Taking inventory: finding what actually repeats

Automating routine (no-code + AI) · Lesson 1 / 23

Until routine is written down, it may as well not exist

People almost never remember what they spend their day on. Ask an operations manager where the time goes and you will hear two or three tasks out of ten. Everything else — small data transfers, forwarding files, copying rows into a spreadsheet by hand — slips past memory because each operation takes ninety seconds. Automating on a hunch is pointless: you will fix what is visible rather than what is expensive.

How to build the list in a week

Keep a log, not a recollection. For five days straight, write down every action that repeated at least twice: what came in, what went out, where you put it.
Record frequency, not duration. You will estimate duration badly and frequency almost exactly. Ten times a day at one minute hurts more than once a month at an hour.
Mark the context switches. Every jump between the mail window, the spreadsheet and the messenger costs more than the operation itself.
Flag interruptions. A task that lands unannounced breaks the rest of your work and costs double.

A workable log format

date  | operation                    | in       | out  | per week | who else
12.04 | copy request from mail to sheet | email  | row  | 40       | Anna
12.04 | assemble the sales report       | 3 files| file | 1        | only me
12.04 | chase a client about payment    | debtors| email| 12       | Anna, Ian

After a week the picture shifts. Usually it turns out that the most visible task — the quarterly report — is not worth automating at all, while the invisible "copy the request over" eats six hours a week across two people. Start with what is frequent and dull, not with what is heavy and rare.

Insight. A task that several people do in slightly different ways is the richest candidate. Automating it also forces everyone onto one method.

Common mistake. The list gets written from memory in ten minutes, and then a month goes into automating the wrong thing. A week of observation is cheaper than a month of wasted work.

Pro tip. Add a column called what happens if nobody does this. Half the rows will turn out to be rituals that nobody actually needs.

Cheat sheet

A week of logging beats a month of reasoning.
Count frequency, not the feeling of heaviness.
One task done by several people is a priority candidate.
The what-if-nobody-does-it column filters out rituals.

1. Why can't the routine inventory be built from memory?

2. Which do people estimate more accurately, frequency or duration?

3. Which task in the log is the strongest automation candidate?

🔒 Answer the question correctly to move on to the next lesson.