Automating a bad process only makes the mistake faster

There is a strong temptation, once you discover an automation tool, to point it at the heaviest process in the company and automate it exactly as it stands. It is fast, it is visible, and it is almost always waste.

First, eliminate

The question that comes before "how do we automate this?" is "why does this exist?". In nearly every SMB we walk into there are steps that survive out of inertia: a report nobody has read in two years, a double check created after one mistake in 2019, a manual backup of data that is already backed up automatically.

Automating those steps means paying to keep them forever.

Then, simplify

A process with seven approvals does not get better by having seven automatic approvals. It gets better when someone decides three are enough. Technology does not fix design problems — it amplifies them.

Only then, automate

Once the process is stripped to the essentials, automation becomes simple and cheap. More importantly, it becomes stable: fewer steps means fewer failure points and less maintenance over time.

A practical test

Draw the current process on a whiteboard with every step. Then mark each step with one of three letters: E (eliminate), S (simplify) or A (automate). If more than half the steps get an A on the first pass, you probably have not looked critically enough.