YC OS · Chapter 02
Do Things That Don't Scale
Every startup's early growth comes from something manual, unglamorous, and temporary. Do that thing on purpose.
After this lesson
- Recruit your first users by hand instead of waiting for them to arrive
- Give a small number of people an experience that is disproportionately good
- Know when to stop doing the unscalable thing and start automating
“The most common unscalable thing founders have to do at the start is to recruit users manually. Nearly all startups have to. You can't wait for users to come to you.”
Startups don't take off by themselves
Nearly every startup that later looked like it grew virally actually pushed a boulder by hand at the start. Someone recruited the first hundred users one at a time, set the product up for them personally, and fixed their problems in real time. That manual phase isn't a shortcut around growth — it's how you learn what's worth scaling.
What 'doesn't scale' looks like
- Onboarding users yourself instead of writing a self-serve flow
- Recruiting in one company or one city before opening up broadly
- Doing the service by hand behind an automated-looking front end
- Delighting a handful of users far beyond what's 'reasonable'
- Sitting next to a customer while they use the product for the first time
Case
Case: the Collison install
When someone agreed to try Stripe in its earliest days, the Collison brothers didn't send a signup link — they said 'give me your laptop' and set the product up on the spot, on the couch, right then.
Takeaway: Every point of friction you remove by hand for the first 20 users is a lesson about what your onboarding actually needs to do, not just a nice gesture.
The compounding math
- 10 users becoming 11 looks unimpressive; 10% weekly growth compounding for a year is roughly 14,000
- The unscalable phase is short by design — its job is to find the loop worth scaling
- Weekly growth rate, not absolute headcount, is the number worth watching this early
Push the boulder yourself this week
- List 20 real people who could use this today — actual names, not a persona
- Reach out to at least 10 of them yourself and offer to set it up on the spot, Collison-style
- Do one thing that's obviously beneath the size of company you imagine having — deliver it by hand, onboard someone in person, assemble it yourself
- Send at least one genuinely personal thank-you — Wufoo hand-wrote every one for years — to a user this week
- Write down this week's user count and next week's target; 10% weekly growth compounds faster than it feels like it should
Which unscalable thing should you do this week?
Pick based on what you don't know yet, not what feels productive.
Read the original
Do Things That Don't ScaleRead past the Recruit and Delight sections to Fire and Manual — the Facebook 'contained fire' and Stripe manual-merchant-account tricks are the two ideas most founders skip.
This chapter is a working summary. When the idea matters, read the source once — then come back and do the practice.
Read next
This chapter is theory until you've recruited someone by hand. Open the outreach practice below, add 10 real names, and reach out to them before you read further.