Automation instead of a sticky note on the fridge

How to tell what in your business can be automated, where to start, and why good automation must check that it worked. Based on an on-call schedule.

The sticky note that sometimes falls off

The team I built Dyżur for has one phone number for customers and a schedule of who is on call. Every day someone had to switch call forwarding to the phone of whoever was on duty that day. Usually they remembered. Sometimes they didn’t, and then a customer called someone who had been home for an hour.

Nobody was to blame. A task that has to be done every day, by hand, at a fixed time will sooner or later be skipped. It isn’t a matter of discipline, it’s statistics.

How to tell something can be automated

You don’t need to know programming to judge this. Just check five things.

  • Someone does the same thing regularly: every day, every week or after every order.
  • It can be described in steps without the word “depends”. If the decision needs judgement, automation can at most prepare it, not make it.
  • A mistake costs something: a missed call, a forgotten appointment, an invoice sent twice.
  • Data is retyped from one place to another, for example from a form to a spreadsheet, and from the spreadsheet to an email.
  • The information exists, but nobody gets it on time: a service reminder, an expiring deadline, an empty schedule for next week.

Examples from small businesses

  • An appointment reminder sent the day before, instead of phone calls from reception.
  • A form submission that lands in a spreadsheet or system by itself and immediately gets a confirmation email.
  • A schedule that shows by itself who answers the phone today, who opens up and who goes out on a service call.
  • A weekly summary from a spreadsheet that arrives by email on Monday morning, instead of being put together by hand.
  • A warning when something is missing: an empty shift in the schedule, an unconfirmed booking, a backup that did not run.

Where to start

Before anyone writes a line of code, write down how it works today: who does what, when, where the data comes from, what they do with it and where they enter it. Often this alone shows that some of the steps are unnecessary.

Then count two things: how much time it takes per month, and what one mistake costs. Only those numbers tell you whether automation pays off and how big it can be.

Not every automation needs a programmer. Simple flows, like a form to a spreadsheet plus a confirmation email, can be put together from ready-made tools. A custom program makes sense when you need to control something the ready-made tools don’t support, like call forwarding on a phone, or when something important depends on the automation and it has to be fully under your control.

Automation has to check itself

The most important lesson from Dyżur: the fact that the automation carried out a command doesn’t mean the command worked. The phone dials a forwarding code at the operator, but the operator doesn’t always accept it. That’s why after every change the app asks the network whether forwarding is really on, up to three times, and only then reports that everything is fine.

The second thing is a signal when the automation goes quiet. The battery saver could silently put the app to sleep. So the panel shows when the phone last checked in, and warns when it stops responding. Automation that breaks without a word is worse than a sticky note, because at least you can see the sticky note.

What is left for people

Automation doesn’t replace the team, it takes the remembering off their hands. In Dyżur people still make the schedule, and when someone falls ill, the shift can be changed by hand with one click in the panel. The difference is that nobody has to remember at four in the afternoon to switch something over.

Key takeaway

Automate what is repetitive, describable in steps and costly when it goes wrong. And always add a check that the automation really worked, plus an alarm for when it goes silent.

All articles

Ready to start?

Let’s start with a conversation

Describe what you want to build. The reply contains the scope, the price and the timeline. The inquiry costs nothing and commits you to nothing.