The quickest way to waste money on automation is to automate a process nobody has written down.
What you get is the same mess, faster, and harder to change.
The job as it really happens
Every process has two versions: the one in the procedures document, and the one people actually do. The second one is the one that matters, and the only way to find it is to sit with the person doing the work and watch.
Write down every step, including the ones that feel too small to mention:
- Where does the information come from?
- Who touches it, and what do they do to it?
- Where does it wait, and for whom?
- What happens when something's missing or wrong?
- Where does it end up?
Look for the waits
When it's written down, the interesting bits jump out. Usually it's not the steps that cost the time, it's the gaps between them: the email sitting in an inbox, the form waiting for a signature, the report waiting for someone to get back from lunch.
Split human from machine
Go through the list and mark each step:
- Machine: copying, checking against a rule, moving information from one place to another, sending a reminder.
- Human: a judgement call, a conversation, anything with an exception you can't write down yet.
The machine steps are what you automate. The human steps are where you put the time you get back.
Then build the smallest version
Automate the machine steps for one job, with a person checking the output for the first few weeks. Measure it against the before. Only then do the next one.
It's slower to start than buying a platform and switching it on. It's much faster to finish.