Most founders treat the early growth problem as a systems problem. Set up the funnel. Optimize the conversion. Wait for the numbers.
That is the wrong model entirely.
The most important work in the early days is work that doesn’t scale. And the reason most founders avoid it is that it feels like the wrong kind of work. It feels temporary. Inefficient. Like something you shouldn’t still be doing once you’re really building.
But this is exactly backward. I have watched this pattern repeat enough times to say it plainly.
Recruit them by hand
The most common version of this mistake is how founders think about user acquisition. They want a system. A campaign. An ad that runs while they sleep.
The founders who build something real do the opposite. They find their first users the same way you’d find a collaborator for a project: one at a time, in person, with full attention on that specific person.
I think of Airbnb. They went door to door in New York. Not because they couldn’t imagine running paid ads. Because that was the only way to find out what was actually wrong. Thirty days of showing up at apartments, meeting hosts face to face, helping them improve their listings. That turned a flat line into a trendline.
You don’t learn what you need to learn from a dashboard. You learn it from a person who is trying to use what you built and struggling in ways you didn’t predict.
Make the first users delirious
There’s another thing the unscalable work is for, and it isn’t data.
It’s delight.
When you only have ten users, you can give each of them something that would be impossible to give ten thousand. You can know their names. You can know their problems. You can notice when something breaks for them before they report it.
Wufoo used to send each new user a handwritten card. Absurd at any real scale. But those early users didn’t just stick around. They became the kind of users who tell other people. That’s the entire point.
The goal of the unscalable period is not just to acquire users. It’s to acquire users who believe in what you’re doing so completely that they want to see it succeed. That kind of early user is worth fifty of the passive kind you’ll acquire once you have a real growth engine.
What the unscalable work teaches you
There’s a third reason, and I think it’s the most important one.
You’ll find out exactly what to build.
When Stripe’s founders set up merchants manually, they weren’t being inefficient. They were learning every friction point in the process they would eventually need to eliminate. When you’ve done something by hand fifty times, you know exactly what to automate. If you try to automate it first, you build the wrong thing.
The unscalable phase is not a phase you survive. It’s the phase where you find out whether you have something worth scaling.
What this looks like when you have ten customers
The question is not: how do I build a system to acquire ten thousand?
The question is: what would I have to do to make all ten of those people deeply happy? Not satisfied. Delirious.
Then do that. By hand. For as long as it takes.
The systems come later. The systems only work when you know enough to build the right ones. And you don’t know enough yet.
You learn by recruiting your first users yourself. You learn by fixing their problems personally, before you have a process. You learn by being present for the part of the journey where your startup is fragile enough to break if you look away.
There is no automated version of this. And if there were, it would teach you nothing.
Start with the work that doesn’t scale. That’s where everything else comes from.