Writing

What to automate first in a small business

Most small businesses automate the wrong thing first. They automate what is annoying: the task that interrupts the day, the one everyone complains about. It feels like progress because the irritation goes away. But the irritating task and the expensive task are rarely the same thing, and only one of them is worth building for.

Automate what is expensive, not what is loud

The task worth automating first is the one that is both repetitive and expensive in time, and that gates everything downstream. It is usually quiet. You stopped noticing it because you do it on autopilot, several times a day, and the whole pipeline waits on it.

A 30-second way to find it

For one week, every time you catch yourself doing something that feels repetitive, write down two numbers: how long it took, and how many times you did it that day. At the end of the week, multiply minutes by frequency. The biggest number is your first leverage point. It will almost never be the task you would have guessed.

A task that takes 4 minutes but happens 20 times a day is 80 minutes a day, more than 300 hours a year. The 45-minute task you hate but do once a week is 39 hours a year. Annoyance is a bad ranking system. Math is a good one.

Three quick tests before you build

Once you have a candidate, run it through three questions:

1. Is it rules-based? If a clear set of steps gets you from input to output most of the time, a machine can do it. If every case needs fresh human judgment, automate the gathering and let the human decide.

2. Is the input already digital, or can it be? Email, PDFs, spreadsheets, form entries: all workable. If the input lives only in someone's head, fix the intake first.

3. Does it gate other work? The best automations remove a bottleneck that other steps are waiting on, so the whole line speeds up, not just one station.

Build one thing, all the way

The mistake after picking the right task is trying to automate five things halfway. One automation, built end to end and actually shipped, beats a folder of half-finished ones every time. Pick the single biggest number from your week of tracking, build that, let it run, then find the next one. Leverage compounds when you finish.