AI summary by TwinMind
Startup Experts Discuss Doing Things That Don't Scale
Y Combinator
25m 41s
Historical Context and Paul Graham's Essay
- Early 2000s startup culture obsessed with "scalability" due to technical limitations
- Slow processors, limited bandwidth made scaling websites difficult
- Google's success warped minds of founders/investors - everyone wanted scalable solutions
- Couldn't raise funding without demonstrating scalability
- Paul Graham wrote "Do Things That Don't Scale" (2013) to solve this problem
- Essay transformed Silicon Valley culture over following years
- Contrarian advice analogous to Zuckerberg's "move fast and break things"
- Focus on biggest current problem rather than future scaling issues
- Most startups' real problem: can't get users, not building what people want
- Many build beautiful, scalable products with zero demand
Core Philosophy and Principles
- Worry about zero-to-one problem first before scaling concerns
- Optimize for learning rather than theoretical scalability
- Nothing is beneath founder - no task too small/manual if serves customers
- Best founders dive into unglamorous work regardless of background
- Focus on product-market fit, serving customers, getting users
- Mindset: "nothing is beneath me" for early-stage founders
Airbnb Case Study - Photo Taking
- Airbnb founders had no users, needed flywheel turning
- Paul Graham advised: go take professional photos of listings yourself
- High-quality photos most important part of listings
- Founders personally photographed properties to create catalog
- Clearly unscalable but got flywheel started
- Alternative would have been missing opportunity due to scalability concerns
Fleek Marketplace Example
- Winter 22 batch company connecting secondhand clothing wholesalers to shops
- Started with just 3 guys, no website, no inventory
- Went to London wholesalers, borrowed boxes of clothes for 6 hours
- Sold clothes to shops around London, returned with $6,000 for wholesaler
- Repeated process for 4 months to learn:
- What sells vs. doesn't sell
- Best timing for bringing clothing in
- Pricing strategies
- Demand elasticity
- Eventually transitioned to website-based marketplace
- Learned lessons "into their bones" through manual process
Stripe and Algolia Implementation Stories
- Stripe founders implemented their API directly in customers' software
- Algolia implemented search for Product Hunt via GitHub pull request
- Ryan (Product Hunt founder) lacked resources for self-implementation
- Created stronger customer connections than traditional sales approach
- Enabled more candid conversations and better product development context
- Optimized for learning rather than hiding behind computers
Founder Face-Time and Sales Advantage
- Nothing like founder face-time in early days for recruiting customers/employees
- Ryan from Vendor (B2B SaaS): "sell yourself when product is janky"
- Give personal cell phone number, available 24/7
- Message: "basing future success on making you successful"
- Advantage over big competitors - middle managers won't give personal numbers
- Make customers feel cared for, that founder's career depends on their success
Instacart Launch Strategy
- Launched without single grocery store partnership
- Instead of negotiating corporate deals (takes years):
- Bought one of every item at Trader Joe's with YC money
- Photographed all products in studio over weekend
- Listed everything online Monday morning
- Eventually struck deals after gaining scale and negotiating leverage
- Demonstrated idea's value to grocery stores through proven demand
DoorDash One-Day Build
- Entire product built in one afternoon by Stanford engineers/MBAs
- Tech stack:
- Google Drive for menu uploads
- HTML/CSS for displaying menus
- Tony's personal phone for orders
- Find My Friends for fake real-time dispatch
- Google Forms for order taking
- One founder picked up orders, another tracked via Find My Friends
- Texted customers ETAs manually
- Could have built sophisticated system but prioritized validation over perfection
Startup vs. Big Company Mentality
- Big companies optimize for error-free execution, take forever
- Startups should embrace errors as features
- Analogy: checking pipes
- Big company: check each pipe individually (2 years)
- Startup: turn on all water, fix where leaks appear
- High motivation to fix problems when too many users crash system
- Startups rarely die from scaling problems - 100% solvable with right incentives
When to Stop Doing Unscalable Things
- Critical transition point - many founders get stuck at zero-to-one
- Need good advisors/investors to identify switch point
- Example: Optimizely consulting trap
- Started by manually building A/B tests for clients
- Made money but wouldn't scale to big business
- Risk: getting addicted to consulting revenue
- Solution: Set ambitious growth targets (10x annually)
- Can't grow consultancy 10x in year
- Can grow real software business 10x
- Investors use growth targets to distinguish consultancies from scalable startups
Key Benefits of Unscalable Approach
- Intimately learn about customers and their problems
- Tie early users more tightly to company
- Create opportunities to delight customers
- Enable quick experimentation and fast failure
- Gain competitive advantage over larger competitors
- Test assumptions before months of engineering investment
- Freedom to pivot quickly with minimal technical debt
