Which processes to pick first
- The ones that repeat: every day or every week, not once a year.
- The ones with clear rules: if this happens, do that. The fewer judgement calls, the better.
- The ones that copy data: from an email to the ERP, from one spreadsheet to another, from a form to the CRM.
- The ones that fail because someone forgets: payment reminders, renewals, deadlines, stock running out.
Make a list with the hours you spend on each one every week. The first one to automate is the one that saves the most hours with the least risk, not the most eye-catching one.
Map it out before you automate it
Sit down with the person who does the work and write down every step: where the information comes from, where it is written, who decides what, and what happens when something does not add up. Almost always, steps turn up that no longer make sense.
Automating a messy process only makes it go faster in the wrong direction. First you put it in order; then you automate it.
Step by step: how to automate a process
- Pick the process and write down how many hours a week go into it.
- Map it as it is done today, with the person who does it.
- Remove the steps that no longer make sense and fix the rules: what comes in, what is done with it and what comes out.
- Decide where a person has to look: the doubtful cases, the large amounts, anything that goes to the customer.
- Choose the tool: a feature of the program you already have, an automation tool or a custom integration.
- Run it for a few days alongside the usual work, with real cases.
- Add an alert for when it fails and measure the hours it saves.
Three ways to automate
- With what you already have. Many ERPs, CRMs and online shops come with rules, alerts and connections that nobody has switched on. It is the first place to look.
- With an automation tool. Platforms that connect applications with building blocks work well for simple flows between well-known programs, with low volume.
- With a custom integration. When there is volume, there are many exceptions, the data cannot fail or a program has no connector, an integration built for your case is more reliable and easier to maintain.
The mistakes that cost dearly
- Automating without anyone noticing when it fails: every automation has to send an alert when something does not go through.
- Depending on a single person: if only they know how it works, the day they are away, it stops.
- Chaining too many tools together: when a piece of data gets lost, nobody knows in which one.
- Not measuring it: without knowing how many hours it took before, you cannot tell whether it was worth it.