Do Things That Don't Scale (Then Automate Later)

There's a famous piece of startup advice: "Do things that don't scale."

It sounds counterintuitive. Isn't the whole point of building a business to create leverage? To work smarter, not harder? To automate everything?

Yes. Eventually. But not at the start.

The Wisdom of Manual Work

When you do things manually, you learn things that automation can't teach you:

• You discover the edge cases that break your assumptions
• You feel the friction points that need smoothing
• You understand what customers actually want, not what you think they want
• You identify the steps that matter and the ones that don't

Automate too early, and you bake in your mistakes. You scale the wrong process. You build elaborate systems for things that shouldn't exist at all.

The Onboarding Example

Imagine you're setting up a new client onboarding process.

The premature automation approach:

1. Build complex automated email sequences
2. Create elaborate workflows
3. Integrate multiple tools
4. Launch to customers
5. Discover the whole thing is wrong because you didn't understand what customers actually needed

The "do things that don't scale" approach:

1. Onboard each customer manually, one by one
2. Take notes on what questions they ask
3. Notice where they get stuck
4. See which information they actually use
5. After 20 customers, you understand the process deeply
6. Now automate what you've learned works

The manual approach takes longer initially but results in better automation that actually serves customers.

Learning Before Optimising

Every process has hidden complexity. You don't see it until you do it yourself, multiple times, with different customers and situations.

Manual work forces you to engage with that complexity. You can't ignore it. You have to solve it in real time.

That knowledge is invaluable when you finally do automate. You know what matters, what doesn't, what breaks, and what delights.

When to Automate

The right time to automate is when:

1. You've done it enough times to understand the process deeply
2. The process is stable and not changing every week
3. You're confident you're automating the right thing
4. The volume justifies the investment in automation

For most businesses, this means going through a process at least 20-50 times before automating it.

Practical Application

Client Communication

Don't start with complex automated email sequences. Start by personally emailing every client. Notice which messages get responses. Notice what questions they ask. Notice what language resonates.

Then automate the patterns you've discovered.

Lead Qualification

Don't build elaborate lead scoring systems. Start by personally talking to every lead. Learn which ones become customers and why. Understand the signals that matter.

Then automate based on what you've learned.

Content Creation

Don't set up complex content management systems. Start by manually creating and posting content. See what works. Understand your workflow.

Then build systems around what you've proven.

The NotLuck Way

Right now, we go over and over with customers. Every issue. Every pain point. One-on-one attention.

Does that scale? No. Not yet.

But every time we onboard someone, we learn something. We make our processes a little better. We understand our customers a little more deeply.

When we do automate more (and we will), we'll be automating something that actually works.

The Balance

This doesn't mean manual forever. The goal is eventual automation. But automation based on understanding, not assumption.

Think of it like annealing metal: each pass through the process refines it. You work it manually, again and again, until it's strong enough to scale.

The Bottom Line

Resist the urge to automate too early. The efficiency gains aren't worth the learning you miss.

Do things manually until you understand them deeply. Then automate what you've proven works.

The businesses that scale well aren't the ones that automated fastest. They're the ones that understood their processes before they automated them.

Do things that don't scale. Then automate later.