Process before automation

Automate a mess and you get a faster mess. The tool does not fix the work. It just runs whatever you give it, at speed, including the broken bits.

This is the most expensive mistake I see owners make, and it is expensive precisely because it feels like progress. The onboarding is chaotic, so you automate the onboarding. Now the chaos arrives in the client’s inbox on a schedule, looking official. You have not solved anything. You have industrialised the problem and put a logo on it.

Here is why it happens. A broken process is uncomfortable to look at. Sitting down to map how a job actually runs, with all its exceptions and workarounds, is dull and slightly embarrassing, because you find out how held-together things really are. Buying a tool feels better. It is action, it has a price tag, it looks like you are doing something modern. So owners reach for automation as a way to avoid the boring work of fixing the process first.

I worked with a firm that wanted to automate their quoting. We started to map it and found there was no process. There were four people quoting four different ways, three spreadsheets, and a pricing rule that lived in the founder’s head. If we had automated that, we would have automated the confusion and made it permanent. The win was not the tool. The win was agreeing one way to quote. The automation took an afternoon after that. The agreement took two weeks and was the whole job.

The sharper way to think about it is that automation is an amplifier. It makes whatever you point it at louder and faster. Point it at a clean, simple process and you get speed and consistency. Point it at a mess and you get more mess, more often, with less chance of catching it because now it runs on its own.

There is a useful test in this. If you cannot describe the process on one page so that a new starter could follow it, it is not ready to automate. Not because the tool cannot handle complexity, but because you do not yet understand the job well enough to hand it over safely.

The practical takeaway. Before you automate anything, run it by hand and write down every step, including the bits where someone “just knows” what to do. Cut the steps that exist for no reason. Fix the ones that are broken. Get it down to something clean and repeatable. Then, and only then, automate it. You will spend more time on the process than the tool, and that is exactly the right way round.