Paul Graham · 2013 · 8 min
Do Things That Don't Scale
The most common cause of startup failure is founders refusing to do unscalable manual work in the early days.
Key Takeaways
- 01Recruit users one by one, manually. Stripe's 'Collison installation' meant the founders literally typed in your API keys for you.
- 02Make every early user feel like they got a custom experience. Airbnb founders flew to NYC to take photos for hosts.
- 03Founders refuse to do this because it feels embarrassing. That embarrassment is exactly why it's a moat.
- 04Most startups don't have a growth problem. They have a 'no one wants this yet' problem disguised as a growth problem.
Distilled
PG's most important essay for first-time founders. The thesis: startups don't take off because of some virality switch — they take off because the founders pushed them, manually, at every step, until the thing built enough momentum to roll on its own.
The corollary nobody wants to hear: if you're hoping launch day or a Product Hunt bump will save you, you've already misunderstood the game. Growth is a habit, not an event.
Practical version: spend the first 6–12 months talking to every single user. Onboard them by Zoom. Fix their bugs personally. Read every support ticket. The data you get is unfair leverage no competitor has.
Source
paulgraham.com — Do Things That Don't Scale →We summarize so you can decide whether to read the full piece. Always read the source for context.