Doing things that don't scale to win your first users

Paul Graham wrote "Do Things that Don't Scale" in 2013. It remains one of the most cited pieces of Y Combinator advice. Its thesis is simple. Startups take off because founders make them take off. The first users almost always come from manual, unglamorous effort.

Recruit users by hand

Graham calls manual recruiting the most common unscalable thing founders must do. He uses Stripe as the example. When someone agreed to try the product, the Collison brothers did not send a link. They asked for the person's laptop and set them up on the spot. YC calls this the "Collison installation."

Graham names two reasons founders avoid this work. They would rather write code than face rejection, and the early numbers look too small to matter. He adds that at least one founder, usually the CEO, has to spend a lot of time on sales and marketing.

Weekly growth rate and the compound math

YC asks founders to measure progress by weekly growth rate. Graham gives the arithmetic. A company with 100 users that grows 10% a week has about 14,000 users after a year and about 2 million after two years. The small absolute numbers at the start are the base for that curve.

Airbnb, Wufoo and the value of delight

Airbnb's founders went door to door in New York, recruiting hosts and improving their listings. Graham writes that about 30 days of in-person work made the difference between success and failure. Wufoo sent each new user a handwritten thank-you note for as long as it could.

Graham's argument is that early users should feel that signing up was one of the best decisions they made. A small company can offer a level of attention no large company can. He states that he has never seen a startup misled by trying too hard to make its first users happy.

Start with a narrow market

Facebook began only at Harvard. That limited the market to a few thousand people, but it let the product reach critical mass. Graham suggests founders look for a subset of the market where they can win users quickly. Among businesses, he notes that other startups are usually the best early adopters.

Consult for one customer, then automate

For B2B startups, Graham recommends picking a single user and building as if you were their consultant. Keep adjusting until the product fits that user. Other users with similar needs usually follow. He also describes doing the work by hand behind a software front end. Stripe's first "instant" merchant accounts were set up manually by the founders behind the scenes.

The advantage is learning. When you automate yourself out of the loop later, you know exactly what to build.

Big launches and partnerships rarely work

Graham warns against the big launch and against early partnerships with large companies. He argues that how a startup is doing a few months after launch depends on how happy it made its first users, not on how many signed up on day one. Partnerships, in his experience, take more work than founders expect and return little.

Founder takeaways

• Book the meetings yourself. Recruit your first 20 users one at a time, in person or on a call.
• Set a weekly growth target. Track it every Monday and report it to your team.
• Write down every manual step. The list becomes your automation roadmap.
• Pick one beachhead. Choose a segment small enough to dominate in a quarter.
• Skip the launch event. Spend the budget on direct outreach instead.

AI tools make it easier to build a product quickly. They do not remove the need to find and serve the first customers by hand. Graham's advice applies with more force when the product itself is cheap to make.

Sources

1. Do Things that Don't Scale
   Paul Graham (Y Combinator co-founder) · Jul 2013