What Paul Graham Really Meant by “Do Things That Don’t Scale” — And What He Didn’t Mean

Paul Graham’s golden rule for startups: Do things that don’t scale. 🤔 Confused? Don’t be—here’s why your unscalable efforts today lay the foundation for massive growth tomorrow.

Frederik van Lierde - Nov 15, 2024

What Paul Graham Really Meant by “Do Things That Don’t Scale” — And What He Didn’t Mean

Articles

Summary

Building early relationships with users helps you understand their real needs before scaling.
Doing unscalable tasks isn’t a permanent strategy but a foundation for sustainable growth.
Unscalable efforts provide insights that data alone can’t reveal, making your product genuinely valuable.

Startups get advice from all directions, especially in our tech-centered world. But there’s one piece of wisdom that’s practically legendary, thanks to Paul Graham from Y Combinator: “Do things that don’t scale.”

It sounds strange at first. Why would a big name in tech tell founders to do things that don’t immediately set them up for explosive growth?

In a world focused on fast growth and huge scaling, what’s the point of doing things that don’t scale? To get to the heart of this, let’s look closer at what Graham was really saying — and, just as important, what he wasn’t.

What Does “Doing Things That Don’t Scale” Really Mean?
What Graham Didn’t Mean by This Advice
Real-World Examples of “Doing Things That Don’t Scale”
Linking It to “Fail Fast” Culture

What Does “Doing Things That Don’t Scale” Really Mean?

When Paul Graham said to “do things that don’t scale,” he was addressing early-stage founders directly. Startups are often so focused on rapid growth and scalability that they miss out on the very activities that can lay the foundation for long-term success.

Graham’s advice is simple but profound: in the early days, it’s okay, even essential, to give personal attention to your first users. You need to create something people really want, and to do that, you may need to go above and beyond in ways that don’t fit neatly into an automated, scalable process.

Think of Airbnb’s founders. In their early days, they spent time going door-to-door in New York City, taking professional-quality photos of rental listings.

This personal approach wasn’t scalable, but it helped them better understand their users and added an irreplaceable quality to their listings.

It’s the type of effort that yields insights and relationships. By listening to users and solving problems firsthand, Airbnb’s founders could build a product that met actual needs.

Why It Works:
Doing things that don’t scale forces you to get close to your customers and understand their pain points and preferences. It enables you to create strong user relationships and discover insights that are often invisible in data alone.

Once you truly understand your customers, you’re in a better position to create scalable solutions that genuinely resonate.

What Graham Didn’t Mean by This Advice

Now, let’s clear up some common misunderstandings. Paul Graham wasn’t suggesting you get stuck in unscalable tasks forever. He’s not advocating for founders to take on every task manually without ever figuring out a way to automate or scale.

Instead, it’s about knowing when to let go. Doing things that don’t scale is a strategic phase, not a permanent state. When you’re just starting, these small, unscalable acts can set you apart and allow you to understand your market better. But as you grow, they need to become processes and structures that others can take over or that can be automated.

Consider the example of Stripe, the online payment processor. In its early days, Stripe’s founders would manually integrate their payment software for new users.

They could have spent time building complex self-serve systems immediately, but by doing these integrations manually, they gained a deep understanding of their clients’ needs and pain points.

Later, they scaled their model to streamline the process, but only after learning everything they could from these unscalable efforts.

The Key Distinction:
Doing unscalable things is a phase that sets the stage for scalability. Graham is not advising founders to take on never-ending, unscalable work but to do these tasks temporarily to build a strong foundation.

At some point, founders need to systematize or hand over parts of the process. The goal is to balance personalized attention with scalability as the business grows.

Real-World Examples of “Doing Things That Don’t Scale”

Instacart’s Door-to-Door Deliveries:
Early on, Instacart’s founder, Apoorva Mehta, would personally deliver groceries to his customers. This one-on-one interaction helped him understand customer expectations, gather immediate feedback, and iterate on the product.

Today, Instacart has a streamlined, scalable delivery process, but the lessons Mehta learned in those early days were invaluable.
Amazon’s Personal Customer Emails:
In Amazon’s early years, Jeff Bezos was known to personally respond to customer complaints and even reached out when issues arose.

He used these interactions to refine Amazon’s customer experience and learn what mattered most to customers.

That level of personal touch wasn’t sustainable, but it gave him a deeper understanding of customer expectations, fueling Amazon’s legendary customer service.
Dropbox’s Initial Manual Onboarding:
Before Dropbox built the sophisticated onboarding processes it has today, founders would manually guide new users through the setup, helping them see the value of cloud storage.

This close interaction gave them insights into usability challenges and the support needed, which they later used to build scalable onboarding processes.

These founders didn’t waste time building complex systems before knowing what their users truly needed. Their unscalable acts taught them crucial lessons that they eventually built into scalable systems.

Linking It to “Fail Fast” Culture

At first glance, Graham’s advice to “do things that don’t scale” might seem at odds with the popular “fail fast” mantra, but it’s actually quite complementary.

While “fail fast” encourages startups to quickly test, learn, and adapt, doing unscalable things helps founders learn what truly matters to users before scaling.

Once they understand their audience through these personal efforts, they can test and scale in ways that have a higher chance of success.

As we discussed in the “Fail Fast Mantra Debunked”, “fail fast” isn’t a license for reckless experimentation. Like Graham’s advice, it emphasizes smart, meaningful action — experimentation that genuinely serves your users and informs your strategy.