When to Build It Yourself

If you're anything like me, you love building solutions to your problems.

But how do you decide when you should do it yourself, or when you should outsource it?

We live in a world where the cost of building has gone to zero. Specialization still exists, but the rules around it have shifted. This is the fundamental question that has led to the saaspocalypse; in short, founders are discovering it's sometimes easier to build it in-house. But other times, they realize it was a terrible mistake.

This newsletter is designed to give you a framework to help you know what you can build in-house, and to help you avoid the disastrous mistake of trying to replicate the wrong service.

It's like deciding to fix your own car vs. going to the mechanic. The constraints at play are:

• Your time vs. their time
• Your expertise vs. their expertise
• Your tools vs. their tools

Time

Time is an interesting one. You wouldn't mind paying someone to do something slower than you, if the price of their man-hours is still less than yours. Think simple video edits, VA tasks, etc.

Expertise

Expertise is the easiest to screw up. A good service provider makes their work appear effortless. This might lead you to believe replication is easy. But the devil is in the details, and small gaps in skill compound into massive gaps in outcome.

"If you're operating with 1,000 times leverage and somebody is right 80% of the time, and somebody else is right 90% of time, the person who's right 90% of the time will literally get paid hundreds of times more by the market because of the leverage and because of the compounding factors and being correct."

Naval Ravikant

Tools

The last constraint of replication is tool availability. For physical services, this is immediately apparent: think nail techs, mechanics, doctors. They can't exactly do their job without their physical equipment.

How about the knowledge economy? Tool availability gets replaced by tool selection. A founder building blindly will accidentally use a server-side-rendered framework for his landing page when a static site would have shipped in an afternoon.

Plus every service provider now has his own AI workflows that can't be replicated by merely looking at his output. The visible deliverable is the tip of an iceberg of prompts, n8n workflows, and Claude-Coded dashboards. And these same constraints apply when he decides to build his own tools vs. buy someone else's.

This is how I see the dynamics of a world where building is nearly free. The decision rule I keep coming back to:

Only build it in-house when you can match the service provider on at least two of the three constraints. Miss on two, and you'll spend more time and money than you would have just paying them.

Hopefully I've given you some clarity in your decisions.