{"id":12485,"date":"2026-07-10T10:00:00","date_gmt":"2026-07-10T10:00:00","guid":{"rendered":"https:\/\/meriaky.com\/blog\/automate-a-process\/"},"modified":"2026-07-25T07:42:47","modified_gmt":"2026-07-25T07:42:47","slug":"automate-a-process","status":"publish","type":"post","link":"https:\/\/meriaky.com\/en\/blog\/automate-a-process\/","title":{"rendered":"Automate a Process: The #1 Mistake to Avoid"},"content":{"rendered":"\n

Before you automate a process, don’t automate it yet.<\/strong> I know, that sounds backwards. But it’s the first thing I tell a client who shows up wanting to automate everything right now. Mistake number one, the one I see nine times out of ten, is jumping straight to automation. And automating a mess just gives you a mess that moves faster.<\/p>\n\n\n\n

A client came to me all fired up the other day. “Edouard, I need to automate this, it’s urgent, it’s eating my whole day.” My first answer disappointed him a little: “Hold on. We’re not automating that yet.”<\/p>\n\n\n\n

The guy comes to me to<\/em> automate, and I tell him to stop. Because if I let him run with it as-is, I’d be doing him a disservice.<\/p>\n\n\n\n

Why you shouldn’t automate a process too early<\/strong><\/h2>\n\n\n\n

When someone comes to me to automate a process, nine times out of ten, one of three things is true.<\/p>\n\n\n\n

First, the process isn’t documented. Nobody could run it without the right person in the room.<\/p>\n\n\n\n

Second, the process has ten exceptions that make no sense. “Oh no, for that client we do it differently.” Why? “Uh… good question.”<\/p>\n\n\n\n

Third, and worst of all: it all lives in the founder’s head. And it depends on the day, his mood, whether he slept well.<\/p>\n\n\n\n

So if you automate that as-is, you’re setting the mess in concrete. You’re automating fuzziness. And automated fuzziness is still fuzziness, just harder to fix.<\/p>\n\n\n\n

That’s why there’s an order to follow. Automation is the last step, not the first. So here are the four that come before it.<\/p>\n\n\n\n

1. Question every step<\/h3>\n\n\n\n

First, you ask why each step exists.<\/p>\n\n\n\n

And here’s the surprise. Half the steps are there “because that’s how it’s always been done.” Nobody remembers why. In other words, it’s a reflex, not a decision.<\/p>\n\n\n\n

So if you can’t explain what a step is for, that’s already a red flag.<\/p>\n\n\n\n

2. Cut the waste, not the value<\/h3>\n\n\n\n

Next, you remove what serves no purpose. Duplicate approvals, ghost steps, exceptions that exist for a client who left two years ago.<\/p>\n\n\n\n

A quick warning here, because it matters. Cutting doesn’t mean cutting corners.<\/p>\n\n\n\n

Removing a useless step is optimization. Removing a step that protects your client or delivers your value is sabotage. That’s exactly the move of the salesperson in a hurry who promises faster, lighter, and ends up saying “oh, sorry about that” when it breaks.<\/p>\n\n\n\n

Bottom line: cut the fat. Never touch the muscle. It’s the same principle behind lean manufacturing<\/a>: eliminate what doesn’t add value, keep everything that does.<\/p>\n\n\n\n

3. Simplify and document what’s left<\/h3>\n\n\n\n
\"Hands<\/figure>\n\n\n\n

Whatever survives, you make crystal clear. One way to do it. As few exceptions as possible. And above all, you write it down.<\/p>\n\n\n\n

Because if the process only lives in your head, it doesn’t really live anywhere. Documenting it is what lets you get the process out of you. It’s the difference between a business and a one-person show held together by luck.<\/p>\n\n\n\n

4. Speed it up by hand before you automate a process<\/h3>\n\n\n\n

Then, before you even think about tools, you check whether you can do it faster by hand.<\/p>\n\n\n\n

Often, a template, a shortcut, or a simple checklist already gets you 80% of the gain. Without a single line of code.<\/p>\n\n\n\n

5. Only then can you automate a process<\/h3>\n\n\n\n

Finally, once the process has been questioned, trimmed, documented and sped up… that’s when it gets easy.<\/p>\n\n\n\n

Because now you’re automating something clear. Something clean. Something built to last, not just to work today.<\/p>\n\n\n\n

Because you’re automating something clear. Something clean. Something built to last, not just to work today.<\/p>\n\n\n\n

My real job: refusing to start too early<\/h2>\n\n\n\n

Actually, that’s exactly where my job lives. My job isn’t to sell you step 5 as fast as possible. It’s the opposite: to refuse to start until the first four are done.<\/p>\n\n\n\n

Because if we jump straight to automation, we automate your fuzziness, and I become just another guy who sold you a system that doesn’t hold up.<\/p>\n\n\n\n

So I’d rather cost you a week clarifying your process than save you a month on a system built on sand. That’s a hard thing to hear when you’re in a rush. But it’s exactly what separates a system still running on its own two years from now from a gadget you abandoned in three months.<\/p>\n\n\n\n

So if you want to see which steps you need to question or trim before you even talk about automation, start with the diagnostic<\/strong><\/a>. You’ll know where to begin, in the right order.<\/p>\n\n\n\n

Frequently asked questions<\/h2>\n\n\n\n
Do you need to document a process before automating it?<\/strong><\/strong>

Yes, it’s a non-negotiable step. A process that only lives in the founder’s head can’t be automated properly. Documenting it means getting it out of your head so it can run without you.<\/p> <\/div>

Why is automating a process too early a mistake?<\/strong><\/strong>

Because you set your own mess in concrete. If the process is fuzzy, full of exceptions and undocumented, automating it just makes that fuzziness faster and harder to fix. Automation is the last step, not the first.<\/p> <\/div>

Where do I start if I want to automate a process?<\/strong><\/strong>

Start by questioning every step, cutting the waste, simplifying and documenting what’s left, then speeding it up by hand. Once the process is clear and clean, automating it becomes simple and durable.<\/p> <\/div> <\/div>\n\n\n\n