notice if you stopped doing it. Sometimes the answer is no. That's the cheapest automation in the world.

The example that passed the test

I worked with a wholesale distributor that had been chasing order acknowledgments, proof of deliveries, and ship dates through three-ring binders, email threads, and spreadsheets. They'd been doing it that way since 1989. Their staff was burning hours on it every week.

Run that through the test: happens daily, follows the same pattern, requires no real judgment, and was costing them roughly 15 hours a week across the team. Easy yes. We built a Smartsheet system — no AI, just the right tool for the problem — and they got back 30 percent of the time they'd been losing.

The point isn't that automation is good or bad. It's that most "I should automate this" instincts are wrong, and the people selling automation aren't going to tell you that. I will.

If you want a second opinion on something you're thinking about automating — or buying — [book a free conversation](https://outlook.office.com/owa/calendar/Brincore@brincore.com/bookings/). No pitch. I'll tell you if it's worth it, and if it isn't, I'll tell you that too.