Two minutes
I took a 5pm call this Monday. I do that because of the timezone difference between US and Europe.
It was Jacob, one of my favorite clients.
On the call he asked me if I could automate their subscription reporting. Pull the data, format it, send it weekly to his email. Pretty standard request.
I paused and asked, "How long does it take you to do that now?"
He replied: "I don't know, two minutes?"
Two minutes. Once a week.
I told him the build would take 5 or 6 hours and the API we'd integrate changes every year or so, so we'd be updating it again when that happens. Forever.
I really like the guy. But still, he pushed back on this. He wanted the automation.
So I broke it down to him.
Two minutes a week is 100 minutes a year. The build alone eats three years of time savings before we touch maintenance.
He agreed. We moved on.
A version of me from 2021 would have built it that same afternoon and felt great about it.
Back in 2021, I built an invoicing tool for one of our clients. I spent 40 hours of work on automating this. It saved us maybe an hour a month. I hated that reporting, so automation felt like the obvious solution. I had to update it twice in year one alone. I killed it a year later and asked our accountant to handle it instead.
Those 40 hours should have been billable work for the client. Or time spent with my family. Probably both.
I didn't know what those 40 hours actually cost until later.
Now I do. So when a client asks me to build something with bad math, I tell them. Even when they want it anyway. Especially then.
Automation Checklist
Whenever I or my client want to automate something, I run it through these three questions:
How long does it take to do manually, right now?
Time it. Don't guess. Guesses are always higher.
What breaks the automation over time?
Third-party APIs, edge cases, the person who built it leaving. Also, add the maintenance hours.
Who's paying the build cost?
If it's you, what billable work or family time does it replace? If it's a client, are you charging them more than the automation will ever save them?
Most of what I've automated passed all three. The ones that failed always failed on question 2. And they always broke when I had no time to fix them.
If the math is bad, do the task manually. Two minutes a week is not a problem to solve. The hardest part of automation is saying no.
What's a two minute task you've tried to automate? Reply and tell me. I might feature this story in the next issue.
That's all for this week. See you next Thursday.
P.S. I sent a special link to Jacob that skips all the logins, clicking and redirects him to the subscription report directly. So it's now one click, on demand. No emails needed.