Make it work first, fix it later

Yesterday I launched something from a blank file. Sixteen pages, a contact form, DNS pointed, SSL green, live on the internet by end of day. No beta. No soft launch. No waiting until it felt finished.

It is not perfect. The copy could be sharper. A few layout decisions I'd change with another hour. The gallery is thinner than I want it to be.

I shipped it anyway.

There's a sequence Kent Beck laid out decades ago for writing code, but it applies to everything a founder touches: make it work, make it right, make it fast — in that order. The reason most people invert this, or try to do all three at once, is fear. If it's not right, someone will notice. If it's not fast, someone will leave. If it's not finished, someone will judge.

But nothing can be made right until it exists. You cannot optimize a thing that isn't there yet.

The cost of the wrong sequence

I've watched founders — and I've been this founder — spend six weeks perfecting a homepage that zero people have seen. Every revision is done in a vacuum. There's no feedback loop, no real user, no actual signal. You're just moving furniture in an empty house.

The sequence breaks down like this. Working means: does it do the thing. Right means: does it do the thing well. Fast means: does it do the thing at scale. Skipping to right or fast before working is not ambition — it's avoidance dressed up as standards.

Yesterday I also spent time cleaning up a second project that has real users. Not adding features. Not redesigning. Cleaning up the parts that were built fast the first time, now that I know what the users actually need. That's what the middle step looks like. You earn the right to make it right by first making it work and putting it in front of real people.

The recursive thing about running a business alone is that you have to resist perfectionism twice: once before launch, once after. Before launch, perfectionism is procrastination. After launch, perfectionism is distraction from the next thing that needs to merely work first.

The bar for shipped is lower than you think. The bar for done is higher than you'll ever reach anyway.

So make it work. Put it out. Then go back for right.