There’s a moment every founder hits, usually around user number 12, where a voice in their head says: “This won’t work when we have 10,000 users.” They’re manually onboarding people. They’re sending personal welcome emails. They’re hopping on calls to walk someone through a feature that should be self-explanatory but isn’t yet.
And so they stop. They build an automated onboarding flow. They record a Loom video. They set up a drip campaign. They “scale” themselves out of the picture.
This is one of the most common mistakes in early-stage startups. Not because automation is bad (it’s essential, eventually) but because you skipped the part where you actually learn what your users need.
The Unscalable Advantage
Paul Graham coined the phrase “do things that don’t scale” back in 2013, and it’s become startup gospel. But most founders treat it like a temporary inconvenience, something you endure until you can afford not to. They miss the deeper point.
Unscalable work isn’t a phase you push through. It’s a competitive advantage you’ll never have again.
When you personally onboard someone, you hear the questions they’re too polite to put in a feedback form. When you manually deliver your service, you discover the gap between what you built and what they actually needed. When you text a user to ask how their first week went, you get the kind of candid response that no NPS survey will ever produce.
Stripe’s founders famously did “Collison installations.” Instead of asking interested developers to sign up online, they would say “let me set it up for you right now” and install Stripe on the spot. It wasn’t efficient. It was brilliant. Every installation was a conversation. Every conversation was product research.
You can’t get that from a dashboard.
Why Founders Resist It
If unscalable work is so valuable, why do founders rush past it? Three reasons.
First, it feels unproductive. We’re wired to measure progress in output: features shipped, campaigns launched, systems built. Spending 45 minutes on a call with one user doesn’t feel like building a company. It feels like customer support.
Second, investors ask about scale. “How does this work at 100x?” is a question you’ll hear in every pitch meeting. So founders pre-optimize, building for a scale they haven’t earned yet. They’re solving tomorrow’s problems while today’s users quietly churn.
Third, it’s emotionally exposing. When you’re personally delivering the experience, you can’t hide behind the product. If it’s not working, you see it on someone’s face in real time. That’s uncomfortable. An automated flow lets you stay comfortably distant from the truth.
A Framework for Your First 100
Here’s how to think about the unscalable phase, not as busywork, but as a structured learning engine.
Users 1-10: Be the product.
At this stage, you might not even need software. Deliver the outcome your product promises, but do it manually. If you’re building a tool that helps founders write investor updates, write the updates for them. If you’re building a scheduling tool, coordinate their calendar by hand. You’ll learn more about the problem in two weeks than you would in two months of building.
The goal here isn’t revenue. It’s pattern recognition. What do users 1 through 10 have in common? Where do they get stuck? What do they ask for that you didn’t anticipate?
Users 11-50: Build the bridge.
Now you start building, but only the parts that your manual experience proved matter. You’ve seen the workflow ten times. You know where the friction is. Build just enough product to handle the most painful manual step, and keep doing the rest by hand.
This is also where you test your messaging. You’ve been explaining your product in conversations. Which descriptions made people lean in? Which ones got blank stares? The language that worked in person is the language that should be on your landing page.
Users 51-100: Systematize the magic.
By now you’ve got real patterns. You know which onboarding steps actually matter and which ones you can cut. You know the “aha moment,” the point where users go from trying it to relying on it. Now you can start automating with confidence, because you’re encoding something you’ve observed dozens of times, not guessing.
The key discipline: every time you automate something, you lose a feedback channel. Be deliberate about which channels you close and which you keep open.
The Data Backs This Up
This isn’t just philosophical. There’s real evidence that high-touch early engagement predicts long-term success. A First Round Capital survey of their portfolio found that founders who personally spoke with their first 50 customers had significantly higher retention rates at the 12-month mark than those who relied on self-serve from the start.
It makes sense. Users who feel heard become advocates. They forgive bugs. They send feature requests instead of cancellation emails. They tell their friends not just “try this product” but “the founder is incredible, they actually listen.”
That kind of loyalty doesn’t come from a well-designed drip sequence.
The Hardest Part Is Knowing When to Stop
There’s a real risk of staying in unscalable mode too long. Some founders use high-touch engagement as a way to avoid building the hard parts of the product. If you’re still manually onboarding user number 300, something’s wrong.
The signal to watch: when you can predict what a new user will need before they tell you. When the onboarding call becomes repetitive. When you’re saying the same things and solving the same problems. That’s when you’ve learned enough to automate well.
Until then, stay close. Stay manual. Stay uncomfortable.
The founders who win aren’t the ones who scale fastest. They’re the ones who learn fastest. And learning happens at the speed of conversation, not the speed of code.
Your first 100 users aren’t a growth metric. They’re your unfair advantage, but only if you’re actually in the room with them.
Inpaceline OS was built from exactly this kind of founder-to-user learning, designed around the real workflows that early-stage founders actually use, not the ones that look clean on a whiteboard. If you want a system that grows with you from user 1 to user 10,000, check it out at www.inpaceline.com.
