I Built the Automation Before I Had Enough Customers

I made a mistake while building my digital product business.

I built the automation before I had enough customers to need it.

At first, this felt like the right thing to do.

I was using Payhip to sell digital products.

Make was handling the workflows.

Telegram was useful for notifications.

Notion was where I wanted to keep track of orders and customers.

Later, I added email automation as well.

The architecture started looking pretty nice.

Payhip
↓
Make
↓
Notion
↓
Telegram
↓
Email follow-up

Everything had a place.

And then I realized something.

There weren't that many orders.

So why was I spending so much time maintaining the system?

The problem wasn't the automation

The workflows themselves were not the problem.

Most of them worked.

The problem was that I was building for a future version of the business.

I kept thinking:

"What if I get more orders?"

"What if I need another notification?"

"What if I want to track another field?"

"What if I add another product?"

Those are reasonable questions.

But when you're still trying to get people to actually find and buy your product, they can become distractions.

I could spend an hour improving an automation that might save me ten minutes a week.

That doesn't make much sense.

I started looking at automation differently

Before building a new workflow now, I try to ask a few basic questions.

Does this happen often?

Is the process predictable?

Am I doing it manually often enough that it annoys me?

What happens if the automation fails?

And probably the most useful question:

Would I build this if my business stayed small for the next few months?

That last question has stopped me from building quite a few things.

A smaller workflow is often easier

For example, digital product delivery is a pretty good automation candidate.

A customer buys something.

The same type of event happens again.

The delivery process is predictable.

There isn't much value in manually checking every order.

So automating that makes sense.

But something that happens once every few weeks?

I might just do it manually.

It's not worth creating another workflow, another webhook, another database field and another thing to debug.

I also learned something about maintenance

This was the part I underestimated.

When you connect several services together, you don't just create an automation.

You create dependencies.

A webhook can change.

A field can be missing.

An API can return something unexpected.

A workflow can run twice.

A database can have a different structure than you expected.

None of this is particularly scary.

But every additional workflow adds another thing you eventually have to understand.

That's why I'm becoming much more conservative about adding automation.

I'm not trying to build the biggest workflow anymore.

I'm trying to build the smallest one that solves the problem.

What I'm doing now

My current approach is pretty simple.

Build one workflow.

Use it.

See if it actually saves time.

Leave it alone for a while.

Only then decide whether another automation is necessary.

This is also changing how I'm building NomadOS.

Instead of asking:

"What else can I automate?"

I'm trying to ask:

"What repetitive problem am I actually having right now?"

That question is much more useful.

I'm still figuring this out, so I'm curious about other people's experience.

Have you ever built an automation that turned out to be more work than the manual process it replaced?