What's actually worth automating first? A 15-minute audit

The most expensive automation is the one you built for the wrong task.

I've watched owners spend weekends wiring up elaborate systems for things that barely mattered, while the actual leak kept draining money quietly in the background: the follow-up that never gets sent, the quote that sits for four days.

So before any building, I run the same short audit. You can do it yourself in about fifteen minutes. You need a piece of paper and an honest look at last week.

Step 1: List the repeaters (5 minutes)

Write down every task you or your team did more than three times last week. Not the big projects. The repeaters:

• Answering the same customer questions
• Copying information from one place to another
• Sending follow-ups and reminders
• Assembling reports, invoices, or quotes
• Routing things: emails to the right person, leads to the right list

Don't filter yet. Ten to twenty items is normal.

Step 2: Score each one two ways (5 minutes)

For each task, two quick scores from 1–3:

Time: how much does it eat weekly? 1 = minutes · 2 = an hour or two · 3 = half a day or more

Money: what happens when it slips? 1 = mild annoyance · 2 = someone waits, quality dips · 3 = revenue is lost: a lead goes cold, an invoice ages, a customer churns

That second score is the one most people underrate. A task can be small on time and huge on money: a five-minute follow-up that doesn't happen is a deal that doesn't happen.

Step 3: Read the corners (5 minutes)

Plot them mentally on a 2×2:

	Low money impact	High money impact
Low time	Ignore for now	Automate first ←
High time	Batch it or template it	Automate next

The winner is almost always in the top-right: small, boring tasks where slipping costs revenue. Lead follow-ups. Quote reminders. Review requests. Invoice chasing. They're unglamorous, cheap to automate, and they pay for themselves in weeks, because the machine's superpower isn't speed. It's never forgetting.

The classic mistake is starting bottom-left or bottom-right: automating the thing that's merely irritating, or the big messy process that should be simplified before it's ever automated. A bad process, automated, is just bad faster.

What "automated" should actually mean

One more thing, because it matters: an automation you can't see into is a liability with a subscription fee. Whatever gets built for you should:

• Run inside the tools you already use. No rip-and-replace.
• Fail loudly. You hear about problems before your customer does.
• Come with plain-English documentation, so your team can run it, adjust it, and switch it off without calling anyone.

That's the standard I build to, and it's what the audit-first approach protects: we find where the hours leak and the revenue stalls before anything gets built, so the first system is the one that pays.

Run the fifteen minutes. If your top-right corner has something in it (and it almost always does), that's your first automation.